Skip to content

Latest commit

 

History

History
235 lines (214 loc) · 14.4 KB

File metadata and controls

235 lines (214 loc) · 14.4 KB

📊 AutoResearch — Strategy Report

Generated: 2026-03-22T14:10:11.750Z Engine: @darksol/autoresearch | Pairs: 4 | Model: claude-sonnet-4.5

Summary

Metric Value
Total experiments 228
Kept / Reverted 59 ✅ / 169 ❌
Hit rate 25.9%
Best score 8.176 (exp199)
Best Sharpe 8.176
Total return 6.64%
Max drawdown 2.18%
Trades 134
Win rate 53.7%
Score target 20

Strategy Evolution

Each row is a kept experiment — a change that improved the strategy.

# Score Δ Hypothesis
1 0.421
4 0.486 +0.065
11 0.523 +0.037
12 0.531 +0.008
17 0.540 +0.009
29 0.564 +0.024
30 0.610 +0.046
baseline 0.610 0.000 Initial strategy — baseline measurement
baseline 0.610 0.000 Initial strategy — baseline measurement
baseline 0.610 0.000 Initial strategy — baseline measurement
exp037 0.615 +0.005 I will decrease the atrPeriod from 14 to 7 to make the position sizing more reactive to recent vol
baseline 0.615 0.000 Initial strategy — baseline measurement
baseline 0.615 0.000 Initial strategy — baseline measurement
exp053 0.671 +0.056 I will increase the exitThreshold from 0.01 to 0.015 to hold winning positions longer before takin
exp065 0.714 +0.043 I will reduce the deviationThreshold from 0.025 to 0.022 to capture earlier mean-reversion entries
exp070 0.726 +0.012 Reduce the rsiPeriod from 14 to 10 to make RSI more responsive to recent price action, allowing ea
exp074 0.740 +0.014 Reduce the atrPeriod from 7 to 5 to make volatility-based position sizing more responsive to recen
baseline 0.740 0.000 Initial strategy — baseline measurement
baseline 0.157 -0.583 Initial strategy — baseline measurement
baseline -1.460 -1.617 Initial strategy — baseline measurement
baseline -1.460 Initial strategy — baseline measurement
baseline -1.460 Initial strategy — baseline measurement
baseline -1.460 Initial strategy — baseline measurement
exp117 2.838 Complete strategy redesign: switched from VWAP reversion (overfit to synthetic data) to adaptive tre
baseline 2.838 0.000 Initial strategy — baseline measurement
baseline 2.838 0.000 Initial strategy — baseline measurement
baseline 2.838 0.000 Initial strategy — baseline measurement
baseline 2.838 0.000 Initial strategy — baseline measurement
exp126 2.919 +0.081 Add regime-based position sizing by increasing exposure in trending regimes (high Hurst exponent) an
exp127 2.923 +0.004 Reduce Hurst lookback from 50 to 30 bars to make regime detection more responsive to recent market c
exp128 3.668 +0.745 Reduce ATR trail multiple from 2.0 to 1.5 to exit winners earlier and capture profits more aggressiv
baseline 3.668 0.000 Initial strategy — baseline measurement
exp137 3.741 +0.073 Replace the Hurst exponent regime detection with a simpler trend-following approach using price mome
exp151 3.777 +0.036 Add a multi-timeframe trend filter that calculates a longer-term 50-period EMA slope and only takes
baseline 3.777 0.000 Initial strategy — baseline measurement
exp161 4.512 +0.735 STRUCTURAL: Ensemble voting (Donchian+RSI+MACD) + macro 100-EMA trend filter + conviction-weighted s
baseline 4.512 0.000 Initial strategy — baseline measurement
exp163 5.310 +0.798 Replace the ensemble voting system with a pure trend-following breakout strategy that only enters wh
baseline 5.310 0.000 Initial strategy — baseline measurement
exp170 7.327 +2.017 Replace the fixed volatility expansion filter (ATR > ATR_SMA) with a percentile-based expansion trig
baseline 7.327 0.000 Initial strategy — baseline measurement
exp180 7.875 +0.548 Replace the fixed 3.0x ATR profit target with an adaptive profit target that scales with trend stren
baseline 7.875 0.000 Initial strategy — baseline measurement
exp183 7.991 +0.116 Replace the fixed 20-period Donchian breakout with a dynamic breakout lookback that adapts to volati
baseline 7.991 0.000 Initial strategy — baseline measurement
baseline 7.991 0.000 Initial strategy — baseline measurement
exp199 8.176 +0.185 Implement a dual-strategy portfolio allocation system that runs a pure breakout strategy and a count
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement
baseline 8.176 0.000 Initial strategy — baseline measurement

Score Progression

1        ███ 0.421
4        ███ 0.486
11       ███ 0.523
12       ███ 0.531
17       ███ 0.540
29       ███ 0.564
30       ████ 0.610
baseline ████ 0.610
baseline ████ 0.610
baseline ████ 0.610
exp037   ████ 0.615
baseline ████ 0.615
baseline ████ 0.615
exp053   ████ 0.671
exp065   ████ 0.714
exp070   ████ 0.726
exp074   █████ 0.740
baseline █████ 0.740
baseline █ 0.157
baseline █ -1.460
baseline █ -1.460
baseline █ -1.460
baseline █ -1.460
exp117   █████████████████ 2.838
baseline █████████████████ 2.838
baseline █████████████████ 2.838
baseline █████████████████ 2.838
baseline █████████████████ 2.838
exp126   ██████████████████ 2.919
exp127   ██████████████████ 2.923
exp128   ██████████████████████ 3.668
baseline ██████████████████████ 3.668
exp137   ███████████████████████ 3.741
exp151   ███████████████████████ 3.777
baseline ███████████████████████ 3.777
exp161   ████████████████████████████ 4.512
baseline ████████████████████████████ 4.512
exp163   ████████████████████████████████ 5.310
baseline ████████████████████████████████ 5.310
exp170   █████████████████████████████████████████████ 7.327
baseline █████████████████████████████████████████████ 7.327
exp180   ████████████████████████████████████████████████ 7.875
baseline ████████████████████████████████████████████████ 7.875
exp183   █████████████████████████████████████████████████ 7.991
baseline █████████████████████████████████████████████████ 7.991
baseline █████████████████████████████████████████████████ 7.991
exp199   ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176
baseline ██████████████████████████████████████████████████ 8.176

Pair Analysis

Per-pair breakdown not available — run a backtest with --verbose to generate pair-level stats.

Current Strategy

Parameter Value
hurstLookback 50
trendEmaPeriod 50
breakoutLookbackLow 15
breakoutLookbackHigh 25
volRegimeThreshold 70
atrPeriod 10
atrPercentileLookback 50
atrPercentileThreshold 60
rocPeriod 10
atrTrailMultiple 1.5
atrProfitMultipleMin 2.0
atrProfitMultipleMax 4.0
trendStrengthThreshold 0.10
bbPeriod 20
bbStdDev 2.0
rsiPeriod 14
rsiOversold 30
rsiOverbought 70
mrProfitTarget 0.015
mrStopLoss 0.01
basePositionSize 0.15
maxPositions 3
cooldown 3

Active Trading Pairs

Pair DEX Fee Source
ETH/USDC uniswap 0.05% 🔒 built-in
ETH/USDC-30 uniswap 0.30% 🔒 built-in
cbETH/WETH uniswap 0.05% 🔒 built-in
AERO/USDC aerodrome 🔒 built-in

Top 10 Experiments

Rank ID Score Sharpe Status Hypothesis
1 exp199 8.176 8.176 Implement a dual-strategy portfolio allocation system that runs a pure breakout
2 exp201 8.176 8.176 Manual mutation — implement via OpenClaw agent
3 baseline 8.176 8.176 Initial strategy — baseline measurement
4 baseline 8.176 8.176 Initial strategy — baseline measurement
5 baseline 8.176 8.176 Initial strategy — baseline measurement
6 baseline 8.176 8.176 Initial strategy — baseline measurement
7 baseline 8.176 8.176 Initial strategy — baseline measurement
8 baseline 8.176 8.176 Initial strategy — baseline measurement
9 baseline 8.176 8.176 Initial strategy — baseline measurement
10 baseline 8.176 8.176 Initial strategy — baseline measurement

Failure Patterns

Last 10 rejected experiments — avoid repeating these:

  • exp210 (7.843): Implement a volatility-regime-adaptive position scaling system that dynamically adjusts position sizes based on real-tim
  • exp213 (7.137): Implement a cross-pair momentum divergence system that detects when a pair shows strong breakout signals while the baske
  • exp214 (1.090): Implement a multi-pair relative strength ranking system that only enters breakout trades on pairs showing top-quartile m
  • exp215 (-999.000): Implement a volatility-adaptive stop-loss layering system that uses multiple ATR-based stops at different distances (tig
  • exp216 (5.575): Implement a volatility-adaptive multi-timeframe momentum filter that synthesizes 3 different lookback periods (short=5,
  • exp219 (4.457): Implement a cross-pair volume-weighted momentum scoring system that ranks pairs by their volume-adjusted ROC strength an
  • exp220 (-999.000): Implement a position-sizing pyramid system that adds to winning breakout trades in thirds (entry at breakout, +33% at +1
  • exp221 (-999.000): Implement a volatility-adjusted partial profit-taking system that scales out 50% of breakout positions at +2 ATR profit
  • exp223 (8.150): Implement a cross-pair correlation-based signal generator that detects divergence between ETH/USDC and cbETH/WETH price
  • exp226 (7.758): Implement an adaptive volatility regime filter that uses dual ATR measurements (fast 5-period and slow 20-period) to det

Configuration

Setting Value
Mutation model claude-sonnet-4.5
Max experiments 200
Score target 20
Backtest capital $100,000
Trade floor 50 trades
Max drawdown limit 50%
Bankr LLM ✅ Enabled
Live mode 📝 Paper

Generated by node scripts/report.js — @darksol/autoresearch 🌑