v4: Bitcoin Accumulation Zone Monitor — on-chain metrics + backtest engine

COMPLETE PIVOT from ML trading optimizer to on-chain metrics monitor. Architecture: - Playwright scrapes LookIntoBitcoin Plotly Dash charts for real on-chain data - 10 proven metrics: Puell Multiple, MVRV Z-Score, Fear & Greed, Reserve Risk, RHODL Ratio, NUPL, LTH Realized Price, 200W SMA, Hash Ribbons, Drawdown - Each metric scores 0-10, composite 0-100 - No ML, no black box — every signal transparent and traceable - Historical backtest validates scoring against actual BTC forward returns - Recency-weighted analysis accounts for diminishing cycle returns Full documentation in ARCHITECTURE.md
feat: add historical backtest engine and dashboard page
2026-03-20 23:07:53 +00:00 · 2026-03-20 22:50:57 +00:00 · 2026-03-20 22:31:35 +00:00 · 2026-03-20 22:31:29 +00:00 · 2026-03-20 21:51:05 +00:00 · 2026-03-19 23:55:51 +00:00
23 changed files with 4487 additions and 915 deletions
--- a/.gitignore
+++ b/.gitignore
@ -0,0 +1,9 @@
+__pycache__/
+*.pyc
+data/cache.json
+data/history.json
+config/llm_settings.json
+results/
+*.log
+.env
+node_modules/
--- a/ARCHITECTURE.md
+++ b/ARCHITECTURE.md
@ -0,0 +1,262 @@
+# Bitcoin Accumulation Zone Monitor — Architecture & Logic
+
+## Overview
+
+This is **NOT** a trading bot or ML predictor. It monitors proven Bitcoin on-chain metrics that have historically signaled optimal accumulation (buying) zones for long-term holders. Each metric scores 0-10 points, producing a composite score of 0-100.
+
+**Philosophy:** Every signal is transparent and traceable. No black box. The metrics used have correctly identified every major Bitcoin cycle bottom since 2010.
+
+## How It Works
+
+### Data Pipeline
+
+```
+LookIntoBitcoin.com ──┐
+  (Playwright scraper) │
+                       ├──> data/cache.json (current values, refreshed every 15min)
+alternative.me API ────┤    data/history.json (full history back to 2010, refreshed weekly)
+                       │
+CoinGecko API ─────────┘
+         │
+         ▼
+   Scoring Engine (scoring/engine.py)
+         │
+         ▼
+   Composite Score 0-100
+         │
+    ┌────┴────┐
+    ▼         ▼
+ Dashboard   Backtest Engine
+ (live)      (historical validation)
+```
+
+### Data Sources
+
+All data is scraped or fetched from free sources — **no API keys required**.
+
+| Source | Method | Data |
+|--------|--------|------|
+| LookIntoBitcoin / BitcoinMagazinePro | Playwright browser scraping of Plotly Dash charts | Puell Multiple, MVRV Z-Score, Reserve Risk, RHODL Ratio, NUPL, 200W SMA, LTH Realized Price, LTH Supply, Hash Ribbons, Pi Cycle |
+| alternative.me | Free REST API | Fear & Greed Index (daily, back to Feb 2018) |
+| CoinGecko | Free REST API | BTC price, market cap, 24h change |
+
+#### Scraping Method (LookIntoBitcoin)
+
+The site uses Plotly Dash charts. We intercept the `_dash-update-component` XHR response which contains the full chart data as JSON:
+
+```python
+page.on("response", handler)  # Intercept XHR
+page.goto("https://www.lookintobitcoin.com/charts/puell-multiple/")
+# Response contains: response['chart']['figure']['data'] → list of trace objects
+# Each trace: {name: str, x: [dates], y: [values]}
+```
+
+This gives us the **complete historical time series** (5000+ data points per metric going back to 2010) without needing any API key.
+
+## Scoring System
+
+### Individual Metrics (0-10 each)
+
+#### 1. Fear & Greed Index (source: alternative.me)
+Measures market sentiment from social media, surveys, and momentum.
+
+| F&G Value | Classification | Score |
+|-----------|---------------|-------|
+| 0-10 | Extreme Fear | 10 |
+| 11-25 | Fear | 7 |
+| 26-45 | Neutral-low | 4 |
+| 46-55 | Neutral | 2 |
+| 56-75 | Greed | 1 |
+| 76-100 | Extreme Greed | 0 |
+
+**Logic:** "Be fearful when others are greedy, be greedy when others are fearful." — Buffett. Extreme Fear has historically coincided with cycle bottoms.
+
+#### 2. Puell Multiple (source: LookIntoBitcoin)
+Measures miner revenue relative to 365-day average. When miners earn very little (low Puell), they're capitulating — historically a bottom signal.
+
+| Puell Value | Meaning | Score |
+|-------------|---------|-------|
+| < 0.3 | Deep miner capitulation | 10 |
+| 0.3-0.5 | Miner stress | 8 |
+| 0.5-0.8 | Below average revenue | 5 |
+| 0.8-1.2 | Normal | 3 |
+| 1.2-2.0 | Above average | 1 |
+| > 2.0 | Miner euphoria | 0 |
+
+**Historical accuracy:** Puell < 0.5 identified the Dec 2018, Mar 2020, and Jun 2022 bottoms.
+
+#### 3. MVRV Z-Score (source: LookIntoBitcoin)
+Compares market value to realized value. Negative Z-Score means the market is valued below what everyone paid — extreme undervaluation.
+
+| Z-Score | Meaning | Score |
+|---------|---------|-------|
+| < 0 | Below realized value | 10 |
+| 0-0.5 | Undervalued | 8 |
+| 0.5-1.5 | Fair value | 5 |
+| 1.5-3.0 | Overvalued | 2 |
+| 3.0-5.0 | Very overvalued | 1 |
+| > 5.0 | Extreme overvaluation | 0 |
+
+**Historical accuracy:** Every time MVRV Z-Score went below 0, buying led to >200% returns within 2 years (100% hit rate across all cycles).
+
+#### 4. Drawdown from ATH (calculated from price)
+How far BTC has fallen from its all-time high. Larger drawdowns = better buying opportunity historically.
+
+| Drawdown | Score |
+|----------|-------|
+| > 70% | 10 |
+| 50-70% | 8 |
+| 30-50% | 6 |
+| 20-30% | 4 |
+| 10-20% | 2 |
+| < 10% | 0 |
+
+#### 5. Price vs 200-Week SMA (source: LookIntoBitcoin)
+The 200-week moving average has historically acted as the absolute floor in bear markets.
+
+| Position | Score |
+|----------|-------|
+| Below 200W SMA | 10 |
+| 0-20% above | 6 |
+| 20-50% above | 3 |
+| 50-100% above | 1 |
+| > 100% above | 0 |
+
+#### 6. Reserve Risk (source: LookIntoBitcoin)
+Measures the confidence of long-term holders relative to the price. Low Reserve Risk = high confidence among HODLers + low price = excellent time to buy.
+
+| Reserve Risk | Score |
+|--------------|-------|
+| < 0.002 | 10 |
+| 0.002-0.005 | 7 |
+| 0.005-0.01 | 4 |
+| 0.01-0.02 | 2 |
+| > 0.02 | 0 |
+
+#### 7. RHODL Ratio (source: LookIntoBitcoin)
+Ratio of 1-week old coins to 1-2 year old coins. Low ratio = long-term holders dominating (accumulation). High ratio = short-term speculation (distribution).
+
+| RHODL | Score |
+|-------|-------|
+| < 100 | 10 |
+| 100-500 | 7 |
+| 500-2000 | 4 |
+| 2000-10000 | 1 |
+| > 10000 | 0 |
+
+#### 8. NUPL — Net Unrealized Profit/Loss (source: LookIntoBitcoin)
+Shows what fraction of market cap is unrealized profit. Negative = market is at a loss (capitulation). Above 0.75 = euphoria.
+
+| NUPL | Phase | Score |
+|------|-------|-------|
+| < 0 | Capitulation | 10 |
+| 0-0.25 | Hope/Fear | 7 |
+| 0.25-0.5 | Optimism | 4 |
+| 0.5-0.75 | Belief/Greed | 1 |
+| > 0.75 | Euphoria | 0 |
+
+#### 9. LTH Realized Price vs Spot (source: LookIntoBitcoin)
+Long-Term Holder Realized Price = average cost basis of coins held >155 days. When spot price drops below this, even diamond hands are underwater — extreme value.
+
+| Position | Score |
+|----------|-------|
+| Price below LTH RP | 10 |
+| 0-20% above | 6 |
+| 20-50% above | 3 |
+| > 50% above | 1 |
+
+#### 10. Hash Ribbons / Miner Capitulation (source: LookIntoBitcoin)
+When miners capitulate (hash rate declining), it signals maximum pain. The recovery signal (hash rate resuming growth) has been a reliable buy signal.
+
+| Signal | Score |
+|--------|-------|
+| Active buy signal | 10 |
+| Recent recovery | 6 |
+| Normal | 3 |
+| Miner euphoria | 0 |
+
+### Composite Score
+
+```
+Total Score = Sum of all individual metric scores (0-100)
+```
+
+| Score Range | Assessment | Action |
+|-------------|------------|--------|
+| 85-100 | Extreme Accumulation Zone | Strong buy — historically rare, ~4x per decade |
+| 70-84 | Strong Accumulation | Buy — excellent long-term entry |
+| 55-69 | Moderate Opportunity | Consider buying — decent entry |
+| 40-54 | Neutral | Hold — not compelling either way |
+| 25-39 | Caution | Reduce or wait — market heating up |
+| 0-24 | Extreme Caution | Do NOT buy — historically the worst times |
+
+## Backtest Engine
+
+### Purpose
+Reconstruct the composite score historically and compare against actual BTC forward returns to validate the scoring system's accuracy.
+
+### Methodology
+
+1. **Historical Reconstruction:** Using scraped historical data (2010-present), calculate what each metric's score would have been on every day
+2. **Forward Returns:** For each historical day, calculate what BTC actually returned over the next 30, 90, 180, and 365 days
+3. **Score Bracket Analysis:** Group days by score bracket and calculate average forward returns, win rates, max drawdowns
+4. **Recency Weighting:** More recent cycles weighted higher because BTC's cycle-over-cycle returns diminish as it matures:
+   - 2022-present: 4x weight
+   - 2020-2021: 3x weight
+   - 2018-2019: 2x weight
+   - Before 2018: 1x weight
+5. **Cycle-Separated Results:** Returns shown per cycle (Cycle 3: 2016-2019, Cycle 4: 2020-2023, Cycle 5: 2024+)
+
+### Diminishing Returns Adjustment
+
+Bitcoin's gains decrease every cycle. A score of 90 in 2018 led to different outcomes than a score of 90 in 2022:
+- The backtest separates results by cycle
+- Current expectations are based on the 2 most recent comparable cycles
+- Adaptive thresholds recalculate based on rolling 2-year windows
+
+## Architecture
+
+```
+/opt/apps/btc-ml-optimizer/
+├── dashboard/
+│   └── server.py              # FastAPI + inline HTML/JS dashboard
+├── scrapers/
+│   ├── __init__.py
+│   ├── lookintobitcoin.py     # Playwright scraper for on-chain charts
+│   ├── history_collector.py   # Full historical data collection
+│   ├── fear_greed.py          # alternative.me Fear & Greed API
+│   └── price.py               # CoinGecko BTC price API
+├── scoring/
+│   ├── __init__.py
+│   └── engine.py              # Scoring logic and thresholds
+├── backtesting/
+│   ├── __init__.py
+│   └── engine.py              # Historical backtest calculations
+├── data/
+│   ├── cache.json             # Current metric values (refreshed every 15min)
+│   └── history.json           # Full historical data (refreshed weekly)
+├── config/
+│   ├── thresholds.json        # Configurable scoring thresholds
+│   └── llm_settings.json      # Optional LLM provider config for AI commentary
+├── llm_client/
+│   └── analyzer.py            # Optional LLM integration for signal analysis
+└── README.md
+```
+
+## Infrastructure
+
+- **Server:** Main VPS (Hostinger), Tailscale IP 100.94.106.120
+- **Port:** 3088
+- **Process Manager:** pm2 (`btc-ml-optimizer`)
+- **Dashboard URL:** http://100.94.106.120:3088
+- **Backtest URL:** http://100.94.106.120:3088/backtest
+- **Git Repo:** https://git.bizzle.lol/bizzle/btc-accumulation-monitor
+
+## Dependencies
+
+- Python 3.13
+- FastAPI + uvicorn
+- Playwright (Chromium, headless)
+- requests
+- No ML libraries required
+- No paid API keys required
--- a/pycache/orchestrator.cpython-313.pyc
+++ b/pycache/orchestrator.cpython-313.pyc
--- a/backtesting/init.py
+++ b/backtesting/init.py
--- a/backtesting/engine.py
+++ b/backtesting/engine.py
@ -0,0 +1,412 @@
+"""Historical backtest engine for Bitcoin Accumulation Zone scoring."""
+
+import json
+import logging
+import os
+import sys
+from collections import defaultdict
+from datetime import datetime, timedelta
+
+log = logging.getLogger(__name__)
+
+BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+sys.path.insert(0, BASE_DIR)
+
+HISTORY_PATH = os.path.join(BASE_DIR, "data", "history.json")
+CACHE_PATH = os.path.join(BASE_DIR, "data", "cache.json")
+
+# Score brackets matching the dashboard assessment levels
+BRACKETS = [
+    (0, 20, "Extreme Caution"),
+    (21, 40, "Caution"),
+    (41, 55, "Neutral"),
+    (56, 70, "Moderate Opportunity"),
+    (71, 85, "Strong Accumulation"),
+    (86, 100, "Extreme Accumulation"),
+]
+
+# Scoring thresholds — replicated from scoring/engine.py for standalone use
+METRIC_SCORERS = {
+    "fear_greed": {
+        "ranges": [[None, 10, 10], [10, 25, 7], [25, 45, 4], [45, 55, 2], [55, 75, 1], [75, None, 0]],
+    },
+    "puell_multiple": {
+        "ranges": [[None, 0.3, 10], [0.3, 0.5, 8], [0.5, 0.8, 5], [0.8, 1.2, 3], [1.2, 2.0, 1], [2.0, None, 0]],
+    },
+    "mvrv_zscore": {
+        "ranges": [[None, 0, 10], [0, 0.5, 8], [0.5, 1.5, 5], [1.5, 3, 2], [3, 5, 1], [5, None, 0]],
+    },
+    "reserve_risk": {
+        "ranges": [[None, 0.002, 10], [0.002, 0.005, 7], [0.005, 0.01, 4], [0.01, 0.02, 2], [0.02, None, 0]],
+    },
+    "rhodl_ratio": {
+        "ranges": [[None, 100, 10], [100, 500, 7], [500, 2000, 4], [2000, 10000, 1], [10000, None, 0]],
+    },
+    "nupl": {
+        "ranges": [[None, 0, 10], [0, 0.25, 7], [0.25, 0.5, 4], [0.5, 0.75, 1], [0.75, None, 0]],
+    },
+}
+
+# Ratio-based metrics: score based on price vs reference value
+RATIO_SCORERS = {
+    "price_vs_200w_sma": {
+        # pct_above ranges
+        "ranges": [[None, 0, 10], [0, 20, 6], [20, 50, 3], [50, 100, 1], [100, None, 0]],
+        "price_key": "btc_price",
+        "ref_key": "200w_sma",
+    },
+    "lth_realized_price": {
+        "ranges": [[None, 0, 10], [0, 20, 6], [20, 50, 3], [50, None, 1]],
+        "price_key": "btc_price",
+        "ref_key": "lth_realized_price",
+    },
+}
+
+# Drawdown scoring
+DRAWDOWN_RANGES = [[70, None, 10], [50, 70, 8], [30, 50, 6], [20, 30, 4], [10, 20, 2], [None, 10, 0]]
+
+
+def _score_range(value, ranges):
+    """Score a value using range-based thresholds."""
+    if value is None:
+        return None
+    for low, high, score in ranges:
+        low_ok = low is None or value >= low
+        high_ok = high is None or value < high
+        if low_ok and high_ok:
+            return score
+    return 0
+
+
+def _build_daily_index(history):
+    """Build a dict mapping metric_key -> {date_str: value} for fast lookup."""
+    index = {}
+    for key, data in history.items():
+        if key.startswith("_") or not isinstance(data, dict) or "dates" not in data:
+            continue
+        lookup = {}
+        for d, v in zip(data["dates"], data["values"]):
+            lookup[d] = v
+        index[key] = lookup
+    return index
+
+
+def _get_all_dates(index):
+    """Get sorted union of all dates across all metrics."""
+    all_dates = set()
+    for lookup in index.values():
+        all_dates.update(lookup.keys())
+    return sorted(all_dates)
+
+
+def _last_known_value(lookup, date, max_lookback=30):
+    """Get value for date, or most recent prior value within lookback window."""
+    if date in lookup:
+        return lookup[date]
+    d = datetime.strptime(date, "%Y-%m-%d")
+    for i in range(1, max_lookback + 1):
+        prev = (d - timedelta(days=i)).strftime("%Y-%m-%d")
+        if prev in lookup:
+            return lookup[prev]
+    return None
+
+
+def _compute_ath_series(price_lookup, dates):
+    """Compute running ATH and drawdown for each date."""
+    ath = 0
+    drawdowns = {}
+    for d in dates:
+        p = price_lookup.get(d)
+        if p is None:
+            continue
+        if p > ath:
+            ath = p
+        if ath > 0:
+            drawdowns[d] = ((ath - p) / ath) * 100
+    return drawdowns
+
+
+def score_day(date, index, drawdowns):
+    """Score a single day using all available metrics. Returns (composite_score, individual_scores, n_metrics)."""
+    scores = []
+    details = {}
+
+    # Simple range-based metrics
+    for metric_key, cfg in METRIC_SCORERS.items():
+        val = _last_known_value(index.get(metric_key, {}), date)
+        if val is not None:
+            s = _score_range(val, cfg["ranges"])
+            if s is not None:
+                scores.append(s)
+                details[metric_key] = {"value": val, "score": s}
+
+    # Ratio-based metrics (price vs reference)
+    for metric_key, cfg in RATIO_SCORERS.items():
+        price_val = _last_known_value(index.get(cfg["price_key"], {}), date)
+        # Try alternate price keys
+        if price_val is None:
+            for pk in ["btc_price_coingecko", "btc_price_sma", "btc_price_lth"]:
+                price_val = _last_known_value(index.get(pk, {}), date)
+                if price_val is not None:
+                    break
+        ref_val = _last_known_value(index.get(cfg["ref_key"], {}), date)
+        if price_val is not None and ref_val is not None and ref_val > 0:
+            pct_above = ((price_val - ref_val) / ref_val) * 100
+            s = _score_range(pct_above, cfg["ranges"])
+            if s is not None:
+                scores.append(s)
+                details[metric_key] = {"value": pct_above, "score": s}
+
+    # Drawdown
+    dd = drawdowns.get(date)
+    if dd is not None:
+        s = _score_range(dd, DRAWDOWN_RANGES)
+        if s is not None:
+            scores.append(s)
+            details["drawdown"] = {"value": dd, "score": s}
+
+    if not scores:
+        return None, details, 0
+
+    composite = sum(scores) / len(scores) * 10
+    return round(composite, 1), details, len(scores)
+
+
+def compute_forward_returns(price_lookup, dates_sorted):
+    """Precompute forward returns for all dates."""
+    periods = [30, 90, 180, 365]
+    returns = {}
+    for d in dates_sorted:
+        p0 = price_lookup.get(d)
+        if p0 is None or p0 <= 0:
+            continue
+        r = {}
+        dt = datetime.strptime(d, "%Y-%m-%d")
+        for days in periods:
+            future = (dt + timedelta(days=days)).strftime("%Y-%m-%d")
+            pf = price_lookup.get(future)
+            if pf is not None:
+                r[f"{days}d"] = round(((pf - p0) / p0) * 100, 2)
+        if r:
+            returns[d] = r
+    return returns
+
+
+def compute_max_drawdown_forward(price_lookup, date, window=90):
+    """Compute max drawdown within N days after a given date."""
+    dt = datetime.strptime(date, "%Y-%m-%d")
+    p0 = price_lookup.get(date)
+    if p0 is None or p0 <= 0:
+        return None
+    peak = p0
+    max_dd = 0
+    for i in range(1, window + 1):
+        future = (dt + timedelta(days=i)).strftime("%Y-%m-%d")
+        pf = price_lookup.get(future)
+        if pf is None:
+            continue
+        if pf > peak:
+            peak = pf
+        dd = ((peak - pf) / peak) * 100
+        if dd > max_dd:
+            max_dd = dd
+    return round(max_dd, 2) if max_dd > 0 else 0
+
+
+def run_backtest():
+    """Run the full backtest and return comprehensive results."""
+    log.info("Loading historical data...")
+    if not os.path.exists(HISTORY_PATH):
+        return {"error": "No historical data found. Run history collector first."}
+
+    with open(HISTORY_PATH) as f:
+        history = json.load(f)
+
+    index = _build_daily_index(history)
+
+    # Build price lookup (prefer coingecko for completeness)
+    price_lookup = {}
+    for pk in ["btc_price_coingecko", "btc_price", "btc_price_sma", "btc_price_lth"]:
+        if pk in index:
+            for d, v in index[pk].items():
+                if d not in price_lookup:
+                    price_lookup[d] = v
+
+    all_dates = _get_all_dates(index)
+    if not all_dates:
+        return {"error": "No date data available."}
+
+    log.info("Date range: %s to %s (%d days)", all_dates[0], all_dates[-1], len(all_dates))
+
+    # Compute drawdowns
+    drawdowns = _compute_ath_series(price_lookup, all_dates)
+
+    # Precompute forward returns
+    log.info("Computing forward returns...")
+    fwd_returns = compute_forward_returns(price_lookup, all_dates)
+
+    # Score each day
+    log.info("Scoring %d days...", len(all_dates))
+    daily_scores = []
+    for d in all_dates:
+        composite, details, n_metrics = score_day(d, index, drawdowns)
+        if composite is not None and n_metrics >= 3:  # Require at least 3 metrics
+            price = price_lookup.get(d)
+            entry = {
+                "date": d,
+                "score": composite,
+                "n_metrics": n_metrics,
+                "price": price,
+                "forward_returns": fwd_returns.get(d, {}),
+            }
+            daily_scores.append(entry)
+
+    if not daily_scores:
+        return {"error": "No scored days (insufficient metric overlap)."}
+
+    log.info("Scored %d days with 3+ metrics", len(daily_scores))
+
+    # --- Bracket statistics ---
+    bracket_stats = []
+    for low, high, label in BRACKETS:
+        days_in = [d for d in daily_scores if low <= d["score"] <= high]
+        if not days_in:
+            bracket_stats.append({
+                "range": f"{low}-{high}", "label": label, "days": 0,
+            })
+            continue
+
+        stats = {"range": f"{low}-{high}", "label": label, "days": len(days_in)}
+        for period in ["30d", "90d", "180d", "365d"]:
+            returns = [d["forward_returns"][period] for d in days_in if period in d["forward_returns"]]
+            if returns:
+                returns_sorted = sorted(returns)
+                stats[f"avg_{period}"] = round(sum(returns) / len(returns), 2)
+                stats[f"median_{period}"] = round(returns_sorted[len(returns_sorted) // 2], 2)
+                stats[f"win_rate_{period}"] = round(len([r for r in returns if r > 0]) / len(returns) * 100, 1)
+                stats[f"max_gain_{period}"] = round(max(returns), 2)
+                stats[f"max_loss_{period}"] = round(min(returns), 2)
+                stats[f"n_{period}"] = len(returns)
+
+        # Average max drawdown within 90 days
+        dd_list = []
+        for d in days_in:
+            dd = compute_max_drawdown_forward(price_lookup, d["date"], 90)
+            if dd is not None:
+                dd_list.append(dd)
+        if dd_list:
+            stats["avg_max_drawdown_90d"] = round(sum(dd_list) / len(dd_list), 2)
+
+        bracket_stats.append(stats)
+
+    # --- Peak signal events ---
+    signal_events = []
+    thresholds = [90, 80, 70]
+    for thresh in thresholds:
+        prev_score = 0
+        for d in daily_scores:
+            if d["score"] >= thresh and prev_score < thresh:
+                event = {
+                    "date": d["date"],
+                    "score": d["score"],
+                    "threshold": thresh,
+                    "price": d["price"],
+                    "forward_returns": d["forward_returns"],
+                }
+                # Add future prices
+                if d["price"]:
+                    dt = datetime.strptime(d["date"], "%Y-%m-%d")
+                    for days_ahead in [30, 90, 365]:
+                        future = (dt + timedelta(days=days_ahead)).strftime("%Y-%m-%d")
+                        fp = price_lookup.get(future)
+                        if fp:
+                            event[f"price_{days_ahead}d"] = round(fp, 2)
+                signal_events.append(event)
+            prev_score = d["score"]
+
+    signal_events.sort(key=lambda e: e["date"])
+
+    # --- Current signal context ---
+    all_scores_list = [d["score"] for d in daily_scores]
+    all_scores_list.sort()
+
+    # Get current score from cache
+    current_score = None
+    current_price = None
+    if os.path.exists(CACHE_PATH):
+        try:
+            with open(CACHE_PATH) as f:
+                cache = json.load(f)
+            scored = cache.get("_scored", {})
+            current_score = scored.get("composite_score")
+            current_price = cache.get("price", {}).get("price")
+        except Exception:
+            pass
+
+    # If no cache, use latest daily score
+    if current_score is None and daily_scores:
+        current_score = daily_scores[-1]["score"]
+        current_price = daily_scores[-1].get("price")
+
+    current_context = None
+    if current_score is not None:
+        # Percentile
+        below = len([s for s in all_scores_list if s <= current_score])
+        percentile = round(below / len(all_scores_list) * 100, 1)
+
+        # Find comparable historical periods
+        comparable = []
+        margin = 5
+        for d in daily_scores:
+            if abs(d["score"] - current_score) <= margin and d["forward_returns"]:
+                comparable.append(d)
+
+        avg_1yr = None
+        if comparable:
+            yr_returns = [d["forward_returns"]["365d"] for d in comparable if "365d" in d["forward_returns"]]
+            if yr_returns:
+                avg_1yr = round(sum(yr_returns) / len(yr_returns), 2)
+
+        # Best comparable examples (most recent 5)
+        examples = []
+        for d in comparable[-5:]:
+            examples.append({
+                "date": d["date"],
+                "score": d["score"],
+                "price": d["price"],
+                "forward_returns": d["forward_returns"],
+            })
+
+        current_context = {
+            "current_score": current_score,
+            "current_price": current_price,
+            "percentile": percentile,
+            "comparable_days": len(comparable),
+            "avg_1yr_return": avg_1yr,
+            "examples": examples,
+        }
+
+    # --- Build time series for charting ---
+    # Downsample to weekly for chart efficiency
+    chart_data = []
+    for i, d in enumerate(daily_scores):
+        # Include every 7th day + last day
+        if i % 7 == 0 or i == len(daily_scores) - 1:
+            chart_data.append({
+                "date": d["date"],
+                "score": d["score"],
+                "price": d["price"],
+            })
+
+    result = {
+        "date_range": {"start": daily_scores[0]["date"], "end": daily_scores[-1]["date"]},
+        "total_days_scored": len(daily_scores),
+        "bracket_stats": bracket_stats,
+        "signal_events": signal_events,
+        "current_context": current_context,
+        "chart_data": chart_data,
+        "computed_at": datetime.utcnow().isoformat() + "Z",
+    }
+
+    log.info("Backtest complete: %d days, %d signal events", len(daily_scores), len(signal_events))
+    return result
--- a/config/best_config.json
+++ b/config/best_config.json
@ -0,0 +1,68 @@
+{
+  "model_type": "xgboost",
+  "features": {
+    "use_price_position": true,
+    "use_momentum": true,
+    "use_volatility": true,
+    "use_volume": true,
+    "use_cycle": true,
+    "use_pca": false,
+    "pca_variance": 0.95,
+    "use_scaler": true
+  },
+  "target": {
+    "type": "regression",
+    "forward_periods_1h": [
+      168,
+      720,
+      2160
+    ],
+    "forward_periods_4h": [
+      42,
+      180,
+      540
+    ],
+    "weights": [
+      0.2,
+      0.3,
+      0.5
+    ],
+    "score_range": [
+      0,
+      100
+    ]
+  },
+  "hyperparameters": {
+    "learning_rate": 0.01,
+    "max_depth": 4,
+    "n_estimators": 300,
+    "subsample": 0.8,
+    "colsample_bytree": 0.8,
+    "min_child_weight": 20,
+    "gamma": 0.3,
+    "reg_alpha": 0.5,
+    "reg_lambda": 3.0,
+    "lstm_hidden_size": 128,
+    "lstm_num_layers": 2,
+    "lstm_dropout": 0.3,
+    "lstm_epochs": 100,
+    "lstm_batch_size": 64,
+    "lstm_sequence_length": 30,
+    "lstm_patience": 10
+  },
+  "strategy": {
+    "strong_buy_threshold": 65,
+    "good_buy_threshold": 55,
+    "poor_threshold": 35
+  },
+  "training": {
+    "rolling_window": true,
+    "rolling_train_size": 2500,
+    "rolling_test_size": 300,
+    "walk_forward_windows": 5,
+    "train_pct": 0.7,
+    "validation_pct": 0.15,
+    "test_pct": 0.15
+  },
+  "timeframe": "4h"
+}
--- a/config/current_config.json
+++ b/config/current_config.json
@ -0,0 +1,68 @@
+{
+  "model_type": "xgboost",
+  "features": {
+    "use_price_position": true,
+    "use_momentum": true,
+    "use_volatility": true,
+    "use_volume": true,
+    "use_cycle": true,
+    "use_pca": false,
+    "pca_variance": 0.85,
+    "use_scaler": true
+  },
+  "target": {
+    "type": "regression",
+    "forward_periods_1h": [
+      168,
+      720,
+      2160
+    ],
+    "forward_periods_4h": [
+      42,
+      180,
+      540
+    ],
+    "weights": [
+      0.2,
+      0.3,
+      0.5
+    ],
+    "score_range": [
+      0,
+      100
+    ]
+  },
+  "hyperparameters": {
+    "learning_rate": 0.005,
+    "max_depth": 5,
+    "n_estimators": 800,
+    "subsample": 0.7,
+    "colsample_bytree": 0.7,
+    "min_child_weight": 15,
+    "gamma": 0.5,
+    "reg_alpha": 0.3,
+    "reg_lambda": 1.0,
+    "lstm_hidden_size": 64,
+    "lstm_num_layers": 2,
+    "lstm_dropout": 0.4,
+    "lstm_epochs": 80,
+    "lstm_batch_size": 64,
+    "lstm_sequence_length": 30,
+    "lstm_patience": 15
+  },
+  "strategy": {
+    "strong_buy_threshold": 55,
+    "good_buy_threshold": 35,
+    "poor_threshold": 20
+  },
+  "training": {
+    "rolling_window": true,
+    "rolling_train_size": 3500,
+    "rolling_test_size": 300,
+    "walk_forward_windows": 5,
+    "train_pct": 0.7,
+    "validation_pct": 0.15,
+    "test_pct": 0.15
+  },
+  "timeframe": "4h"
+}
--- a/config/initial_config.json
+++ b/config/initial_config.json
@ -1,59 +1,68 @@
 {
  "model_type": "xgboost",
  "features": {
-    "technical_indicators": [
-      "RSI_14", "RSI_7", "RSI_21",
-      "MACD_line", "MACD_signal", "MACD_hist",
-      "BB_upper", "BB_lower", "BB_width",
-      "ATR_14",
-      "SMA_5", "SMA_10", "SMA_20", "SMA_50", "SMA_200",
-      "EMA_5", "EMA_10", "EMA_20", "EMA_50",
-      "OBV",
-      "stoch_k", "stoch_d",
-      "williams_r",
-      "CCI_20",
-      "ROC_10",
-      "keltner_upper", "keltner_lower"
-    ],
-    "lookback_periods": [3, 5, 10, 20],
-    "use_volume_features": true,
-    "use_volatility_features": true,
-    "use_candle_patterns": true,
-    "use_lag_features": true,
-    "lag_periods": [1, 2, 3, 5]
+    "use_price_position": true,
+    "use_momentum": true,
+    "use_volatility": true,
+    "use_volume": true,
+    "use_cycle": true,
+    "use_pca": false,
+    "pca_variance": 0.95,
+    "use_scaler": true
  },
  "target": {
-    "type": "classification",
-    "direction": "long",
-    "horizon_candles": 6,
-    "threshold_pct": 1.0
+    "type": "regression",
+    "forward_periods_1h": [
+      168,
+      720,
+      2160
+    ],
+    "forward_periods_4h": [
+      42,
+      180,
+      540
+    ],
+    "weights": [
+      0.2,
+      0.3,
+      0.5
+    ],
+    "score_range": [
+      0,
+      100
+    ]
  },
  "hyperparameters": {
-    "learning_rate": 0.05,
-    "max_depth": 6,
-    "n_estimators": 500,
+    "learning_rate": 0.01,
+    "max_depth": 4,
+    "n_estimators": 300,
    "subsample": 0.8,
    "colsample_bytree": 0.8,
-    "min_child_weight": 5,
-    "gamma": 0.1,
-    "reg_alpha": 0.1,
-    "reg_lambda": 1.0
+    "min_child_weight": 20,
+    "gamma": 0.3,
+    "reg_alpha": 0.5,
+    "reg_lambda": 3.0,
+    "lstm_hidden_size": 128,
+    "lstm_num_layers": 2,
+    "lstm_dropout": 0.3,
+    "lstm_epochs": 100,
+    "lstm_batch_size": 64,
+    "lstm_sequence_length": 30,
+    "lstm_patience": 10
  },
  "strategy": {
-    "entry_threshold": 0.60,
-    "exit_type": "trailing_stop",
-    "stop_loss_pct": 2.0,
-    "take_profit_pct": 4.0,
-    "trailing_stop_pct": 1.5,
-    "position_sizing": "confidence_scaled",
-    "max_position_pct": 100,
-    "min_confidence_to_trade": 0.55
+    "strong_buy_threshold": 65,
+    "good_buy_threshold": 55,
+    "poor_threshold": 35
  },
  "training": {
+    "rolling_window": true,
+    "rolling_train_size": 2500,
+    "rolling_test_size": 300,
    "walk_forward_windows": 5,
    "train_pct": 0.7,
    "validation_pct": 0.15,
    "test_pct": 0.15
  },
  "timeframe": "4h"
-}
+}
--- a/config/llm_settings.json
+++ b/config/llm_settings.json
@ -0,0 +1,21 @@
+{
+  "provider": "ollama",
+  "model": "qwen3.5:27b",
+  "providers": {
+    "ollama": {
+      "base_url": "http://100.100.242.21:11434"
+    },
+    "lmstudio": {
+      "base_url": "http://100.100.242.21:1234"
+    },
+    "openai": {
+      "api_key": ""
+    },
+    "anthropic": {
+      "api_key": ""
+    },
+    "openrouter": {
+      "api_key": ""
+    }
+  }
+}
--- a/config/thresholds.json
+++ b/config/thresholds.json
@ -0,0 +1,35 @@
+{
+  "fear_greed": {
+    "ranges": [[0, 10, 10], [11, 25, 7], [26, 45, 4], [46, 55, 2], [56, 75, 1], [76, 100, 0]]
+  },
+  "puell_multiple": {
+    "ranges": [[null, 0.3, 10], [0.3, 0.5, 8], [0.5, 0.8, 5], [0.8, 1.2, 3], [1.2, 2.0, 1], [2.0, null, 0]]
+  },
+  "mvrv_zscore": {
+    "ranges": [[null, 0, 10], [0, 0.5, 8], [0.5, 1.5, 5], [1.5, 3, 2], [3, 5, 1], [5, null, 0]]
+  },
+  "drawdown": {
+    "ranges": [[70, null, 10], [50, 70, 8], [30, 50, 6], [20, 30, 4], [10, 20, 2], [null, 10, 0]]
+  },
+  "price_vs_200w_sma": {
+    "ranges": [[null, 0, 10], [0, 20, 6], [20, 50, 3], [50, 100, 1], [100, null, 0]]
+  },
+  "reserve_risk": {
+    "ranges": [[null, 0.002, 10], [0.002, 0.005, 7], [0.005, 0.01, 4], [0.01, 0.02, 2], [0.02, null, 0]]
+  },
+  "rhodl_ratio": {
+    "ranges": [[null, 100, 10], [100, 500, 7], [500, 2000, 4], [2000, 10000, 1], [10000, null, 0]]
+  },
+  "nupl": {
+    "ranges": [[null, 0, 10], [0, 0.25, 7], [0.25, 0.5, 4], [0.5, 0.75, 1], [0.75, null, 0]]
+  },
+  "lth_realized_price": {
+    "ranges": [[null, 0, 10], [0, 20, 6], [20, 50, 3], [50, null, 1]]
+  },
+  "hash_ribbons": {
+    "buy_signal": 10,
+    "recent_recovery": 6,
+    "normal": 3,
+    "euphoria": 0
+  }
+}
--- a/dashboard/server.py
+++ b/dashboard/server.py
--- a/data/score_history.jsonl
+++ b/data/score_history.jsonl
@ -0,0 +1,4 @@
+{"timestamp": "2026-03-20T22:26:50.475811+00:00", "composite_score": 32.5, "scored_count": 8, "metrics": {"fear_greed": {"score": 7, "value": 11}, "puell_multiple": {"score": 5, "value": 0.6602699608966011}, "mvrv_zscore": {"score": 5, "value": 0.5211180167687892}, "drawdown": {"score": 6, "value": 43.891180203045685}, "price_vs_200w_sma": {"score": null, "value": 0.0}, "reserve_risk": {"score": 0, "value": 69871.0}, "rhodl_ratio": {"score": 0, "value": 69871.0}, "nupl": {"score": 0, "value": 69871.0}, "lth_realized_price": {"score": null, "value": null}, "hash_ribbons": {"score": 3, "value": null}}}
+{"timestamp": "2026-03-20T22:30:13.547149+00:00", "composite_score": 51.0, "scored_count": 10, "metrics": {"fear_greed": {"score": 7, "value": 11}, "puell_multiple": {"score": 5, "value": 0.6602699608966011}, "mvrv_zscore": {"score": 5, "value": 0.5211180167687892}, "drawdown": {"score": 6, "value": 43.910215736040605}, "price_vs_200w_sma": {"score": 3, "value": 58895.78086828114}, "reserve_risk": {"score": 10, "value": 0.0012985709697654493}, "rhodl_ratio": {"score": 4, "value": 1230.6243545314708}, "nupl": {"score": 7, "value": 0.22243290955405431}, "lth_realized_price": {"score": 1, "value": 43346.58756410873}, "hash_ribbons": {"score": 3, "value": null}}}
+{"timestamp": "2026-03-20T22:46:34.952569+00:00", "composite_score": 51.0, "scored_count": 10, "metrics": {"fear_greed": {"score": 7, "value": 11}, "puell_multiple": {"score": 5, "value": 0.6602699608966011}, "mvrv_zscore": {"score": 5, "value": 0.5211180167687892}, "drawdown": {"score": 6, "value": 43.931630710659896}, "price_vs_200w_sma": {"score": 3, "value": 58895.78086828114}, "reserve_risk": {"score": 10, "value": 0.0012985709697654493}, "rhodl_ratio": {"score": 4, "value": 1230.6243545314708}, "nupl": {"score": 7, "value": 0.22243290955405431}, "lth_realized_price": {"score": 1, "value": 43346.58756410873}, "hash_ribbons": {"score": 3, "value": null}}}
+{"timestamp": "2026-03-20T22:51:27.724327+00:00", "composite_score": 54.0, "scored_count": 10, "metrics": {"fear_greed": {"score": 7, "value": 11}, "puell_multiple": {"score": 5, "value": 0.6602699608966011}, "mvrv_zscore": {"score": 5, "value": 0.5211180167687892}, "drawdown": {"score": 6, "value": 43.94907994923858}, "price_vs_200w_sma": {"score": 6, "value": 58895.78086828114}, "reserve_risk": {"score": 10, "value": 0.0012985709697654493}, "rhodl_ratio": {"score": 4, "value": 1230.6243545314708}, "nupl": {"score": 7, "value": 0.22243290955405431}, "lth_realized_price": {"score": 1, "value": 43346.58756410873}, "hash_ribbons": {"score": 3, "value": null}}}
--- a/llm_client/pycache/analyzer.cpython-313.pyc
+++ b/llm_client/pycache/analyzer.cpython-313.pyc
--- a/llm_client/analyzer.py
+++ b/llm_client/analyzer.py
@ -1,109 +1,274 @@
 #!/usr/bin/env python3
 """
-LLM Strategy Analyzer — Calls Ollama on Mac Mini to analyze results
+LLM Accumulation Signal Analyzer -- Calls LLM to analyze results
 and suggest config modifications for the next iteration.
+Supports multiple providers: Ollama, LM Studio, OpenAI, Anthropic, OpenRouter.
 """

 import json
+import os
 import re
 import requests

-OLLAMA_URL = "http://100.100.242.21:11434"
-MODEL = "qwen3.5:27b"
+BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+LLM_SETTINGS_PATH = os.path.join(BASE_DIR, "config", "llm_settings.json")

-SYSTEM_PROMPT = """You are a quantitative trading strategy optimizer. You analyze ML model backtesting results for a BTC/USDT trading strategy and suggest precise modifications to improve performance.
+# Fallback defaults
+DEFAULT_OLLAMA_URL = "http://100.100.242.21:11434"
+DEFAULT_MODEL = "qwen3.5:27b"

-## Your Task
-Given the current configuration and results, suggest 1-3 specific, justified changes to the configuration for the next iteration. Be methodical and scientific — change one thing at a time when possible.
+
+def load_llm_settings():
+    """Load LLM settings from config file, with fallback to defaults."""
+    if os.path.exists(LLM_SETTINGS_PATH):
+        with open(LLM_SETTINGS_PATH) as f:
+            return json.load(f)
+    return {
+        "provider": "ollama",
+        "model": DEFAULT_MODEL,
+        "providers": {
+            "ollama": {"base_url": DEFAULT_OLLAMA_URL},
+        },
+    }
+
+
+SYSTEM_PROMPT = """You are a quantitative analyst optimizing a BTC ACCUMULATION SIGNAL model. The goal is NOT day-trading -- it is finding statistically optimal times to BUY BTC for long-term holding.
+
+## Core Question
+"Given current market conditions, is NOW a good time to BUY BTC for long-term holding?"
+
+## What the Model Does
+For each candle, the model predicts an Accumulation Score (0-100):
+- 90-100: STRONG BUY -- historically rare, excellent entry point
+- 70-89: GOOD BUY -- better than average entry
+- 50-69: NEUTRAL -- average time to buy
+- 30-49: WAIT -- price likely to come down
+- 0-29: POOR -- historically bad time to buy (near local tops)
+
+The model is trained on ACTUAL forward returns at 7d, 30d, and 90d horizons, weighted 20/30/50. Times when buying led to the best long-term returns get the highest scores.
+
+## Primary Metric: cost_basis_improvement_pct
+This measures how much better the model's average buy price is vs uniform DCA.
+- 10%+ = good
+- 15%+ = excellent
+- 20%+ = exceptional
+Also require strong_buy_signal_count >= 30 for statistical validity.

 ## Config Parameters You Can Modify

-**model_type**: "xgboost", "lightgbm", "catboost", or "ensemble"
-  - xgboost: Generally best for structured data, fast GPU training
-  - lightgbm: Faster training, good with large feature sets
-  - catboost: Handles feature interactions well, less tuning needed
-  - ensemble: Combines all three, reduces variance but slower
+**model_type**: "xgboost", "lightgbm", "catboost", "lstm", or "hybrid"
+  - hybrid: Average of LSTM + XGBoost regression predictions. Recommended default.
+  - xgboost: Fast GPU training, good for structured features.
+  - lstm: Captures temporal patterns in price sequences.

-**hyperparameters**:
-  - learning_rate (0.001-0.3): Lower = more robust but slower. If overfitting, decrease.
-  - max_depth (3-10): Controls model complexity. Deeper = more overfitting risk.
-  - n_estimators (100-2000): More trees = better fit but diminishing returns.
-  - subsample (0.5-1.0): Row sampling. Lower = more regularization.
-  - colsample_bytree (0.5-1.0): Feature sampling per tree. Lower = more diversity.
-  - min_child_weight (1-20): Higher = more conservative splits.
-  - gamma (0-5): Minimum loss reduction for split. Higher = more pruning.
-  - reg_alpha (0-10): L1 regularization. Encourages sparsity.
-  - reg_lambda (0-10): L2 regularization. Prevents large weights.
+**hyperparameters** (gradient boosting):
+  - learning_rate (0.001-0.1): Lower = more robust. Start conservative.
+  - max_depth (3-8): Controls complexity. Deeper risks overfitting.
+  - n_estimators (200-1500): More trees = better fit but diminishing returns.
+  - subsample (0.5-1.0): Row sampling for regularization.
+  - colsample_bytree (0.5-1.0): Feature sampling per tree.
+  - min_child_weight (5-30): Higher = more conservative (important for noisy targets).
+  - gamma (0-5): Minimum loss reduction for split.
+  - reg_alpha (0-10): L1 regularization.
+  - reg_lambda (1-10): L2 regularization. Higher values prevent overfitting.
+
+**hyperparameters** (LSTM):
+  - lstm_hidden_size (32-256): Hidden units.
+  - lstm_num_layers (1-4): Stacked layers. 2 is usually optimal.
+  - lstm_dropout (0.1-0.5): Regularization.
+  - lstm_epochs (50-200): Max training epochs (early stopping usually triggers).
+  - lstm_batch_size (32-128): Smaller = noisier but better generalization.
+  - lstm_sequence_length (15-60): Past candles the LSTM sees. Longer = more context.
+  - lstm_patience (5-20): Early stopping patience.

 **target**:
-  - direction: "long" or "both"
-  - horizon_candles (1-20): How far ahead to predict. Longer = smoother but lagging.
-  - threshold_pct (0.3-3.0): Minimum move % to label as positive. Higher = fewer but clearer signals.
+  - forward_periods_4h: List of 3 forward periods in 4h candles [short, medium, long].
+    Defaults: [42, 180, 540] = roughly [7d, 30d, 90d]
+  - weights: Weights for each period. Default [0.2, 0.3, 0.5] (emphasize long-term).
+  - score_range: [0, 100] -- do not change.

 **strategy**:
-  - entry_threshold (0.5-0.8): Min prediction probability to enter trade. Higher = fewer trades, higher quality.
-  - stop_loss_pct (0.5-5.0): Max loss before exit. Tighter = more stopped out.
-  - take_profit_pct (1.0-10.0): Target profit. Should be > stop_loss for positive expectancy.
-  - trailing_stop_pct (0.5-3.0): Trailing stop distance. Tighter = locks profit faster but exits early.
-  - min_confidence_to_trade (0.5-0.9): Absolute minimum confidence to consider.
-  - exit_type: "trailing_stop" or "fixed" (just SL/TP)
+  - strong_buy_threshold (70-95): Score above which = STRONG BUY signal. Higher = fewer but better signals.
+  - good_buy_threshold (50-80): Score above which = GOOD BUY. Used for cost basis comparison.
+  - poor_threshold (10-40): Score below which = POOR time to buy.

 **features**:
-  - use_volume_features (true/false): Volume features can be noisy in crypto.
-  - use_candle_patterns (true/false): Candle patterns may or may not help.
-  - use_lag_features (true/false): Lagged features capture momentum.
-  - lag_periods: List of lag periods [1,2,3,5,10]
-  - lookback_periods: List of lookback windows [3,5,10,20]
+  - use_price_position (true/false): Distance from ATH, 52w high/low, percentile.
+  - use_momentum (true/false): RSI, MACD, Stochastic, Williams %R, ROC.
+  - use_volatility (true/false): Bollinger Bands, ATR, consecutive red candles, drawdown.
+  - use_volume (true/false): Volume ratio, OBV, red/green volume ratio.
+  - use_cycle (true/false): MA cross regime, candles since major drawdown.
+  - use_pca (true/false): PCA dimensionality reduction.
+  - pca_variance (0.80-0.99): Variance to retain.
+  - use_scaler (true/false): StandardScaler. Critical for LSTM.

 **training**:
-  - walk_forward_windows (3-10): More windows = more robust but less data per window.
+  - rolling_window (true/false): Rolling vs static walk-forward.
+  - rolling_train_size (1500-5000): Training window candles.
+  - rolling_test_size (100-500): Test window candles.

-## Key Metrics to Optimize (in priority order)
-1. **Sharpe Ratio** (target: > 2.0): Risk-adjusted return. Most important metric.
-2. **Profit Factor** (target: > 1.5): Gross profit / gross loss.
-3. **Max Drawdown** (target: > -15%): Worst peak-to-trough decline.
-4. **Win Rate** (target: > 55%): Percentage of winning trades.
-5. **Trade Count**: Need enough trades for statistical significance (>50).
+## Key Metrics to Analyze
+1. **cost_basis_improvement_pct**: PRIMARY metric. How much better is model buy price vs DCA.
+2. **strong_buy_signal_count**: Must be >= 30 for validity. Too few = raise threshold. Too many = lower it.
+3. **signal_frequency_pct**: Should be 5-15%. If outside, adjust thresholds.
+4. **avg_score_at_actual_bottoms**: Should be high (>70). Model should recognize bottoms.
+5. **avg_score_at_actual_tops**: Should be low (<30). Model should avoid tops.
+6. **model_r2_score**: Regression fit quality. > 0.2 is decent for financial data.
+7. **per_window_cost_improvement**: Consistency across windows. Low variance = robust.

 ## Decision Guidelines
- If Sharpe < 1.0: The strategy is not working well. Consider larger changes.
- If Sharpe 1.0-1.5: Decent. Fine-tune hyperparameters and thresholds.
- If Sharpe 1.5-2.0: Good. Make small, targeted improvements.
- If Sharpe > 2.0: Very good. Be careful not to overfit.
- If win_rate < 0.50 but profit_factor > 1.5: Strategy relies on big wins — ok, tighten SL.
- If win_rate > 0.60 but profit_factor < 1.2: Many small wins but losses are too big — widen TP or tighten SL.
- If trade_count < 30: Not enough trades. Lower entry_threshold or min_confidence.
- If max_drawdown < -20%: Too risky. Increase regularization, tighten stop loss.
- If per_window_sharpe has high variance: Model is not stable. More regularization or simpler model.
- Check feature_importances: If top features make financial sense, good. If random features dominate, possible overfitting.
+- If cost_improvement < 5%: Strategy is barely working. Try: switch model type, enable all features, increase training window, lower good_buy_threshold.
+- If cost_improvement 5-10%: Decent. Fine-tune thresholds and hyperparameters.
+- If cost_improvement 10-15%: Good. Make targeted improvements -- focus on signal consistency.
+- If cost_improvement > 15%: Very good. Be careful not to overfit. Check per_window variance.
+- If signal_count < 30: Not statistically valid. Lower strong_buy_threshold, increase training data.
+- If signal_frequency > 20%: Too many signals = not selective enough. Raise threshold.
+- If signal_frequency < 3%: Too few signals. Lower threshold.
+- If score_at_bottoms < 60: Model is missing bottoms. More features, different model type.
+- If score_at_tops > 40: Model is not avoiding tops. More regularization.
+- If per_window has high variance: Model is unstable. Increase regularization, try hybrid.
+- Check feature_importances: price position features should dominate (distance from ATH, percentile).

 ## Response Format
 You MUST respond with ONLY a JSON object (no markdown, no explanation outside the JSON):
 ```
 {
-  "reasoning": "Explanation of what you observed and why you're making these changes",
+  "reasoning": "Explanation of observations and why you are making these changes",
  "changes": ["Change 1 description", "Change 2 description"],
  "config": { <complete modified config JSON> }
 }
 ```
-The "config" field must contain the COMPLETE config (not just changes) so it can be used directly."""
+The "config" field must contain the COMPLETE config so it can be used directly."""


-def analyze_and_suggest(current_config: dict, results: dict,
-                        iteration_history: list = None) -> tuple[dict, str]:
+def _call_ollama(settings, messages):
+    """Call Ollama API."""
+    provider_cfg = settings.get("providers", {}).get("ollama", {})
+    base_url = provider_cfg.get("base_url", DEFAULT_OLLAMA_URL)
+    model = settings.get("model", DEFAULT_MODEL)
+
+    payload = {
+        "model": model,
+        "messages": messages,
+        "stream": False,
+        "think": False,
+        "options": {"temperature": 0.7, "num_predict": 4096},
+    }
+    print(f"  Calling LLM ({model} via Ollama at {base_url})...")
+    resp = requests.post(f"{base_url}/api/chat", json=payload, timeout=600)
+    resp.raise_for_status()
+    return resp.json()["message"]["content"]
+
+
+def _call_openai_compatible(settings, messages, provider_name):
+    """Call OpenAI-compatible API (LM Studio, OpenAI, OpenRouter)."""
+    provider_cfg = settings.get("providers", {}).get(provider_name, {})
+    model = settings.get("model", "")
+
+    if provider_name == "lmstudio":
+        base_url = provider_cfg.get("base_url", "http://100.100.242.21:1234")
+        url = f"{base_url}/v1/chat/completions"
+        headers = {"Content-Type": "application/json"}
+    elif provider_name == "openai":
+        url = "https://api.openai.com/v1/chat/completions"
+        headers = {
+            "Content-Type": "application/json",
+            "Authorization": f"Bearer {provider_cfg.get('api_key', '')}",
+        }
+    elif provider_name == "openrouter":
+        url = "https://openrouter.ai/api/v1/chat/completions"
+        headers = {
+            "Content-Type": "application/json",
+            "Authorization": f"Bearer {provider_cfg.get('api_key', '')}",
+        }
+    else:
+        raise ValueError(f"Unknown OpenAI-compatible provider: {provider_name}")
+
+    payload = {
+        "model": model,
+        "messages": messages,
+        "temperature": 0.7,
+        "max_tokens": 4096,
+    }
+    print(f"  Calling LLM ({model} via {provider_name})...")
+    resp = requests.post(url, json=payload, headers=headers, timeout=600)
+    resp.raise_for_status()
+    return resp.json()["choices"][0]["message"]["content"]
+
+
+def _call_anthropic(settings, messages):
+    """Call Anthropic Messages API."""
+    provider_cfg = settings.get("providers", {}).get("anthropic", {})
+    model = settings.get("model", "claude-sonnet-4-20250514")
+    api_key = provider_cfg.get("api_key", "")
+
+    # Anthropic uses system as a top-level param, not in messages
+    system_msg = ""
+    api_messages = []
+    for m in messages:
+        if m["role"] == "system":
+            system_msg = m["content"]
+        else:
+            api_messages.append(m)
+
+    payload = {
+        "model": model,
+        "max_tokens": 4096,
+        "messages": api_messages,
+    }
+    if system_msg:
+        payload["system"] = system_msg
+
+    headers = {
+        "Content-Type": "application/json",
+        "x-api-key": api_key,
+        "anthropic-version": "2023-06-01",
+    }
+    print(f"  Calling LLM ({model} via Anthropic)...")
+    resp = requests.post(
+        "https://api.anthropic.com/v1/messages",
+        json=payload,
+        headers=headers,
+        timeout=600,
+    )
+    resp.raise_for_status()
+    data = resp.json()
+    # Extract text from content blocks
+    return "".join(
+        block["text"] for block in data.get("content", []) if block.get("type") == "text"
+    )
+
+
+def call_llm(messages):
+    """Route LLM call to the configured provider."""
+    settings = load_llm_settings()
+    provider = settings.get("provider", "ollama")
+
+    if provider == "ollama":
+        return _call_ollama(settings, messages)
+    elif provider in ("lmstudio", "openai", "openrouter"):
+        return _call_openai_compatible(settings, messages, provider)
+    elif provider == "anthropic":
+        return _call_anthropic(settings, messages)
+    else:
+        raise ValueError(f"Unknown LLM provider: {provider}")
+
+
+def analyze_and_suggest(current_config, results, iteration_history=None):
    """
    Send current results to LLM and get suggested config modifications.
    Returns (new_config, reasoning).
    """
-    # Build the user prompt with context
    history_text = ""
    if iteration_history:
        history_text = "\n## Previous Iterations (most recent last)\n"
        for h in iteration_history[-5:]:
            history_text += (
-                f"- Iteration {h['iteration']}: Sharpe={h['sharpe']}, "
-                f"Return={h['return']}%, WinRate={h['win_rate']}, "
-                f"Trades={h['trades']}, Model={h['model_type']}\n"
+                f"- Iteration {h.get('iteration', '?')}: "
+                f"CostImprovement={h.get('cost_improvement', 0):.1f}%, "
+                f"Signals={h.get('signal_count', 0)}, "
+                f"R2={h.get('r2_score', 0):.4f}, "
+                f"Model={h.get('model_type', '?')}\n"
            )

    user_prompt = f"""## Current Configuration
@ -112,40 +277,31 @@ def analyze_and_suggest(current_config: dict, results: dict,
 ```

 ## Current Results
- Sharpe Ratio: {results.get('sharpe_ratio', 0)}
- Total Return: {results.get('total_return_pct', 0)}%
- Max Drawdown: {results.get('max_drawdown_pct', 0)}%
- Win Rate: {results.get('win_rate', 0)}
- Trade Count: {results.get('trade_count', 0)}
- Profit Factor: {results.get('profit_factor', 0)}
- Avg Trade Duration: {results.get('avg_trade_duration_candles', 0)} candles
- Per-Window Sharpe: {results.get('per_window_sharpe', [])}
+- Cost Basis Improvement: {results.get('cost_basis_improvement_pct', 0):.1f}%
+- Avg Cost (Model): ${results.get('avg_cost_basis_model', 0):,.2f}
+- Avg Cost (DCA): ${results.get('avg_cost_basis_dca', 0):,.2f}
+- Strong Buy Signals: {results.get('strong_buy_signal_count', 0)}
+- Good Buy Signals: {results.get('good_buy_signal_count', 0)}
+- Signal Frequency: {results.get('signal_frequency_pct', 0):.1f}%
+- Quality of Strong Buys: {results.get('pct_quality_strong_buy', 0):.1%}
+- Model R2: {results.get('model_r2_score', 0):.4f}
+- Score at Actual Bottoms: {results.get('avg_score_at_actual_bottoms', 0):.1f}
+- Score at Actual Tops: {results.get('avg_score_at_actual_tops', 0):.1f}
+- Per-Window Improvement: {results.get('per_window_cost_improvement', [])}
+- Score Distribution: {results.get('score_distribution', {})}

 ## Top Feature Importances
 {json.dumps(dict(list(results.get('feature_importances', {}).items())[:15]), indent=2)}
 {history_text}
 Analyze these results and suggest 1-3 specific modifications to the config. Return ONLY valid JSON."""

-    # Call Ollama
-    payload = {
-        "model": MODEL,
-        "messages": [
-            {"role": "system", "content": SYSTEM_PROMPT},
-            {"role": "user", "content": user_prompt},
-        ],
-        "stream": False,
-        "options": {
-            "temperature": 0.7,
-            "num_predict": 4096,
-        },
-    }
+    messages = [
+        {"role": "system", "content": SYSTEM_PROMPT},
+        {"role": "user", "content": user_prompt},
+    ]

-    print(f"  Calling LLM ({MODEL} on Mac Mini)...")
-    resp = requests.post(f"{OLLAMA_URL}/api/chat", json=payload, timeout=300)
-    resp.raise_for_status()
-    content = resp.json()["message"]["content"]
+    content = call_llm(messages)

-    # Parse JSON from response (handle markdown code blocks)
    # Strip thinking tags if present
    content = re.sub(r"<think>.*?</think>", "", content, flags=re.DOTALL).strip()

@ -153,8 +309,6 @@ Analyze these results and suggest 1-3 specific modifications to the config. Retu
    if json_match:
        parsed = json.loads(json_match.group(1))
    else:
-        # Try parsing the whole response as JSON
-        # Find the outermost JSON object
        brace_start = content.find("{")
        if brace_start >= 0:
            depth = 0
@ -164,7 +318,7 @@ Analyze these results and suggest 1-3 specific modifications to the config. Retu
                elif content[i] == "}":
                    depth -= 1
                    if depth == 0:
-                        parsed = json.loads(content[brace_start:i + 1])
+                        parsed = json.loads(content[brace_start : i + 1])
                        break
            else:
                raise ValueError("Could not find complete JSON in LLM response")
@ -175,8 +329,14 @@ Analyze these results and suggest 1-3 specific modifications to the config. Retu
    changes = parsed.get("changes", [])
    new_config = parsed.get("config", current_config)

-    # Validate that config has required fields
-    required_keys = ["model_type", "features", "target", "hyperparameters", "strategy", "training"]
+    required_keys = [
+        "model_type",
+        "features",
+        "target",
+        "hyperparameters",
+        "strategy",
+        "training",
+    ]
    for key in required_keys:
        if key not in new_config:
            new_config[key] = current_config[key]
@ -186,22 +346,36 @@ Analyze these results and suggest 1-3 specific modifications to the config. Retu


 if __name__ == "__main__":
-    # Test with dummy data
    import sys
+
    config_path = sys.argv[1] if len(sys.argv) > 1 else "config/initial_config.json"
    with open(config_path) as f:
        config = json.load(f)

    dummy_results = {
-        "sharpe_ratio": 1.2,
-        "total_return_pct": 15.3,
-        "max_drawdown_pct": -12.5,
-        "win_rate": 0.55,
-        "trade_count": 120,
-        "profit_factor": 1.4,
-        "avg_trade_duration_candles": 7.2,
-        "feature_importances": {"RSI_14": 0.15, "MACD_hist": 0.12, "BB_width": 0.10},
-        "per_window_sharpe": [1.0, 1.3, 1.5, 0.9, 1.1],
+        "cost_basis_improvement_pct": 8.5,
+        "avg_cost_basis_model": 65000,
+        "avg_cost_basis_dca": 71000,
+        "strong_buy_signal_count": 45,
+        "good_buy_signal_count": 120,
+        "signal_frequency_pct": 7.2,
+        "pct_quality_strong_buy": 0.72,
+        "model_r2_score": 0.22,
+        "avg_score_at_actual_bottoms": 68.5,
+        "avg_score_at_actual_tops": 35.2,
+        "per_window_cost_improvement": [7.1, 9.3, 8.8, 10.2, 7.0],
+        "score_distribution": {
+            "0-20": 80,
+            "20-40": 150,
+            "40-60": 200,
+            "60-80": 130,
+            "80-100": 40,
+        },
+        "feature_importances": {
+            "dist_from_ath_pct": 0.18,
+            "RSI_14": 0.12,
+            "price_percentile_365": 0.10,
+        },
    }

    new_config, reasoning = analyze_and_suggest(config, dummy_results)
--- a/ml_engine/train_and_backtest.py
+++ b/ml_engine/train_and_backtest.py
--- a/orchestrator.py
+++ b/orchestrator.py
@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 """
-BTC ML Trading Strategy Optimizer — Orchestrator
+BTC Accumulation Signal Optimizer -- Orchestrator
 Coordinates the optimization loop across VPS, Windows PC (GPU), and Mac Mini (LLM).
 """

@ -28,7 +28,8 @@ MAC_MINI_HOST = "bizzle@bizzles-mac-mini-1"
 MAX_ITERATIONS = 50
 CONVERGENCE_WINDOW = 5
 CONVERGENCE_THRESHOLD = 0.01  # 1% improvement
-TARGET_SHARPE = 3.0
+TARGET_COST_IMPROVEMENT = 20.0  # 20% cost basis improvement = exceptional
+MIN_SIGNAL_COUNT = 30  # Minimum strong buy signals for valid results
 ML_TIMEOUT = 600  # 10 minutes

 # Colors
@ -98,7 +99,6 @@ def run_ml_training():
    )
    if result.returncode != 0:
        raise RuntimeError(f"ML training failed:\n{result.stderr}\n{result.stdout}")
-    # Print training output
    for line in result.stdout.strip().split("\n"):
        log(f"  {C.DIM}{line}", C.DIM)
    return True
@ -127,45 +127,53 @@ def check_convergence(history):
    if len(history) < CONVERGENCE_WINDOW + 1:
        return False, "Not enough iterations"

-    recent = history[-CONVERGENCE_WINDOW:]
-    sharpes = [h["sharpe"] for h in recent]
+    # Only consider valid results (enough signals)
+    valid = [h for h in history if h.get("signal_count", 0) >= MIN_SIGNAL_COUNT]

-    # Check if best sharpe exceeds target
-    best_sharpe = max(h["sharpe"] for h in history)
-    if best_sharpe >= TARGET_SHARPE:
-        return True, f"Target Sharpe reached: {best_sharpe:.3f}"
+    if not valid:
+        return False, "No valid results yet"
+
+    recent = history[-CONVERGENCE_WINDOW:]
+    scores = [h.get("cost_improvement", 0) for h in recent]
+
+    # Check if best score exceeds target
+    best_score = max(h.get("cost_improvement", 0) for h in valid)
+    if best_score >= TARGET_COST_IMPROVEMENT:
+        return True, f"Target cost improvement reached: {best_score:.1f}%"

    # Check if improvement has stalled
-    best_recent = max(sharpes)
-    worst_recent = min(sharpes)
+    best_recent = max(scores)
+    worst_recent = min(scores)
    if best_recent > 0 and (best_recent - worst_recent) / best_recent < CONVERGENCE_THRESHOLD:
-        return True, f"Converged: Sharpe variance < {CONVERGENCE_THRESHOLD*100}% over {CONVERGENCE_WINDOW} iterations"
+        return True, f"Converged: variance < {CONVERGENCE_THRESHOLD*100}% over {CONVERGENCE_WINDOW} iterations"

    return False, ""


 def print_header():
    print(f"""
-{C.BOLD}{C.CYAN}╔══════════════════════════════════════════════════╗
-║       BTC ML Trading Strategy Optimizer          ║
-║     VPS → Windows GPU → Mac Mini LLM → Loop      ║
-╚══════════════════════════════════════════════════╝{C.RESET}
+{C.BOLD}{C.CYAN}========================================================
+       BTC Accumulation Signal Optimizer
+     VPS -> Windows GPU -> Mac Mini LLM -> Loop
+========================================================{C.RESET}
 """)


 def print_results(results, iteration):
-    sharpe = results.get("sharpe_ratio", 0)
-    sharpe_color = C.GREEN if sharpe > 1.5 else C.YELLOW if sharpe > 1.0 else C.RED
+    cost_imp = results.get("cost_basis_improvement_pct", 0)
+    color = C.GREEN if cost_imp > 15 else C.YELLOW if cost_imp > 10 else C.RED
    print(f"""
-{C.BOLD}━━━ Iteration {iteration} Results ━━━{C.RESET}
-  Sharpe Ratio:    {sharpe_color}{C.BOLD}{sharpe:.3f}{C.RESET}
-  Total Return:    {results.get('total_return_pct', 0):.1f}%
-  Max Drawdown:    {results.get('max_drawdown_pct', 0):.1f}%
-  Win Rate:        {results.get('win_rate', 0):.1%}
-  Trade Count:     {results.get('trade_count', 0)}
-  Profit Factor:   {results.get('profit_factor', 0):.3f}
-  Avg Duration:    {results.get('avg_trade_duration_candles', 0):.1f} candles
-  Window Sharpes:  {results.get('per_window_sharpe', [])}
+{C.BOLD}--- Iteration {iteration} Results ---{C.RESET}
+  Cost Improvement: {color}{C.BOLD}{cost_imp:.1f}%{C.RESET}
+  Avg Cost (Model): ${results.get('avg_cost_basis_model', 0):,.2f}
+  Avg Cost (DCA):   ${results.get('avg_cost_basis_dca', 0):,.2f}
+  Strong Signals:   {results.get('strong_buy_signal_count', 0)}
+  Signal Frequency: {results.get('signal_frequency_pct', 0):.1f}%
+  Quality Score:    {results.get('pct_quality_strong_buy', 0):.1%}
+  Model R2:         {results.get('model_r2_score', 0):.4f}
+  Score@Bottoms:    {results.get('avg_score_at_actual_bottoms', 0):.1f}
+  Score@Tops:       {results.get('avg_score_at_actual_tops', 0):.1f}
+  Window Improvements: {results.get('per_window_cost_improvement', [])}
 """)


@ -173,14 +181,11 @@ def main():
    print_header()
    os.makedirs(RESULTS_DIR, exist_ok=True)

-    # Step 1: Ensure data
    ensure_data()

-    # Step 2: Load or create initial config
    config_path = os.path.join(CONFIG_DIR, "initial_config.json")
    best_config_path = os.path.join(CONFIG_DIR, "best_config.json")

-    # Resume from best config if it exists
    if os.path.exists(best_config_path):
        log("Resuming from best_config.json", C.GREEN)
        with open(best_config_path) as f:
@ -191,29 +196,24 @@ def main():

    history = load_iteration_history()
    start_iter = len(history) + 1
-    best_sharpe = max((h["sharpe"] for h in history), default=0)
+    best_score = max((h.get("cost_improvement", 0) for h in history), default=0)

-    log(f"Starting at iteration {start_iter}, best Sharpe so far: {best_sharpe:.3f}", C.BOLD)
+    log(f"Starting at iteration {start_iter}, best cost improvement so far: {best_score:.1f}%", C.BOLD)

-    # Step 3: Setup Windows remote
    setup_windows_remote()

-    # SCP the ML engine script (once)
    log("Uploading ML engine to Windows...", C.CYAN)
    scp_to_windows(os.path.join(BASE_DIR, "ml_engine", "train_and_backtest.py"), "train_and_backtest.py")

-    # SCP data files (once)
    for tf in ["1h", "4h"]:
        data_file = os.path.join(DATA_DIR, f"btc_{tf}.csv")
        if os.path.exists(data_file):
            log(f"Uploading btc_{tf}.csv to Windows...", C.CYAN)
            scp_to_windows(data_file, f"btc_{tf}.csv")

-    # Import LLM analyzer
    sys.path.insert(0, os.path.join(BASE_DIR, "llm_client"))
    from analyzer import analyze_and_suggest

-    # Main optimization loop
    for iteration in range(start_iter, MAX_ITERATIONS + 1):
        log(f"\n{'='*50}", C.BOLD)
        log(f"ITERATION {iteration}/{MAX_ITERATIONS}", f"{C.BOLD}{C.CYAN}")
@ -222,13 +222,11 @@ def main():
            f"Depth: {config.get('hyperparameters', {}).get('max_depth', '?')}", C.DIM)
        log(f"{'='*50}", C.BOLD)

-        # Write current config to temp file and SCP
        tmp_config = os.path.join(BASE_DIR, "config", "current_config.json")
        with open(tmp_config, "w") as f:
            json.dump(config, f, indent=2)
        scp_to_windows(tmp_config, "config.json")

-        # Run ML training on Windows
        try:
            run_ml_training()
        except (RuntimeError, subprocess.TimeoutExpired) as e:
@ -238,7 +236,6 @@ def main():
                config = history[-1].get("config", config)
            continue

-        # Fetch results from Windows
        results_local = os.path.join(RESULTS_DIR, f"results_iter_{iteration}.json")
        scp_from_windows("results.json", results_local)

@ -247,34 +244,35 @@ def main():

        print_results(results, iteration)

-        # Track best
-        current_sharpe = results.get("sharpe_ratio", 0)
-        is_best = current_sharpe > best_sharpe
+        current_score = results.get("cost_basis_improvement_pct", 0)
+        signal_count = results.get("strong_buy_signal_count", 0)
+        is_best = current_score > best_score and signal_count >= MIN_SIGNAL_COUNT

        if is_best:
-            best_sharpe = current_sharpe
+            best_score = current_score
            with open(best_config_path, "w") as f:
                json.dump(config, f, indent=2)
-            log(f"NEW BEST! Sharpe: {best_sharpe:.3f}", f"{C.BOLD}{C.GREEN}")
+            log(f"NEW BEST! Cost Improvement: {best_score:.1f}%", f"{C.BOLD}{C.GREEN}")

-        # Log iteration
        iter_data = {
            "iteration": iteration,
            "timestamp": datetime.now(timezone.utc).isoformat(),
-            "sharpe": current_sharpe,
-            "return": results.get("total_return_pct", 0),
-            "max_drawdown": results.get("max_drawdown_pct", 0),
-            "win_rate": results.get("win_rate", 0),
-            "trades": results.get("trade_count", 0),
-            "profit_factor": results.get("profit_factor", 0),
+            "cost_improvement": current_score,
+            "avg_30d_return": results.get("avg_quality_score_strong_buy", 0),
+            "avg_90d_return": results.get("pct_quality_strong_buy", 0),
+            "signal_count": signal_count,
+            "signal_frequency": results.get("signal_frequency_pct", 0),
+            "r2_score": results.get("model_r2_score", 0),
+            "score_at_bottoms": results.get("avg_score_at_actual_bottoms", 0),
+            "score_at_tops": results.get("avg_score_at_actual_tops", 0),
            "model_type": config.get("model_type", "unknown"),
            "is_best": is_best,
            "config": config,
+            "results": results,
        }
        save_iteration(iter_data)
        history.append(iter_data)

-        # Check convergence
        converged, reason = check_convergence(history)
        if converged:
            log(f"\nOptimization converged: {reason}", f"{C.BOLD}{C.GREEN}")
@ -284,17 +282,15 @@ def main():
            log(f"\nMax iterations ({MAX_ITERATIONS}) reached.", C.YELLOW)
            break

-        # Ask LLM for next config
        log("\nConsulting LLM for strategy modifications...", C.MAGENTA)
        try:
            summary_history = [
                {
                    "iteration": h["iteration"],
-                    "sharpe": h["sharpe"],
-                    "return": h["return"],
-                    "win_rate": h["win_rate"],
-                    "trades": h["trades"],
-                    "model_type": h["model_type"],
+                    "cost_improvement": h.get("cost_improvement", 0),
+                    "signal_count": h.get("signal_count", 0),
+                    "r2_score": h.get("r2_score", 0),
+                    "model_type": h.get("model_type", "unknown"),
                }
                for h in history
            ]
@ -304,37 +300,34 @@ def main():
        except Exception as e:
            log(f"LLM call failed: {e}", C.RED)
            log("Continuing with current config + random perturbation...", C.YELLOW)
-            # Small random perturbation as fallback
            import random
            hp = config.get("hyperparameters", {})
-            hp["learning_rate"] = hp.get("learning_rate", 0.05) * random.uniform(0.8, 1.2)
-            hp["max_depth"] = max(3, min(10, hp.get("max_depth", 6) + random.choice([-1, 0, 1])))
+            hp["learning_rate"] = hp.get("learning_rate", 0.01) * random.uniform(0.8, 1.2)
+            hp["max_depth"] = max(3, min(10, hp.get("max_depth", 5) + random.choice([-1, 0, 1])))
            config["hyperparameters"] = hp

-    # Final summary
    print(f"""
-{C.BOLD}{C.GREEN}╔══════════════════════════════════════════════════╗
-║              Optimization Complete!               ║
-╚══════════════════════════════════════════════════╝{C.RESET}
+{C.BOLD}{C.GREEN}========================================================
+              Optimization Complete!
+========================================================{C.RESET}

-  Total Iterations: {len(history)}
-  Best Sharpe:      {C.BOLD}{best_sharpe:.3f}{C.RESET}
-  Best Config:      {best_config_path}
-  Iteration Log:    {ITERATIONS_LOG}
+  Total Iterations:     {len(history)}
+  Best Cost Improvement: {C.BOLD}{best_score:.1f}%{C.RESET}
+  Best Config:          {best_config_path}
+  Iteration Log:        {ITERATIONS_LOG}
 """)


 # --- Library API for dashboard integration ---

-# Shared state for dashboard
 _stop_event = threading.Event()
 _status = {
-    "state": "idle",  # idle, running, completed, error
+    "state": "idle",
    "iteration": 0,
    "max_iterations": MAX_ITERATIONS,
-    "best_sharpe": 0.0,
+    "best_score": 0.0,
    "error": None,
-    "llm_suggestions": [],  # list of {iteration, reasoning, changes}
+    "llm_suggestions": [],
 }
 _status_lock = threading.Lock()

@ -352,15 +345,9 @@ def update_status(**kwargs):


 def run_optimization_loop(callback=None, config_override=None):
-    """
-    Run the optimization loop. Designed to be called from a background thread.
-
-    Args:
-        callback: Called after each iteration with (iteration_number, iter_data_dict).
-        config_override: Optional dict to use instead of loading from disk.
-    """
+    """Run the optimization loop from a background thread."""
    _stop_event.clear()
-    update_status(state="running", iteration=0, error=None, best_sharpe=0.0)
+    update_status(state="running", iteration=0, error=None, best_score=0.0)

    try:
        os.makedirs(RESULTS_DIR, exist_ok=True)
@ -380,8 +367,8 @@ def run_optimization_loop(callback=None, config_override=None):

        history = load_iteration_history()
        start_iter = len(history) + 1
-        best_sharpe = max((h["sharpe"] for h in history), default=0)
-        update_status(best_sharpe=best_sharpe)
+        best_score = max((h.get("cost_improvement", 0) for h in history), default=0)
+        update_status(best_score=best_score)

        setup_windows_remote()
        scp_to_windows(os.path.join(BASE_DIR, "ml_engine", "train_and_backtest.py"), "train_and_backtest.py")
@ -418,23 +405,26 @@ def run_optimization_loop(callback=None, config_override=None):
            with open(results_local) as f:
                results = json.load(f)

-            current_sharpe = results.get("sharpe_ratio", 0)
-            is_best = current_sharpe > best_sharpe
+            current_score = results.get("cost_basis_improvement_pct", 0)
+            signal_count = results.get("strong_buy_signal_count", 0)
+            is_best = current_score > best_score and signal_count >= MIN_SIGNAL_COUNT
+
            if is_best:
-                best_sharpe = current_sharpe
+                best_score = current_score
                with open(best_config_path, "w") as f:
                    json.dump(config, f, indent=2)
-                update_status(best_sharpe=best_sharpe)
+                update_status(best_score=best_score)

            iter_data = {
                "iteration": iteration,
                "timestamp": datetime.now(timezone.utc).isoformat(),
-                "sharpe": current_sharpe,
-                "return": results.get("total_return_pct", 0),
-                "max_drawdown": results.get("max_drawdown_pct", 0),
-                "win_rate": results.get("win_rate", 0),
-                "trades": results.get("trade_count", 0),
-                "profit_factor": results.get("profit_factor", 0),
+                "cost_improvement": current_score,
+                "signal_count": signal_count,
+                "signal_frequency": results.get("signal_frequency_pct", 0),
+                "r2_score": results.get("model_r2_score", 0),
+                "score_at_bottoms": results.get("avg_score_at_actual_bottoms", 0),
+                "score_at_tops": results.get("avg_score_at_actual_tops", 0),
+                "quality": results.get("pct_quality_strong_buy", 0),
                "model_type": config.get("model_type", "unknown"),
                "is_best": is_best,
                "config": config,
@ -459,10 +449,10 @@ def run_optimization_loop(callback=None, config_override=None):
                update_status(state="completed")
                return

-            # LLM suggestion
            try:
                summary_history = [
-                    {k: h[k] for k in ("iteration", "sharpe", "return", "win_rate", "trades", "model_type")}
+                    {k: h[k] for k in ("iteration", "cost_improvement", "signal_count", "r2_score", "model_type")
+                     if k in h}
                    for h in history
                ]
                new_config, reasoning = analyze_and_suggest(config, results, summary_history)
@ -471,12 +461,25 @@ def run_optimization_loop(callback=None, config_override=None):
                        "iteration": iteration,
                        "reasoning": reasoning,
                    })
+                # Also persist LLM suggestion to iteration log
+                iter_data["llm_reasoning"] = reasoning
+                iter_data["llm_applied"] = True
                config = new_config
-            except Exception:
-                import random
+            except Exception as e:
+                import random, traceback
+                err_msg = f"LLM call failed: {type(e).__name__}: {e}"
+                print(f"  WARNING: {err_msg}")
+                traceback.print_exc()
+                with _status_lock:
+                    _status["llm_suggestions"].append({
+                        "iteration": iteration,
+                        "reasoning": f"ERROR: {err_msg} — using random perturbation",
+                    })
+                iter_data["llm_reasoning"] = err_msg
+                iter_data["llm_applied"] = False
                hp = config.get("hyperparameters", {})
-                hp["learning_rate"] = hp.get("learning_rate", 0.05) * random.uniform(0.8, 1.2)
-                hp["max_depth"] = max(3, min(10, hp.get("max_depth", 6) + random.choice([-1, 0, 1])))
+                hp["learning_rate"] = hp.get("learning_rate", 0.01) * random.uniform(0.8, 1.2)
+                hp["max_depth"] = max(3, min(10, hp.get("max_depth", 5) + random.choice([-1, 0, 1])))
                config["hyperparameters"] = hp

        update_status(state="completed")
--- a/scoring/init.py
+++ b/scoring/init.py
--- a/scoring/engine.py
+++ b/scoring/engine.py
@ -0,0 +1,418 @@
+"""Scoring engine for Bitcoin accumulation zone metrics."""
+
+import json
+import os
+import logging
+
+log = logging.getLogger(__name__)
+
+THRESHOLDS_PATH = os.path.join(
+    os.path.dirname(os.path.dirname(os.path.abspath(__file__))),
+    "config",
+    "thresholds.json",
+)
+
+
+def load_thresholds():
+    """Load scoring thresholds from config."""
+    try:
+        with open(THRESHOLDS_PATH) as f:
+            return json.load(f)
+    except Exception:
+        return {}
+
+
+def _score_range(value, ranges):
+    """Score a value using range-based thresholds.
+    Each range is [low, high, score]. null means unbounded.
+    """
+    if value is None:
+        return None
+    for low, high, score in ranges:
+        low_ok = low is None or value >= low
+        high_ok = high is None or value < high
+        if low_ok and high_ok:
+            return score
+    return 0
+
+
+def _score_range_inverted(value, ranges):
+    """Score where higher value = lower range index (for drawdown)."""
+    if value is None:
+        return None
+    for low, high, score in ranges:
+        low_ok = low is None or value >= low
+        high_ok = high is None or value < high
+        if low_ok and high_ok:
+            return score
+    return 0
+
+
+def score_fear_greed(value, thresholds=None):
+    """Score Fear & Greed index (0-100 input, 0-10 output)."""
+    if value is None:
+        return None, "No data"
+    t = (thresholds or load_thresholds()).get("fear_greed", {})
+    ranges = t.get("ranges", [[0, 10, 10], [11, 25, 7], [26, 45, 4], [46, 55, 2], [56, 75, 1], [76, 100, 0]])
+    score = _score_range(value, ranges)
+
+    if value <= 10:
+        desc = "Extreme Fear — historically excellent buying"
+    elif value <= 25:
+        desc = "Fear — good accumulation territory"
+    elif value <= 45:
+        desc = "Low neutral — moderate opportunity"
+    elif value <= 55:
+        desc = "Neutral"
+    elif value <= 75:
+        desc = "Greed — caution"
+    else:
+        desc = "Extreme Greed — poor time to accumulate"
+    return score, desc
+
+
+def score_puell_multiple(value, thresholds=None):
+    if value is None:
+        return None, "No data"
+    t = (thresholds or load_thresholds()).get("puell_multiple", {})
+    ranges = t.get("ranges", [[None, 0.3, 10], [0.3, 0.5, 8], [0.5, 0.8, 5], [0.8, 1.2, 3], [1.2, 2.0, 1], [2.0, None, 0]])
+    score = _score_range(value, ranges)
+
+    if value < 0.3:
+        desc = "Deep value — miners under extreme stress"
+    elif value < 0.5:
+        desc = "Low — miners selling below average"
+    elif value < 0.8:
+        desc = "Below average miner revenue"
+    elif value < 1.2:
+        desc = "Average miner revenue"
+    elif value < 2.0:
+        desc = "Above average — miners profiting well"
+    else:
+        desc = "Elevated — potential top signal"
+    return score, desc
+
+
+def score_mvrv_zscore(value, thresholds=None):
+    if value is None:
+        return None, "No data"
+    t = (thresholds or load_thresholds()).get("mvrv_zscore", {})
+    ranges = t.get("ranges", [[None, 0, 10], [0, 0.5, 8], [0.5, 1.5, 5], [1.5, 3, 2], [3, 5, 1], [5, None, 0]])
+    score = _score_range(value, ranges)
+
+    if value < 0:
+        desc = "Below realized value — historically perfect buy zone"
+    elif value < 0.5:
+        desc = "Near realized value — strong accumulation"
+    elif value < 1.5:
+        desc = "Fair value range"
+    elif value < 3:
+        desc = "Above fair value"
+    elif value < 5:
+        desc = "Overvalued territory"
+    else:
+        desc = "Extreme overvaluation — cycle top territory"
+    return score, desc
+
+
+def score_drawdown(value, thresholds=None):
+    """Score drawdown from ATH (value is % drawdown, e.g. 50 = 50% below ATH)."""
+    if value is None:
+        return None, "No data"
+    t = (thresholds or load_thresholds()).get("drawdown", {})
+    ranges = t.get("ranges", [[70, None, 10], [50, 70, 8], [30, 50, 6], [20, 30, 4], [10, 20, 2], [None, 10, 0]])
+    score = _score_range(value, ranges)
+
+    if value > 70:
+        desc = f"{value:.0f}% below ATH — extreme capitulation"
+    elif value > 50:
+        desc = f"{value:.0f}% below ATH — deep bear market"
+    elif value > 30:
+        desc = f"{value:.0f}% below ATH — significant correction"
+    elif value > 20:
+        desc = f"{value:.0f}% below ATH — moderate pullback"
+    elif value > 10:
+        desc = f"{value:.0f}% below ATH — minor dip"
+    else:
+        desc = f"{value:.0f}% below ATH — near all-time high"
+    return score, desc
+
+
+def score_price_vs_200w_sma(price, sma_200w, thresholds=None):
+    """Score price relative to 200-week SMA."""
+    if price is None or sma_200w is None or sma_200w == 0:
+        return None, "No data"
+    pct_above = ((price - sma_200w) / sma_200w) * 100
+    t = (thresholds or load_thresholds()).get("price_vs_200w_sma", {})
+    ranges = t.get("ranges", [[None, 0, 10], [0, 20, 6], [20, 50, 3], [50, 100, 1], [100, None, 0]])
+    score = _score_range(pct_above, ranges)
+
+    if pct_above < 0:
+        desc = f"Below 200W SMA — historically rare buy zone"
+    elif pct_above < 20:
+        desc = f"{pct_above:.0f}% above 200W SMA — good value"
+    elif pct_above < 50:
+        desc = f"{pct_above:.0f}% above 200W SMA — moderate"
+    elif pct_above < 100:
+        desc = f"{pct_above:.0f}% above 200W SMA — extended"
+    else:
+        desc = f"{pct_above:.0f}% above 200W SMA — extremely overheated"
+    return score, desc
+
+
+def score_reserve_risk(value, thresholds=None):
+    if value is None:
+        return None, "No data"
+    t = (thresholds or load_thresholds()).get("reserve_risk", {})
+    ranges = t.get("ranges", [[None, 0.002, 10], [0.002, 0.005, 7], [0.005, 0.01, 4], [0.01, 0.02, 2], [0.02, None, 0]])
+    score = _score_range(value, ranges)
+
+    if value < 0.002:
+        desc = "Very low risk/reward — strong accumulation"
+    elif value < 0.005:
+        desc = "Low risk — good entry"
+    elif value < 0.01:
+        desc = "Moderate risk/reward"
+    elif value < 0.02:
+        desc = "Elevated risk"
+    else:
+        desc = "High risk — cycle top territory"
+    return score, desc
+
+
+def score_rhodl_ratio(value, thresholds=None):
+    if value is None:
+        return None, "No data"
+    t = (thresholds or load_thresholds()).get("rhodl_ratio", {})
+    ranges = t.get("ranges", [[None, 100, 10], [100, 500, 7], [500, 2000, 4], [2000, 10000, 1], [10000, None, 0]])
+    score = _score_range(value, ranges)
+
+    if value < 100:
+        desc = "Extreme low — long-term holders dominate"
+    elif value < 500:
+        desc = "Low — mature holder confidence"
+    elif value < 2000:
+        desc = "Moderate rotation"
+    elif value < 10000:
+        desc = "Elevated — new money entering"
+    else:
+        desc = "Extreme — speculative mania"
+    return score, desc
+
+
+def score_nupl(value, thresholds=None):
+    if value is None:
+        return None, "No data"
+    t = (thresholds or load_thresholds()).get("nupl", {})
+    ranges = t.get("ranges", [[None, 0, 10], [0, 0.25, 7], [0.25, 0.5, 4], [0.5, 0.75, 1], [0.75, None, 0]])
+    score = _score_range(value, ranges)
+
+    if value < 0:
+        desc = "Capitulation — holders underwater"
+    elif value < 0.25:
+        desc = "Hope/Fear — early recovery"
+    elif value < 0.5:
+        desc = "Optimism — moderate profit taking"
+    elif value < 0.75:
+        desc = "Belief/Greed — significant unrealized gains"
+    else:
+        desc = "Euphoria — extreme unrealized profit"
+    return score, desc
+
+
+def score_lth_realized_price(price, lth_rp, thresholds=None):
+    """Score price relative to Long-Term Holder realized price."""
+    if price is None or lth_rp is None or lth_rp == 0:
+        return None, "No data"
+    pct_above = ((price - lth_rp) / lth_rp) * 100
+    t = (thresholds or load_thresholds()).get("lth_realized_price", {})
+    ranges = t.get("ranges", [[None, 0, 10], [0, 20, 6], [20, 50, 3], [50, None, 1]])
+    score = _score_range(pct_above, ranges)
+
+    if pct_above < 0:
+        desc = f"Below LTH cost basis — LTHs underwater (extreme value)"
+    elif pct_above < 20:
+        desc = f"{pct_above:.0f}% above LTH cost basis — good value"
+    elif pct_above < 50:
+        desc = f"{pct_above:.0f}% above LTH cost basis — moderate"
+    else:
+        desc = f"{pct_above:.0f}% above LTH cost basis — extended"
+    return score, desc
+
+
+def score_hash_ribbons(data, thresholds=None):
+    """Score hash ribbons based on buy signal detection."""
+    if not data:
+        return None, "No data"
+    if data.get("buy_signal"):
+        return 10, "Active buy signal — miner capitulation recovery"
+    return 3, "Normal mining activity"
+
+
+def score_all(metrics):
+    """Score all metrics and return individual + composite scores."""
+    thresholds = load_thresholds()
+    results = []
+
+    # Fear & Greed
+    fg = metrics.get("fear_greed", {})
+    fg_score, fg_desc = score_fear_greed(fg.get("value"), thresholds)
+    results.append({
+        "name": "Fear & Greed Index",
+        "key": "fear_greed",
+        "value": fg.get("value"),
+        "display_value": f"{fg.get('value', 'N/A')} — {fg.get('classification', '')}",
+        "score": fg_score,
+        "description": fg_desc,
+        "recent": fg.get("recent", []),
+    })
+
+    # Puell Multiple
+    pm = metrics.get("puell_multiple", {})
+    pm_score, pm_desc = score_puell_multiple(pm.get("value"), thresholds)
+    results.append({
+        "name": "Puell Multiple",
+        "key": "puell_multiple",
+        "value": pm.get("value"),
+        "display_value": f"{pm.get('value', 'N/A'):.4f}" if pm.get("value") is not None else "N/A",
+        "score": pm_score,
+        "description": pm_desc,
+        "recent": pm.get("recent", []),
+    })
+
+    # MVRV Z-Score
+    mz = metrics.get("mvrv_zscore", {})
+    mz_score, mz_desc = score_mvrv_zscore(mz.get("value"), thresholds)
+    results.append({
+        "name": "MVRV Z-Score",
+        "key": "mvrv_zscore",
+        "value": mz.get("value"),
+        "display_value": f"{mz.get('value', 'N/A'):.2f}" if mz.get("value") is not None else "N/A",
+        "score": mz_score,
+        "description": mz_desc,
+        "recent": mz.get("recent", []),
+    })
+
+    # Drawdown from ATH
+    dd = metrics.get("drawdown", {})
+    dd_score, dd_desc = score_drawdown(dd.get("value"), thresholds)
+    results.append({
+        "name": "Drawdown from ATH",
+        "key": "drawdown",
+        "value": dd.get("value"),
+        "display_value": f"{dd.get('value', 0):.1f}%" if dd.get("value") is not None else "N/A",
+        "score": dd_score,
+        "description": dd_desc,
+        "recent": [],
+    })
+
+    # Price vs 200W SMA
+    sma = metrics.get("200w_sma", {})
+    price_data = metrics.get("price", {})
+    current_price = price_data.get("price") or sma.get("btc_price")
+    sma_val = sma.get("value")
+    sma_score, sma_desc = score_price_vs_200w_sma(current_price, sma_val, thresholds)
+    results.append({
+        "name": "Price vs 200W SMA",
+        "key": "price_vs_200w_sma",
+        "value": sma_val,
+        "display_value": f"${sma_val:,.0f}" if sma_val else "N/A",
+        "score": sma_score,
+        "description": sma_desc,
+        "recent": sma.get("recent", []),
+    })
+
+    # Reserve Risk
+    rr = metrics.get("reserve_risk", {})
+    rr_score, rr_desc = score_reserve_risk(rr.get("value"), thresholds)
+    results.append({
+        "name": "Reserve Risk",
+        "key": "reserve_risk",
+        "value": rr.get("value"),
+        "display_value": f"{rr.get('value', 'N/A'):.6f}" if rr.get("value") is not None else "N/A",
+        "score": rr_score,
+        "description": rr_desc,
+        "recent": rr.get("recent", []),
+    })
+
+    # RHODL Ratio
+    rh = metrics.get("rhodl_ratio", {})
+    rh_score, rh_desc = score_rhodl_ratio(rh.get("value"), thresholds)
+    results.append({
+        "name": "RHODL Ratio",
+        "key": "rhodl_ratio",
+        "value": rh.get("value"),
+        "display_value": f"{rh.get('value', 'N/A'):.0f}" if rh.get("value") is not None else "N/A",
+        "score": rh_score,
+        "description": rh_desc,
+        "recent": rh.get("recent", []),
+    })
+
+    # NUPL
+    nu = metrics.get("nupl", {})
+    nu_score, nu_desc = score_nupl(nu.get("value"), thresholds)
+    results.append({
+        "name": "Net Unrealized Profit/Loss",
+        "key": "nupl",
+        "value": nu.get("value"),
+        "display_value": f"{nu.get('value', 'N/A'):.4f}" if nu.get("value") is not None else "N/A",
+        "score": nu_score,
+        "description": nu_desc,
+        "recent": nu.get("recent", []),
+    })
+
+    # LTH Realized Price
+    lth = metrics.get("lth_realized_price", {})
+    lth_price = lth.get("btc_price") or current_price
+    lth_rp = lth.get("value")
+    lth_score, lth_desc = score_lth_realized_price(lth_price, lth_rp, thresholds)
+    results.append({
+        "name": "LTH Realized Price",
+        "key": "lth_realized_price",
+        "value": lth_rp,
+        "display_value": f"${lth_rp:,.0f}" if lth_rp else "N/A",
+        "score": lth_score,
+        "description": lth_desc,
+        "recent": lth.get("recent", []),
+    })
+
+    # Hash Ribbons
+    hr = metrics.get("hash_ribbons", {})
+    hr_score, hr_desc = score_hash_ribbons(hr, thresholds)
+    results.append({
+        "name": "Hash Ribbons",
+        "key": "hash_ribbons",
+        "value": None,
+        "display_value": "Buy Signal" if hr.get("buy_signal") else "Normal",
+        "score": hr_score,
+        "description": hr_desc,
+        "recent": [],
+    })
+
+    # Compute composite
+    valid_scores = [r["score"] for r in results if r["score"] is not None]
+    if valid_scores:
+        # Scale to 0-100 based on available metrics
+        composite = sum(valid_scores) / len(valid_scores) * 10
+    else:
+        composite = 0
+
+    # Assessment text
+    if composite >= 71:
+        assessment = "STRONG ACCUMULATION ZONE"
+    elif composite >= 51:
+        assessment = "MODERATE OPPORTUNITY"
+    elif composite >= 31:
+        assessment = "NEUTRAL"
+    elif composite >= 15:
+        assessment = "CAUTION — OVERHEATED"
+    else:
+        assessment = "EXTREME CAUTION"
+
+    return {
+        "metrics": results,
+        "composite_score": round(composite, 1),
+        "assessment": assessment,
+        "scored_count": len(valid_scores),
+        "total_count": len(results),
+    }
--- a/scrapers/init.py
+++ b/scrapers/init.py
--- a/scrapers/fear_greed.py
+++ b/scrapers/fear_greed.py
@ -0,0 +1,34 @@
+"""Fear & Greed Index from alternative.me API."""
+
+import logging
+import requests
+
+log = logging.getLogger(__name__)
+
+FNG_URL = "https://api.alternative.me/fng/?limit=30"
+
+
+def fetch():
+    """Fetch Fear & Greed data. Returns dict with value, classification, and recent history."""
+    try:
+        resp = requests.get(FNG_URL, timeout=15)
+        resp.raise_for_status()
+        data = resp.json()
+        entries = data.get("data", [])
+        if not entries:
+            return {"value": None, "error": "No data"}
+
+        current = entries[0]
+        value = int(current["value"])
+        classification = current.get("value_classification", "")
+
+        recent = [int(e["value"]) for e in entries[:30]]
+
+        return {
+            "value": value,
+            "classification": classification,
+            "recent": recent,
+        }
+    except Exception as e:
+        log.error("Fear & Greed fetch error: %s", e)
+        return {"value": None, "error": str(e)}
--- a/scrapers/history_collector.py
+++ b/scrapers/history_collector.py
@ -0,0 +1,276 @@
+"""Collect full historical time series from LookIntoBitcoin charts, CoinGecko, and Fear & Greed."""
+
+import json
+import logging
+import os
+import time
+from datetime import datetime
+
+import requests
+
+log = logging.getLogger(__name__)
+
+BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+HISTORY_PATH = os.path.join(BASE_DIR, "data", "history.json")
+
+# Charts to scrape with expected trace names
+CHART_CONFIGS = {
+    "puell_multiple": {
+        "path": "/charts/puell-multiple/",
+        "traces": {"puell_multiple": "Puell Multiple", "btc_price": "Price"},
+    },
+    "mvrv_zscore": {
+        "path": "/charts/mvrv-zscore/",
+        "traces": {"mvrv_zscore": "Z-Score"},
+    },
+    "reserve_risk": {
+        "path": "/charts/reserve-risk/",
+        "traces": {"reserve_risk": "Reserve Risk"},
+    },
+    "rhodl_ratio": {
+        "path": "/charts/rhodl-ratio/",
+        "traces": {"rhodl_ratio": "RHODL Ratio"},
+    },
+    "nupl": {
+        "path": "/charts/relative-unrealized-profit--loss/",
+        "traces": {"nupl": "NUPL"},
+    },
+    "200w_sma": {
+        "path": "/charts/200-week-moving-average-heatmap/",
+        "traces": {"200w_sma": "200 Week Moving Average", "btc_price_sma": "Price"},
+    },
+    "lth_realized_price": {
+        "path": "/charts/long-term-holder-realized-price/",
+        "traces": {"lth_realized_price": "Long-Term Holder Realized Price", "btc_price_lth": "Price"},
+    },
+    "lth_supply": {
+        "path": "/charts/long-term-holder-supply/",
+        "traces": {"lth_supply": None},  # None = grab first numeric trace
+    },
+}
+
+
+def _find_trace(traces, name):
+    """Find a trace by name (case-insensitive partial match)."""
+    if not traces or not name:
+        return None
+    name_lower = name.lower()
+    for t in traces:
+        trace_name = t.get("name", "").lower()
+        if name_lower in trace_name or trace_name in name_lower:
+            return t
+    words = name_lower.split()
+    for t in traces:
+        trace_name = t.get("name", "").lower()
+        if all(w in trace_name for w in words):
+            return t
+    return None
+
+
+def _extract_series(trace):
+    """Extract (dates, values) from a Plotly trace dict."""
+    if not trace:
+        return [], []
+    x = trace.get("x", [])
+    y = trace.get("y", [])
+    dates = []
+    values = []
+    for i, (d, v) in enumerate(zip(x, y)):
+        if v is None:
+            continue
+        try:
+            val = float(v)
+        except (ValueError, TypeError):
+            continue
+        # Normalize date string to YYYY-MM-DD
+        date_str = str(d)[:10]
+        dates.append(date_str)
+        values.append(val)
+    return dates, values
+
+
+def scrape_chart_history(chart_path):
+    """Scrape a chart and return all trace data."""
+    from scrapers.lookintobitcoin import scrape_chart
+    return scrape_chart(chart_path)
+
+
+def collect_onchain_history(progress_cb=None):
+    """Scrape all on-chain charts and return dict of {metric: {dates, values}}."""
+    result = {}
+    total = len(CHART_CONFIGS)
+
+    for idx, (chart_key, cfg) in enumerate(CHART_CONFIGS.items()):
+        label = f"[{idx+1}/{total}] {chart_key}"
+        log.info("Scraping history: %s", label)
+        if progress_cb:
+            progress_cb(chart_key, idx, total)
+
+        try:
+            traces = scrape_chart_history(cfg["path"])
+            if not traces:
+                log.warning("No traces for %s", chart_key)
+                continue
+
+            for metric_key, trace_name in cfg["traces"].items():
+                if trace_name is None:
+                    # Grab first trace with numeric data
+                    for candidate in traces:
+                        y = candidate.get("y", [])
+                        if y and any(v is not None for v in y[-10:]):
+                            dates, values = _extract_series(candidate)
+                            if dates:
+                                result[metric_key] = {"dates": dates, "values": values}
+                                log.info("  %s: %d data points", metric_key, len(dates))
+                            break
+                else:
+                    t = _find_trace(traces, trace_name)
+                    if not t:
+                        # Fallback: try BTC Price
+                        if "btc_price" in metric_key or "price" in trace_name.lower():
+                            t = _find_trace(traces, "BTC") or _find_trace(traces, "Price")
+                        if not t:
+                            log.warning("  Trace '%s' not found for %s", trace_name, metric_key)
+                            continue
+                    dates, values = _extract_series(t)
+                    if dates:
+                        result[metric_key] = {"dates": dates, "values": values}
+                        log.info("  %s: %d data points (%s to %s)", metric_key, len(dates), dates[0], dates[-1])
+                    else:
+                        log.warning("  %s: no valid data points", metric_key)
+
+        except Exception as e:
+            log.error("Error scraping %s: %s", chart_key, e)
+
+        # Be polite between requests
+        if idx < total - 1:
+            time.sleep(2)
+
+    return result
+
+
+def collect_price_history():
+    """Fetch BTC price history from CoinGecko (max history)."""
+    log.info("Fetching BTC price history from CoinGecko...")
+    try:
+        resp = requests.get(
+            "https://api.coingecko.com/api/v3/coins/bitcoin/market_chart",
+            params={"vs_currency": "usd", "days": "max"},
+            timeout=30,
+        )
+        resp.raise_for_status()
+        data = resp.json()
+        prices = data.get("prices", [])
+        dates = []
+        values = []
+        seen_dates = set()
+        for ts_ms, price in prices:
+            d = datetime.utcfromtimestamp(ts_ms / 1000).strftime("%Y-%m-%d")
+            if d not in seen_dates:
+                seen_dates.add(d)
+                dates.append(d)
+                values.append(round(price, 2))
+        log.info("CoinGecko BTC price: %d days (%s to %s)", len(dates), dates[0] if dates else "?", dates[-1] if dates else "?")
+        return {"dates": dates, "values": values}
+    except Exception as e:
+        log.error("CoinGecko price fetch failed: %s", e)
+        return None
+
+
+def collect_fear_greed_history():
+    """Fetch full Fear & Greed history from alternative.me."""
+    log.info("Fetching Fear & Greed history...")
+    try:
+        resp = requests.get(
+            "https://api.alternative.me/fng/",
+            params={"limit": "0"},
+            timeout=30,
+        )
+        resp.raise_for_status()
+        data = resp.json().get("data", [])
+        dates = []
+        values = []
+        for entry in reversed(data):  # API returns newest first
+            ts = int(entry["timestamp"])
+            d = datetime.utcfromtimestamp(ts).strftime("%Y-%m-%d")
+            dates.append(d)
+            values.append(int(entry["value"]))
+        log.info("Fear & Greed: %d days (%s to %s)", len(dates), dates[0] if dates else "?", dates[-1] if dates else "?")
+        return {"dates": dates, "values": values}
+    except Exception as e:
+        log.error("Fear & Greed fetch failed: %s", e)
+        return None
+
+
+def collect_all_history(progress_cb=None):
+    """Collect all historical data and save to history.json."""
+    log.info("=== Starting full historical data collection ===")
+    history = {}
+
+    # 1. On-chain metrics from LookIntoBitcoin
+    onchain = collect_onchain_history(progress_cb=progress_cb)
+    history.update(onchain)
+
+    # 2. BTC price from CoinGecko
+    price = collect_price_history()
+    if price:
+        history["btc_price_coingecko"] = price
+
+    # 3. Fear & Greed
+    fng = collect_fear_greed_history()
+    if fng:
+        history["fear_greed"] = fng
+
+    # Merge BTC price: prefer the LookIntoBitcoin trace (goes to 2010), fill gaps with CoinGecko
+    btc_keys = [k for k in history if "btc_price" in k]
+    if btc_keys:
+        # Use longest series as base
+        best = max(btc_keys, key=lambda k: len(history[k]["dates"]))
+        history["btc_price"] = history[best]
+        log.info("BTC price source: %s (%d days)", best, len(history[best]["dates"]))
+
+    # Add metadata
+    history["_metadata"] = {
+        "collected_at": datetime.utcnow().isoformat() + "Z",
+        "metrics": list(k for k in history if not k.startswith("_")),
+        "metric_counts": {k: len(v["dates"]) for k, v in history.items() if isinstance(v, dict) and "dates" in v},
+    }
+
+    # Save
+    os.makedirs(os.path.dirname(HISTORY_PATH), exist_ok=True)
+    with open(HISTORY_PATH, "w") as f:
+        json.dump(history, f, separators=(",", ":"))
+
+    size_mb = os.path.getsize(HISTORY_PATH) / 1024 / 1024
+    log.info("=== History saved to %s (%.1f MB) ===", HISTORY_PATH, size_mb)
+    log.info("Metrics collected: %s", ", ".join(k for k in history if not k.startswith("_")))
+
+    return history
+
+
+def load_history():
+    """Load history from disk."""
+    if not os.path.exists(HISTORY_PATH):
+        return None
+    with open(HISTORY_PATH) as f:
+        return json.load(f)
+
+
+def history_status():
+    """Check if history exists and return metadata."""
+    if not os.path.exists(HISTORY_PATH):
+        return {"exists": False}
+    try:
+        stat = os.stat(HISTORY_PATH)
+        with open(HISTORY_PATH) as f:
+            data = json.load(f)
+        meta = data.get("_metadata", {})
+        return {
+            "exists": True,
+            "collected_at": meta.get("collected_at"),
+            "metrics": meta.get("metrics", []),
+            "metric_counts": meta.get("metric_counts", {}),
+            "size_mb": round(stat.st_size / 1024 / 1024, 2),
+        }
+    except Exception as e:
+        return {"exists": True, "error": str(e)}
--- a/scrapers/lookintobitcoin.py
+++ b/scrapers/lookintobitcoin.py
@ -0,0 +1,265 @@
+"""Playwright scraper for LookIntoBitcoin / BitcoinMagazinePro charts."""
+
+import logging
+import traceback
+
+log = logging.getLogger(__name__)
+
+BASE_URL = "https://www.lookintobitcoin.com"
+
+CHARTS = {
+    "puell_multiple": {
+        "path": "/charts/puell-multiple/",
+        "traces": ["Puell Multiple"],
+    },
+    "mvrv_zscore": {
+        "path": "/charts/mvrv-zscore/",
+        "traces": ["Z-Score"],
+    },
+    "reserve_risk": {
+        "path": "/charts/reserve-risk/",
+        "traces": ["Reserve Risk"],
+    },
+    "rhodl_ratio": {
+        "path": "/charts/rhodl-ratio/",
+        "traces": ["RHODL Ratio"],
+    },
+    "nupl": {
+        "path": "/charts/relative-unrealized-profit--loss/",
+        "traces": ["NUPL"],
+    },
+    "200w_sma": {
+        "path": "/charts/200-week-moving-average-heatmap/",
+        "traces": ["200 Week Moving Average"],
+    },
+    "lth_realized_price": {
+        "path": "/charts/long-term-holder-realized-price/",
+        "traces": ["Long-Term Holder Realized Price", "BTC Price"],
+    },
+    "hash_ribbons": {
+        "path": "/charts/hash-ribbons/",
+        "traces": None,
+    },
+    "pi_cycle_bottom": {
+        "path": "/charts/pi-cycle-top-bottom-indicator/",
+        "traces": None,
+    },
+    "lth_supply": {
+        "path": "/charts/long-term-holder-supply/",
+        "traces": None,
+    },
+}
+
+
+def scrape_chart(chart_path, timeout=25000):
+    """Scrape a single chart from LookIntoBitcoin. Returns list of trace dicts or None."""
+    from playwright.sync_api import sync_playwright
+
+    store = {"data": None}
+
+    with sync_playwright() as p:
+        browser = p.chromium.launch(headless=True)
+        page = browser.new_page()
+
+        def handle_response(response):
+            if "_dash-update-component" in response.url:
+                try:
+                    store["data"] = response.json()
+                except Exception:
+                    pass
+
+        page.on("response", handle_response)
+        try:
+            page.goto(f"{BASE_URL}{chart_path}", timeout=timeout)
+            page.wait_for_timeout(6000)
+        except Exception as e:
+            log.warning("Navigation error for %s: %s", chart_path, e)
+        finally:
+            browser.close()
+
+    if store["data"]:
+        try:
+            return store["data"]["response"]["chart"]["figure"]["data"]
+        except (KeyError, TypeError):
+            # Try alternate response structures
+            try:
+                resp = store["data"]
+                if isinstance(resp, dict):
+                    for key in resp:
+                        val = resp[key]
+                        if isinstance(val, dict) and "figure" in val:
+                            return val["figure"]["data"]
+                        if isinstance(val, dict) and "chart" in val:
+                            return val["chart"]["figure"]["data"]
+            except Exception:
+                pass
+    return None
+
+
+def _find_trace(traces, name):
+    """Find a trace by name (case-insensitive partial match)."""
+    if not traces:
+        return None
+    name_lower = name.lower()
+    # First pass: exact or substring match
+    for t in traces:
+        trace_name = t.get("name", "").lower()
+        if name_lower in trace_name or trace_name in name_lower:
+            return t
+    # Second pass: check if all words in name appear in trace name
+    words = name_lower.split()
+    for t in traces:
+        trace_name = t.get("name", "").lower()
+        if all(w in trace_name for w in words):
+            return t
+    return None
+
+
+def _get_latest_value(trace):
+    """Get the most recent non-null y value from a trace."""
+    if not trace:
+        return None
+    y = trace.get("y", [])
+    for val in reversed(y):
+        if val is not None:
+            try:
+                return float(val)
+            except (ValueError, TypeError):
+                continue
+    return None
+
+
+def _get_recent_values(trace, n=30):
+    """Get the last n non-null values from a trace."""
+    if not trace:
+        return []
+    y = trace.get("y", [])
+    values = []
+    for val in reversed(y):
+        if val is not None:
+            try:
+                values.append(float(val))
+            except (ValueError, TypeError):
+                continue
+        if len(values) >= n:
+            break
+    values.reverse()
+    return values
+
+
+def scrape_all():
+    """Scrape all charts and return parsed metric values."""
+    results = {}
+
+    for metric_key, chart_info in CHARTS.items():
+        log.info("Scraping %s ...", metric_key)
+        try:
+            traces = scrape_chart(chart_info["path"])
+            if not traces:
+                log.warning("No data for %s", metric_key)
+                results[metric_key] = {"value": None, "error": "No data returned"}
+                continue
+
+            wanted = chart_info.get("traces")
+
+            if metric_key == "puell_multiple":
+                t = _find_trace(traces, "Puell Multiple")
+                val = _get_latest_value(t)
+                results[metric_key] = {
+                    "value": val,
+                    "recent": _get_recent_values(t),
+                }
+
+            elif metric_key == "mvrv_zscore":
+                t = _find_trace(traces, "Z-Score")
+                val = _get_latest_value(t)
+                results[metric_key] = {
+                    "value": val,
+                    "recent": _get_recent_values(t),
+                }
+
+            elif metric_key == "200w_sma":
+                t = _find_trace(traces, "200 Week Moving Average") or _find_trace(traces, "200 Week MA") or _find_trace(traces, "200W")
+                val = _get_latest_value(t)
+                # Also try to find BTC price trace
+                price_t = _find_trace(traces, "BTC Price") or _find_trace(traces, "Price")
+                price_val = _get_latest_value(price_t)
+                results[metric_key] = {
+                    "value": val,
+                    "btc_price": price_val,
+                    "recent": _get_recent_values(t),
+                }
+
+            elif metric_key == "lth_realized_price":
+                lth_t = _find_trace(traces, "Long-Term Holder Realized Price") or _find_trace(traces, "LTH Realized Price") or _find_trace(traces, "LTH")
+                price_t = _find_trace(traces, "BTC Price") or _find_trace(traces, "Price")
+                lth_val = _get_latest_value(lth_t)
+                price_val = _get_latest_value(price_t)
+                results[metric_key] = {
+                    "value": lth_val,
+                    "btc_price": price_val,
+                    "recent": _get_recent_values(lth_t),
+                }
+
+            elif metric_key == "hash_ribbons":
+                # Look for buy/sell signal traces or MA crossover
+                results[metric_key] = {
+                    "traces": [
+                        {"name": t.get("name", ""), "latest": _get_latest_value(t)}
+                        for t in traces[:6]
+                    ],
+                    "value": None,
+                }
+                # Try to detect buy signal from trace names/colors
+                for t in traces:
+                    name = t.get("name", "").lower()
+                    if "buy" in name or "signal" in name:
+                        results[metric_key]["buy_signal"] = True
+                        break
+
+            elif metric_key == "lth_supply":
+                # Get main supply trace
+                t = traces[0] if traces else None
+                for candidate in traces:
+                    name = candidate.get("name", "").lower()
+                    if "supply" in name or "lth" in name:
+                        t = candidate
+                        break
+                recent = _get_recent_values(t, 60)
+                # Determine trend: compare recent avg to older avg
+                trend = None
+                if len(recent) >= 30:
+                    old_avg = sum(recent[:15]) / 15
+                    new_avg = sum(recent[-15:]) / 15
+                    trend = "increasing" if new_avg > old_avg else "decreasing"
+                results[metric_key] = {
+                    "value": _get_latest_value(t),
+                    "trend": trend,
+                    "recent": _get_recent_values(t),
+                }
+
+            else:
+                # Generic: grab first non-layout trace with numeric data
+                t = None
+                if wanted:
+                    for name in wanted:
+                        t = _find_trace(traces, name)
+                        if t:
+                            break
+                if not t:
+                    for candidate in traces:
+                        y = candidate.get("y", [])
+                        if y and any(v is not None for v in y[-10:]):
+                            t = candidate
+                            break
+                val = _get_latest_value(t)
+                results[metric_key] = {
+                    "value": val,
+                    "recent": _get_recent_values(t),
+                }
+
+        except Exception as e:
+            log.error("Error scraping %s: %s\n%s", metric_key, e, traceback.format_exc())
+            results[metric_key] = {"value": None, "error": str(e)}
+
+    return results
--- a/scrapers/price.py
+++ b/scrapers/price.py
@ -0,0 +1,80 @@
+"""BTC price data from CoinGecko API."""
+
+import logging
+import requests
+
+log = logging.getLogger(__name__)
+
+PRICE_URL = "https://api.coingecko.com/api/v3/simple/price?ids=bitcoin&vs_currencies=usd&include_24hr_change=true"
+HISTORY_URL = "https://api.coingecko.com/api/v3/coins/bitcoin/market_chart?vs_currency=usd&days=365"
+ATH_URL = "https://api.coingecko.com/api/v3/coins/bitcoin?localization=false&tickers=false&market_data=true&community_data=false&developer_data=false"
+
+
+def fetch_current():
+    """Fetch current BTC price and 24h change."""
+    try:
+        resp = requests.get(PRICE_URL, timeout=15)
+        resp.raise_for_status()
+        data = resp.json()
+        btc = data.get("bitcoin", {})
+        return {
+            "price": btc.get("usd"),
+            "change_24h": btc.get("usd_24h_change"),
+        }
+    except Exception as e:
+        log.error("Price fetch error: %s", e)
+        return {"price": None, "error": str(e)}
+
+
+def fetch_historical():
+    """Fetch 365 days of BTC price history. Returns list of [timestamp, price]."""
+    try:
+        resp = requests.get(HISTORY_URL, timeout=30)
+        resp.raise_for_status()
+        data = resp.json()
+        prices = data.get("prices", [])
+        return prices
+    except Exception as e:
+        log.error("Historical price fetch error: %s", e)
+        return []
+
+
+def fetch_ath():
+    """Fetch BTC all-time high from CoinGecko."""
+    try:
+        resp = requests.get(ATH_URL, timeout=15)
+        resp.raise_for_status()
+        data = resp.json()
+        market = data.get("market_data", {})
+        ath = market.get("ath", {}).get("usd")
+        ath_change = market.get("ath_change_percentage", {}).get("usd")
+        return {
+            "ath": ath,
+            "ath_change_pct": ath_change,
+        }
+    except Exception as e:
+        log.error("ATH fetch error: %s", e)
+        return {"ath": None, "error": str(e)}
+
+
+def calculate_200d_sma(prices):
+    """Calculate 200-day SMA from historical price data."""
+    if not prices or len(prices) < 200:
+        return None
+    # prices is [[timestamp, price], ...]
+    recent_200 = [p[1] for p in prices[-200:]]
+    return sum(recent_200) / len(recent_200)
+
+
+def calculate_mayer_multiple(current_price, sma_200d):
+    """Mayer Multiple = current price / 200-day SMA."""
+    if not current_price or not sma_200d or sma_200d == 0:
+        return None
+    return current_price / sma_200d
+
+
+def calculate_drawdown(current_price, ath):
+    """Drawdown from ATH as percentage."""
+    if not current_price or not ath or ath == 0:
+        return None
+    return (ath - current_price) / ath * 100
Author	SHA1	Message	Date
BizzleBot	13bac5f654	v4: Bitcoin Accumulation Zone Monitor — on-chain metrics + backtest engine COMPLETE PIVOT from ML trading optimizer to on-chain metrics monitor. Architecture: - Playwright scrapes LookIntoBitcoin Plotly Dash charts for real on-chain data - 10 proven metrics: Puell Multiple, MVRV Z-Score, Fear & Greed, Reserve Risk, RHODL Ratio, NUPL, LTH Realized Price, 200W SMA, Hash Ribbons, Drawdown - Each metric scores 0-10, composite 0-100 - No ML, no black box — every signal transparent and traceable - Historical backtest validates scoring against actual BTC forward returns - Recency-weighted analysis accounts for diminishing cycle returns Full documentation in ARCHITECTURE.md	2026-03-20 23:07:53 +00:00
BizzleBot	5b3b3811ec	feat: add historical backtest engine and dashboard page - scrapers/history_collector.py: scrapes full time series from 8 LookIntoBitcoin charts + Fear & Greed API, stores to data/history.json (~5700 days back to 2010) - backtesting/engine.py: scores each historical day using same thresholds as live scoring, computes 30d/90d/180d/1yr forward returns, bracket stats, signal events - dashboard/server.py: adds /backtest page with dual-axis score vs price chart, bracket performance table, signal event list, current context box; adds backtest nav link and historical context box on main dashboard; 4 new API endpoints Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-03-20 22:50:57 +00:00
BizzleBot	e3c5aa9f32	chore: add .gitignore for pycache and data dirs Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-03-20 22:31:35 +00:00
BizzleBot	62e32fc655	feat: replace ML optimizer with on-chain accumulation zone monitor Complete rewrite — replaces the ML-based signal optimizer with a transparent on-chain metric monitoring dashboard. Scrapes 10 metrics from LookIntoBitcoin (Playwright) and free APIs, scores each 0-10, composite 0-100. Metrics: Fear & Greed, Puell Multiple, MVRV Z-Score, Drawdown from ATH, Price vs 200W SMA, Reserve Risk, RHODL Ratio, NUPL, LTH Realized Price, Hash Ribbons. Auto-refreshes every 15 minutes. Settings page preserved. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-03-20 22:31:29 +00:00
BizzleBot	aba30f7718	fix: LLM analysis + new run button + settings page support - Fixed LLM failing silently (401 auth error on every iteration) - Reset provider to Ollama (working) from broken OpenRouter config - Added /api/clear endpoint + 'New Run' button to reset history - LLM failures now logged visibly with error details - LLM suggestions persisted to iteration data (survive restarts) - Settings page support via llm_settings.json (multi-provider)	2026-03-20 21:51:05 +00:00
BizzleBot	c17b3b5167	v3: accumulation signal optimizer - lower initial thresholds, disable PCA, simpler model start	2026-03-19 23:55:51 +00:00
BizzleBot	560863fa0d	pivot: rewrite as BTC accumulation signal optimizer Replace day-trading bot with long-term accumulation signal model. Predicts optimal BUY times using forward return analysis at 7d/30d/90d horizons, scoring each candle 0-100. Primary metric is now cost_basis_improvement_pct (model buy price vs DCA). - train_and_backtest.py: regression models (XGBoost/LSTM hybrid), accumulation-focused features (price position, momentum, volatility, volume, cycle), forward return targets, signal quality backtesting - orchestrator.py: cost improvement scoring, signal count validation - analyzer.py: accumulation-focused LLM system prompt - dashboard: cost improvement display, signal metrics table - config: new accumulation-focused parameters Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-03-19 23:51:43 +00:00
BizzleBot	a21e635d9f	feat: add LSTM, hybrid ensemble, PCA, scaler, ATR stops, rolling window Major upgrade to the ML engine: - LSTM model type: 2-layer PyTorch LSTM with early stopping, GPU support - Hybrid mode: LSTM (60%) + XGBoost (40%) with agreement gating - StandardScaler normalization (critical for LSTM) - PCA dimensionality reduction (configurable variance retention) - ATR-based dynamic stop-loss/take-profit adapting to volatility - Rolling window retraining for more realistic time series validation - Updated LLM system prompt with docs for all new parameters - All backward compatible (xgboost/lightgbm/catboost still work) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-03-19 23:02:11 +00:00
BizzleBot	e24b6605d7	fix: disable qwen3.5 thinking mode for analyzer (was consuming all tokens), increase timeout	2026-03-19 22:32:40 +00:00
BizzleBot	d81d1dedac	fix: replace unicode chars that break Windows cp1252 encoding	2026-03-19 22:25:40 +00:00