Crypto Trading Blueprint

Comprehensive analysis of our algorithmic crypto trading systems for external review.

Generated: 2026-06-05 13:30 EST | Data snapshot: 2026-06-05T18:25:28 UTC | 106 systems | 2 active crypto picks

Purpose: Provide enough context for another AI or human trader to evaluate our system, identify weaknesses, and suggest improvements.

Table of Contents
  1. Glossary & Architecture — Key terms, system design, fee assumptions, lifecycle rules
  2. Executive Summary — Overall crypto performance (honest assessment)
  3. System Breakdown — Per-system stats, grades, descriptions
  4. Strategy Detail — Top 30 strategies with BT/FWD metrics
  5. Strategy Logic Reference — What each strategy actually does (entry/exit rules)
  6. Backtest vs Forward — Decay analysis (overfitting detection)
  7. Daily Volume — Signal generation rate, last 7 days
  8. Optimal Picks Simulation — What if we only traded top-scored picks?
  9. Strategy Leaderboard — Top 30 by forward PnL
  10. Questions for Reviewers — Specific feedback requested
  11. Mercury AI Recommendations — External review action items
  12. Forex & Futures — Active forex/futures picks with market hours and strategy details

0. Glossary, Architecture & Assumptions

Key Terms

TermDefinition
WR (Win Rate)% of closed trades that were profitable (hit TP or closed in profit)
PF (Profit Factor)Gross wins / Gross losses. PF > 1.0 = profitable. PF > 1.5 = strong edge.
R:R (Risk:Reward)Distance to take-profit / distance to stop-loss. R:R of 2.0 means TP is 2x further than SL.
Expectancy(WR × Avg Win) − ((1−WR) × Avg Loss). Expected PnL per trade. Must be positive for profitability.
ConfidenceStrategy-specific score 0.0−1.0 generated by each system based on its internal signal strength assessment. NOT calibrated (0.8 confidence ≠ 80% probability).
BT (Backtest)Performance on historical data. Subject to look-ahead bias, curve-fitting, and survivorship bias.
FWD (Forward Test)Performance on live market data after the strategy was deployed. The gold standard for validation.
DecayBT WR − FWD WR. Positive = strategy performs worse live (likely overfitted). Negative = improved or BT data missing.
HealthStrategy status: healthy = performing within expected parameters. degrading = recent drawdown or declining WR. watch = under observation. None = status not yet classified.
AgreementNumber of independent system groups flagging the same symbol+direction. Higher = more consensus. Capped at system group level (deduplicated).
Trust TierPROVEN (1.0x weight) → SANDBOX (0.25x, default for new systems) → PROBATION (0.10−0.15x) → DEMOTED (0.25x, asset-specific failures)

System Architecture

75 CRYPTO SYSTEMS generate signals every 15-30 min via GitHub Actions │ ├── PROVEN: battleground, alpha_engine (forward-validated, full trust) ├── GENETIC: genome_gp, DARWIN variants (GP expression trees evolving buy/sell formulas) │ └── Function set: RSI, MACD, EMA, Volume, ATR, Bollinger — standard TA indicators │ └── Fitness: Sharpe ratio on 6-month rolling window, penalized by trade count │ └── Population: 500 individuals, 100 generations, tournament selection ├── ML: mercury2 (XGBoost), ml_bg_system_a-f (various ML ensembles), crypto_ml_edge ├── REVIVAL: Auto-mutates dormant strategies when stale >2 days (parameter ±10-30%) │ └── ⚠ Known risk: creates untested variants. Safeguard: enters as SANDBOX tier (0.25x weight) ├── KIMI: 81-algorithm scanner (fear/greed, order book, funding rate, whale detection) └── AGGREGATORS: super_signals, aggregated_picks (consensus from multiple systems) │ ▼ SCORING ENGINE: ranks each pick 0-100 based on weighted formula │ ▼ AUDIT DASHBOARD: tracks all picks, monitors TP/SL resolution, computes performance

Fee & Execution Assumptions

ParameterValueNotes
Exchange fees0.10% taker per sideBinance standard. Round-trip cost = 0.20%
Slippage~0.05% estimatedFor liquid pairs (BTC, ETH). Small-cap may be higher.
Funding rateNot includedPerpetual swap funding can be +/- 0.01-0.10% per 8h. Material for holds >24h.
PnL calculationEntry vs TP/SL hitDoes NOT include fees. Actual returns ~0.20-0.40% worse than shown per trade.
Price sourceBinance spot APILive prices fetched on page load. During outages, falls back to CoinGecko.
Avg hold time1h (0.0 days)Computed from current active picks. Historical may differ.

System Lifecycle

StageCriteriaTrust Weight
New / SANDBOXDefault for all new systems. <50 closed trades.0.25x
Under Review10-50 closed trades. Performance monitored.0.25x
PROVEN50+ closed trades, WR ≥ 50%, PF ≥ 1.0, documented forward results.0.8−1.0x
PROBATIONDocumented poor performance: broken risk mgmt, consistently losing, or system errors.0.10−0.15x
DEMOTEDStrategy works on some assets but fails on others (asset-specific demotion).0.25x
KilledAuto-kill when criteria met (see below). 10 of 54 systems are grade F.0x (disabled)

Kill Criteria (Mercury AI Recommended)

A system/strategy should be automatically KILLED (disabled) when ANY of these are met:
• FWD WR < 40% after ≥30 closed trades — consistently below random chance
• FWD PF < 0.8 after ≥30 closed trades — losing more than winning
• BT→FWD Decay > 25% — strategy is severely overfitted to historical data
• Max Drawdown > 30% on $100 position sizing — unacceptable risk
• No trades generated in 14+ days — dormant/broken system
• 3 consecutive losing months — persistent underperformance

⚠️ Current status: Kill mechanism is NOT yet implemented. 10/54 grade-F systems remain active, generating noise and diluting signal quality.

1. Executive Summary — Crypto Portfolio

Generated 2026-06-05 13:30 EST from 106 crypto systems producing 2 active picks.

Active Picks
75
Closed Picks
1,643
Win Rate
41.6%
Total PnL (Realized)
-150.31%
Profit Factor
0.95
Expectancy / Trade
-0.09%
Avg Win
+4.24%
Avg Loss
-3.18%
Honest Assessment: The overall crypto portfolio is unprofitable with a 41.6% win rate and 0.95 profit factor. Expectancy is -0.09% per trade — meaning every trade loses on average. The system generates too many low-quality signals from unproven strategies. However, a subset of proven strategies shows strong positive results (see Section 4).

Key Problem: Avg loss (3.18%) exceeds avg win (4.24%), AND win rate is below 50%. A profitable system needs EITHER a high WR with reasonable R:R, OR a low WR with high R:R. We have neither in aggregate.

2. System-by-System Breakdown (54 crypto systems with data)

Grade: A = proven profitable (WR≥55, PF≥1.5, 20+ trades), B = positive edge, C = marginal, D = insufficient data, F = unprofitable

SystemGradeActiveClosedWL WRTotal PnLPFExpect.Avg WAvg L
mega mutation
Mega Mutation Tournament. 1,000 DNA mutations - 33 crypto symbols with walk-forward validation. Top strategies: MACD+RSI		 A 0 279 9352 64.1% +311.98% 3.19 +2.15% +4.9% -2.7%
kimi riseoftheclaw
KIMI v11.0 - 81 algorithm scanner. Includes crypto acceleration, proven research-backed strategies, ML signal ranking, a		 C 13 1140 403482 45.2% +135.48% 1.07 +0.15% +5.1% -4.0%
stocks competition
Stocks Competition. Multi-asset competition scanner for equities - scans SPY, QQQ, major stocks.		 C 2 1369 108110 49.1% +98.30% 1.35 +0.45% +3.5% -2.6%
rapid fire
Rapid Fire 1h Scanner (NOW.py). High-frequency crypto scanner on 1h timeframe. WARNING: High-frequency noise, limited pr		 A 0 481 137 56.5% +65.95% 3.57 +2.87% +7.0% -3.7%
battleground
Forward-testing arena. Runs proven winners (VWAP Deviation +332%, RSI ETH +291%) against live market. Gold standard for 		A 0 122 6953 56.6% +27.52% 1.76 +0.23% +0.9% -0.7%
trusted genome
No description available		 A 0 43 125 66.7% +20.07% 3.37 +1.11% +2.4% -1.7%
macd dna mutations
MACD DNA Mutations. Specialized mutations of MACD-based strategies with evolved parameters (fast/slow/signal periods).		 C 0 14 70 100.0% +17.26% 0.00 +2.47% +2.5% -0.0%
contrarian evolver
DARWIN Contrarian Evolver. GP evolved for mean-reversion/contrarian signals. Optimized for oversold bounces and overboug		 C 0 10 50 100.0% +15.00% 0.00 +3.00% +3.0% -0.0%
copy trader clones
No description available		 A 0 117 118 57.9% +5.19% 1.59 +0.27% +1.3% -1.1%
copy trader intel
No description available		 C 0 336 10 100.0% +4.33% 0.00 +4.33% +4.3% -0.0%
crypto signal engine
Multi-model crypto signal engine. Combines XGBoost day-trade, LightGBM top-gainer, and ML ensemble models.		 C 0 1 10 100.0% +2.67% 0.00 +2.67% +2.7% -0.0%
dna winner picks
DNA Winner Picks. Top-scoring strategies from the DNA evolution pipeline, filtered by fitness > 0.7.		 C 6 2 11 50.0% +1.50% 1.75 +0.75% +3.5% -2.0%
multitf evolver
DARWIN Multi-Timeframe Evolver. Evolves strategies that combine signals across 1h, 4h, 1d timeframes. Higher timeframe c		 C 0 10 23 40.0% +1.50% 1.33 +0.30% +3.0% -1.5%
alpha engine fast
Alpha Engine with tighter TP/SL and shorter hold periods for more frequent signals. Same proven strategies, faster exits		 D 13 10 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
ensemble evolver
DARWIN Ensemble Coevolution (LEGION). Co-evolves teams of strategies that complement each other. Optimizes portfolio-lev		 F 2 0 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
failure evolver
DARWIN Failure Evolver. Learns specifically from LOSING picks. Mutates failed strategies to find what went wrong and evo		 F 0 3 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
fc crypto pro
FC Crypto Pro. Professional crypto signal feed with curated picks.		 D 0 6 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
genetic programmer
DARWIN GP Engine (GENESIS). Evolves trading strategies using genetic programming - buy/sell formulas are expression tree		 F 5 0 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
genome
DNA Permutation Engine. Core genetic algorithm system - represents strategies as DNA (genes for timeframe, indicators, T		 F 0 4 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
hyperparam dna
DARWIN Hyperparameter DNA. Evolves optimal indicator parameters (RSI period, EMA length, ATR multiplier) using genetic a		 D 0 8 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
inverse mutations
No description available		 D 4 77 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
mape evolver
DARWIN MAP-Elites (ATLAS). Quality-Diversity evolution - fills a grid of (volatility x trend) niches with the best strat		 F 6 0 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
maplestax cbc
No description available		 D 0 16 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
mercury2
Mercury2 XGBoost ML ensemble. Gradient-boosted tree model trained on 100+ technical features. Targets crypto momentum wi		 D 0 469 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
momentum evolver
DARWIN Momentum Evolver. Specialized GP focused on momentum/trend signals. Evolves buy/sell rules optimized for trending		 D 0 8 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
neat neural
DARWIN NEAT Neural Evolution. NeuroEvolution of Augmenting Topologies - evolves neural network topology AND weights simu		 D 0 8 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
pm kalshi signals
No description available		 F 6 0 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
polymarket signals
No description available		 F 1 0 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
prediction market consensus
No description available		 F 1 0 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
rocket scanner
No description available		 F 2 0 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
top gainer predictor
No description available		 D 0 20 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
tsmom strategy
No description available		 F 5 0 00 0.0% +0.00% 0.00 +0.00% +0.0% -0.0%
breakout b ml
Breakout Arena B - ML Breakout Detection. Uses machine learning to score breakout probability at S/R levels. Combines fr		 D 0 44 00 0.0% -0.44% 0.00 -0.01% +0.0% -0.0%
ai challenge scanner
4 AI Challenge: Scanner. Alpha Engine ML-ranked signals entered as tournament competitor. R4 avg +2.40%. Uses bollinger_		 D 0 7 01 0.0% -2.00% 0.00 -2.00% +0.0% -2.0%
ai challenge predictable
4 AI Challenge: Predictable. Highest ML Score ranking - picks the top 3 scanner BUY signals by raw ML confidence. R4 bes		 D 0 7 01 0.0% -2.22% 0.00 -2.22% +0.0% -2.2%
crypto ml edge
Crypto ML Edge. Specialized ML model for identifying statistical edges in crypto price action using feature engineering.		 C 0 8 43 50.0% -2.63% 0.76 -0.33% +2.0% -3.6%
prop firm strategies
Prop Firm Strategies. Specialized for FTMO/MFF challenge rules - daily loss limits, profit targets, consistency.		 D 0 20 37 30.0% -2.72% 0.79 -0.27% +3.4% -1.8%
proven strategies
No description available		 D 0 9 13 25.0% -3.00% 0.50 -0.75% +3.0% -2.0%
ml bg system c
ML Battleground System C - Deep Learning. LSTM + attention-based neural network for sequence prediction. WARNING: 0% WR 		D 0 5 05 0.0% -4.04% 0.00 -0.81% +0.0% -0.8%
copy trader highscore
No description available		 D 0 43 28 20.0% -4.36% 0.46 -0.44% +1.8% -1.0%
paper trading
Paper trading portfolio manager. Simulates real trades with virtual capital, tracks equity curves and drawdowns.		 C 0 34 815 34.8% -4.89% 0.92 -0.21% +7.0% -4.0%
signal aggregator
Signal Aggregator master tracker. Central registry tracking all picks across all systems with consensus scoring.		 D 0 8 03 0.0% -6.00% 0.00 -2.00% +0.0% -2.0%
breakout c spike
Breakout Arena C - Spike Reverse. Detects price spikes using cosine similarity to historical spike archetypes, then trad		 D 0 9 36 33.3% -12.35% 0.16 -1.37% +0.8% -2.5%
alpha engine
Proven Underdog Portfolio. 100 statistically-validated strategies (Connors RSI-2 75% WR, VIX Spike 72% WR). Scans crypto		 C 14 330 8995 48.4% -18.75% 0.93 -0.10% +2.9% -2.9%
ml bg ensemble
ML Battleground Ensemble - combines predictions from Systems A-F using weighted voting. Only fires when multiple ML syst		 D 0 7 07 0.0% -29.64% 0.00 -4.23% +0.0% -4.2%
kimi signal tracking
KIMI historical signal audit log. Tracks TP/SL validation against real Binance prices via signal_tracker.py.		 D 0 1350 1438 23.7% -34.11% 0.55 -0.58% +3.0% -2.0%
ml bg system a
ML Battleground System A - Filter-based ML. Applies layered filters (volume, volatility, trend) before ML prediction. Mo		 D 0 19 217 10.5% -49.84% 0.14 -2.62% +4.2% -3.4%
aggregated picks
Cross-system aggregated picks. Merges picks from all systems with symbol normalization and deduplication.		 D 2 30 719 26.9% -52.44% 0.10 -2.02% +0.8% -3.1%
ml bg system b
ML Battleground System B - Regime Detection. Uses Hidden Markov Model to detect market regimes (trending/ranging/volatil		 D 0 19 117 5.3% -54.70% 0.02 -2.88% +1.4% -3.3%
signal validation
No description available		 D 0 458 068 0.0% -136.00% 0.00 -1.43% +0.0% -2.0%
mercury2 fast
Mercury2 fast variant with shorter timeframes. WARNING: Has shown broken entry prices ($1M+) - on probation.		 D 0 27 34 42.9% -139.53% 0.07 -19.93% +3.5% -37.5%
luxalgo filters
LuxAlgo 5-Filter Confluence Engine. Automated signal generator inspired by LuxAlgo TradingView indicators. Runs 5 Python		 D 12 270 83151 35.5% -144.99% 0.68 -0.62% +3.7% -3.0%
super signals
Cross-system consensus super signals. Only fires when multiple independent systems agree on direction + symbol. High-con		 D 14 171 3890 29.7% -157.91% 0.52 -1.23% +4.4% -3.6%
ml bg system f
ML Battleground System F - Claws of Doom. Proprietary multi-factor ML combining KIMI signals with battleground validatio		 D 0 214 84130 39.3% -173.81% 0.76 -0.81% +6.6% -5.6%

3. Strategy Breakdown — Top 30 by Active Picks

Each strategy may run across multiple systems. Forward WR/PF/Trades reflect live forward-testing performance. BT = backtest.

StrategyActiveDirectionsSymbols BT WRFWD WRFWD TradesFWD PF ConfR:RAvg PnLHealth
tsmom_volscaled 1 SHORT:1 SUI n/a 0.0% 0 0.00 0.76 2.0 +2.35% None:1
kalshi_mtf_consensus 1 SHORT:1 ETH n/a 0.0% 0 0.00 0.79 0.0 +0.00% None:1

3b. Strategy Logic Reference — What Each Strategy Actually Does

For an external reviewer to evaluate signal quality, they need to know the actual entry/exit rules, not just names. Strategies marked with "No documented rules" are a transparency gap.

StrategyActiveFWD WR/TradesEntry/Exit Logic & Parameters
tsmom_volscaled10%/0tNo documented entry/exit rules. Strategy logic needs documentation.
kalshi_mtf_consensus10%/0tNo documented entry/exit rules. Strategy logic needs documentation.
Transparency gap: Many strategies lack documented entry/exit rules. This makes external review impossible for those strategies. A reviewer can only evaluate strategies where the logic is explicitly documented. Undocumented strategies should be treated as untestable black boxes.

4. Backtest vs Forward-Test Performance

Decay = BT WR − FWD WR. Positive decay means strategy performs worse live than in backtest (overfitting). Negative means it improved.

What to look for: Strategies with low decay AND sufficient forward trades (≥10) are the most trustworthy. High decay (>15%) suggests curve-fitting. Negative decay with few trades may be luck.
StrategySystem BT WRBT TradesFWD WRFWD TradesDecay
connors_rsi2 alpha_engine 68.5%851 0.0%0 +0.0%
vwap_mean_reversion alpha_engine 64.5%723 0.0%0 +0.0%
bollinger_mean_reversion alpha_engine 60.4%336 0.0%0 +0.0%
rsi_extreme_reversal alpha_engine 60.6%109 0.0%0 +0.0%
ema_crossover_trend alpha_engine 44.5%317 0.0%0 +0.0%
macd_divergence alpha_engine 68.2%488 0.0%0 +0.0%
supertrend alpha_engine 54.3%35 0.0%0 +0.0%
connors_r3 alpha_engine 71.9%751 0.0%0 +0.0%
double_seven alpha_engine 63.0%818 0.0%0 +0.0%
three_day_low alpha_engine 64.8%1050 0.0%0 +0.0%
williams_r_oversold alpha_engine 59.3%450 0.0%0 +0.0%
keltner_mean_reversion alpha_engine 68.3%101 0.0%0 +0.0%
volatility_scaled alpha_engine 66.2%536 0.0%0 +0.0%
dual_momentum alpha_engine 49.5%660 0.0%0 +0.0%
kama_adaptive_trend alpha_engine 26.1%2662 0.0%0 +0.0%
kama_adx_trend alpha_engine 47.7%640 0.0%0 +0.0%
volatility_anchor_fade alpha_engine 48.0%494 0.0%0 +0.0%
kama_pullback_continuation alpha_engine 33.1%426 0.0%0 +0.0%
williams_pr_trend_mr alpha_engine 49.9%706 0.0%0 +0.0%
cci_exhaustion_reversal alpha_engine 54.0%163 0.0%0 +0.0%
donchian_adx_breakout alpha_engine 41.2%1229 0.0%0 +0.0%
kama_slope_reversal alpha_engine 60.5%43 0.0%0 +0.0%
volume_dryup_fade alpha_engine 0.0%3 0.0%0 +0.0%
short_term_return_reversal alpha_engine 44.5%1006 0.0%0 +0.0%
vwap_volume_mean_reversion alpha_engine 43.8%1394 0.0%0 +0.0%
donchian_turtle_breakout alpha_engine 46.1%1378 0.0%0 +0.0%
triple_ma_trend alpha_engine 48.5%951 0.0%0 +0.0%
kama_mean_reversion alpha_engine 71.2%358 0.0%0 +0.0%
autocorr_reversion alpha_engine 63.9%1060 0.0%0 +0.0%
coinglass_extreme_reversion coinglass 0.0%0 0.0%0 +0.0%
coinglass_funding_confluence coinglass 0.0%0 0.0%0 +0.0%
coinglass_leverage_squeeze coinglass 0.0%0 0.0%0 +0.0%
coinglass_ratio_momentum coinglass 0.0%0 0.0%0 +0.0%
coinglass_roll_yield coinglass 0.0%0 0.0%0 +0.0%
coinglass_sentiment_composite coinglass 0.0%0 0.0%0 +0.0%
coinglass_whale_divergence coinglass 0.0%0 0.0%0 +0.0%

5. Daily Pick Volume — Last 7 Days

How many crypto signals are being generated each day? Is the system too aggressive or too conservative?

DatePicksDirectionsTop SymbolsTop Systems
2026-06-052 SHORT:2 SUI(1), ETH(1) tsmom_strategy(1), pm_kalshi_signals(1)
Volume Analysis: The system generates 2 crypto picks over 7 days (avg 2/day). Volume is manageable but may miss opportunities if too conservative.

6. Optimal Pick Selection — Simulated Results

If we only traded the top-scoring picks, what would performance look like? All PnL values are unrealized (positions still open).

Scoring Formula

score = (0.25 × confidence + 0.20 × min(R:R/5, 1) + 0.25 × fwd_wr + 0.15 × min(fwd_pf/3, 1) + 0.10 × min(fwd_trades/50, 1) + 0.05 × min(agreement/3, 1)) × health_multiplier health_multiplier: healthy=1.0, degrading=0.7, other=0.5

A) Top 10 by Score

2 picks WR: 100.0% Total PnL: +2.35% (unrealized) PF: 99 Avg PnL: +1.18% Worst pick: +0.00% 2 symbols {'SHORT': 2} Hold: avg 1h, max 1h Overlap: Top 20 Score: 2 shared; Top 20 High Agreement (≥3 systems): 1 shared; Mercury Top 20 (FWD-weighted): 2 shared
SymbolDirEntryTPSLPnL (unreal.) Age(h)ConfR:RFWD WRAgreeStrategySystem
SUI SHORT 0.71370.5810570.780021 +2.35% 1h 0.762.0 0% 0 tsmom_volscaled tsmom_strategy
ETH SHORT 1576.831539.67988521599.07557772 +0.00% 0h 0.790.0 0% 7 kalshi_mtf_consensus pm_kalshi_signals

B) Top 20 by Score

2 picks WR: 100.0% Total PnL: +2.35% (unrealized) PF: 99 Avg PnL: +1.18% Worst pick: +0.00% 2 symbols {'SHORT': 2} Hold: avg 1h, max 1h Overlap: Top 10 Score: 2 shared; Top 20 High Agreement (≥3 systems): 1 shared; Mercury Top 20 (FWD-weighted): 2 shared
SymbolDirEntryTPSLPnL (unreal.) Age(h)ConfR:RFWD WRAgreeStrategySystem
SUI SHORT 0.71370.5810570.780021 +2.35% 1h 0.762.0 0% 0 tsmom_volscaled tsmom_strategy
ETH SHORT 1576.831539.67988521599.07557772 +0.00% 0h 0.790.0 0% 7 kalshi_mtf_consensus pm_kalshi_signals

C) Top 20 Vetted Only (Forward WR ≥ 55%, Forward Trades ≥ 10)

No picks matched criteria.

D) Top 20 High Agreement (≥3 systems agree)

1 picks WR: 0% Total PnL: +0.00% (unrealized) PF: 0 Avg PnL: +0.00% Worst pick: +0.00% 1 symbols {'SHORT': 1} Hold: avg 0h, max 0h Overlap: Top 10 Score: 1 shared; Top 20 Score: 1 shared; Mercury Top 20 (FWD-weighted): 1 shared
SymbolDirEntryTPSLPnL (unreal.) Age(h)ConfR:RFWD WRAgreeStrategySystem
ETH SHORT 1576.831539.67988521599.07557772 +0.00% 0h 0.790.0 0% 7 kalshi_mtf_consensus pm_kalshi_signals

E) Mercury AI Recommended Scoring (FWD-weighted)

mercury_score = 0.30 × fwd_wr + 0.25 × min(fwd_pf/3, 1) + 0.20 × confidence + 0.15 × min(agreement/3, 1) + 0.10 × trust_tier_score trust_tier_score: PROVEN=1.0, SANDBOX=0.5, DEMOTED=0.3, PROBATION=0.2 Key change: FWD WR weight increased from 0.25→0.30, R:R removed, trust tier added
2 picks WR: 100.0% Total PnL: +2.35% (unrealized) PF: 99 Avg PnL: +1.18% Worst pick: +0.00% 2 symbols {'SHORT': 2} Hold: avg 1h, max 1h Overlap: Top 10 Score: 2 shared; Top 20 Score: 2 shared; Top 20 High Agreement (≥3 systems): 1 shared
SymbolDirEntryTPSLPnL (unreal.) Age(h)ConfR:RFWD WRAgreeStrategySystem
ETH SHORT 1576.831539.67988521599.07557772 +0.00% 0h 0.790.0 0% 7 kalshi_mtf_consensus pm_kalshi_signals
SUI SHORT 0.71370.5810570.780021 +2.35% 1h 0.762.0 0% 0 tsmom_volscaled tsmom_strategy

F) Mercury Strict Filter (FWD Trades≥30, WR≥55%, PF≥1.2)

Only strategies with substantial forward validation and proven profitability. This is the most conservative filter.

No picks matched criteria.

G) Grok Hard-Gated + Normalized Scoring

HARD GATE (score=0 if ANY fail): fwd_trades < 30, fwd_wr < 55%, fwd_pf < 1.2, decay > 15% normalized_wr = min(1.0, (fwd_wr - 0.45) / 0.55) normalized_pf = min(1.0, (fwd_pf - 1.0) / 2.0) score = 0.30 × norm_wr + 0.25 × norm_pf + 0.20 × confidence + 0.15 × (agreement/max) + 0.10 × trust 55% weight on proven forward metrics. Zero tolerance for unvalidated strategies.

No picks matched criteria.

Important Caveats

All PnL values above are UNREALIZED — these positions are still open and can reverse at any time. An unrealized total PnL is NOT a valid measure of strategy quality. These snapshots change every few minutes.

Portfolio overlap: The four portfolios above are NOT independent — they share many of the same picks (same high-scoring picks appear in multiple selection methods). Do not treat them as independent validation.

No benchmark: Without comparing to a simple BTC buy-and-hold or a 60/40 crypto index over the same period, it is impossible to know if these picks add alpha or are just riding market beta.

Scoring Formula Limitations

Reviewers should note:
• Confidence is uncalibrated — a 0.8 confidence score does NOT mean 80%% probability of profit. Each system generates its own scale.
• Agreement caps at 3 — the formula uses min(agreement/3, 1), so 3-system and 23-system agreement score identically. This may underweight strong consensus.
• Forward WR dominates at 25%% weight, but most strategies have 0 forward trades, making this component 0 for the majority of picks.
• No historical backtest of the formula itself — we have not validated whether this scoring formula would have selected winners in the past.
Recommendation for reviewers: Compare portfolios A-D. Which filtering criteria produces the best risk-adjusted returns? If vetted picks (C) outperform raw score (A), the system should raise minimum forward-trade thresholds. If agreement picks (D) outperform, cross-system consensus is more valuable than individual strategy metrics.

7. Strategy Leaderboard — Top 30 by Forward PnL

Strategies ranked by actual forward-testing profit. Only strategies with ≥3 forward trades shown.

StrategySystems BT WRBT TradesBT PF FWD WRFWD TradesFWD PnLDecay
macd_rsi_m048 mega_mutation 0.0%00.00 68.3% 123 +440.43% -68.3%
macd_rsi_m048 mega_mutation 0.0%00.00 68.3% 123 +440.43% -68.3%
rs-breakout-scout kimi_competition, kimi_riseoftheclaw 0.0%00.00 66.7% 48 +90.11% -66.7%
rs-breakout-scout kimi_riseoftheclaw 0.0%00.00 66.7% 48 +90.11% -66.7%
macd_rsi_m017 mega_mutation 0.0%00.00 77.8% 27 +80.26% -77.8%
macd_rsi_m017 mega_mutation 0.0%00.00 77.8% 27 +80.26% -77.8%
donchian-stock-breakout kimi_competition, kimi_riseoftheclaw 0.0%00.00 73.3% 15 +79.96% -73.3%
donchian-stock-breakout kimi_riseoftheclaw 0.0%00.00 73.3% 15 +79.96% -73.3%
stochrsi_macd_combo rapid_fire 0.0%00.00 65.2% 23 +61.95% -65.2%
stochrsi_macd_combo rapid_fire 0.0%00.00 65.2% 23 +61.95% -65.2%
Bollinger MR stocks_competition 0.0%00.00 51.2% 82 +59.76% -51.2%
Bollinger MR stocks_competition 0.0%00.00 51.2% 82 +59.76% -51.2%
futures_momentum multi_asset_copytrader 0.0%00.00 60.9% 46 +57.81% -60.9%
futures_momentum multi_asset_copytrader 0.0%00.00 60.9% 46 +57.81% -60.9%
prediction_market_consensus alpha_engine 0.0%00.00 81.2% 32 +56.67% -81.2%
prediction_market_consensus alpha_engine 0.0%00.00 81.2% 32 +56.67% -81.2%
gap-and-go-stocks kimi_competition, kimi_riseoftheclaw 0.0%00.00 60.0% 10 +49.89% -60.0%
gap-and-go-stocks kimi_riseoftheclaw 0.0%00.00 60.0% 10 +49.89% -60.0%
price-accel-scout kimi_competition, kimi_riseoftheclaw 0.0%00.00 63.6% 22 +47.13% -63.6%
price-accel-scout kimi_riseoftheclaw 0.0%00.00 63.6% 22 +47.13% -63.6%
multi_period_rsi_confluence_eth battleground, web_ai 0.0%00.00 61.0% 118 +44.29% -61.0%
meme-velocity kimi_competition, kimi_riseoftheclaw 0.0%00.00 47.8% 26 +41.29% -47.8%
meme-velocity kimi_riseoftheclaw 0.0%00.00 47.8% 26 +41.29% -47.8%
crypto_keltner_compression_expansio codex_gpt5 0.0%00.00 54.3% 173 +40.42% -54.3%
mtf-align-scout kimi_competition, kimi_riseoftheclaw 0.0%00.00 50.0% 24 +39.79% -50.0%
mtf-align-scout kimi_riseoftheclaw 0.0%00.00 50.0% 24 +39.79% -50.0%
multi_period_rsi_confluence web_ai 0.0%00.00 47.7% 149 +38.37% -47.7%
ema-ribbon-momentum-scout kimi_competition, kimi_riseoftheclaw 0.0%00.00 63.6% 22 +37.94% -63.6%
ema-ribbon-momentum-scout kimi_riseoftheclaw 0.0%00.00 63.6% 22 +37.94% -63.6%
strong consensus (alpha_engine, qua super_signals 0.0%00.00 38.5% 13 +37.59% -38.5%

8. Questions for AI / Human Reviewers

Q1: Signal Volume

We generate 100-1000+ crypto signals per day across 75 systems. Is this creating too much noise? Should we enforce a hard cap (e.g., max 20 picks/day) based on score ranking?

Q2: Entry Quality

Our avg loss (-19.88%) exceeds avg win (+17.78%). Should we tighten stop-losses, widen take-profits, or implement trailing stops to improve the win/loss ratio?

Q3: Strategy Pruning

Many strategies have <5 forward trades. Should we require a minimum forward trade count (e.g., 20) before allowing a strategy to generate live signals?

Q4: Backtest Decay

Several strategies show 30-50% decay from backtest to forward test. At what decay threshold should we auto-demote or disable a strategy?

Q5: Optimal Filtering

Based on the simulated portfolios in Section 6, which selection criteria (score, vetted, agreement) produces the best risk-adjusted outcome? What filter thresholds would you recommend?

Q6: Position Sizing

Currently all picks are equal-weighted. Should we size positions based on confidence score, R:R ratio, system trust tier, or forward validation strength?

Q7: What's Actually Working?

Looking at the leaderboard (Section 7) and system grades (Section 2), which 3-5 strategies or systems show genuine edge that we should double down on?

9. Mercury AI Recommendations (External Review)

These recommendations were provided by an external AI reviewer (Mercury) after analyzing the full blueprint. They represent actionable improvements prioritized by impact.

Top 5 Strategies to Scale

#StrategyWhy ScaleAction
1crypto_rsi_whaleconfirmed_v1FWD WR 67%, PF 2.1+ — combines RSI with on-chain whale confirmationIncrease position size, add to PROVEN tier
2funding_momentumExploits funding rate trends — unique edge not in standard TAExtend to more perpetual pairs
3crypto_keltner_compression_expansionVolatility squeeze breakout — well-understood mechanicsAdd multi-timeframe confirmation
4crypto_vwap_deviation_reversion_volMean reversion to VWAP with volume filter — institutional flow signalTighten entry window, reduce max hold time
5crypto_kalman_trend_residual_reversionKalman filter trend + residual reversion — adaptive to regime changesValidate with walk-forward optimization

Recommended Filtering Thresholds

Minimum requirements before a strategy generates live signals:
• ≥30 forward trades (current: many strategies have <5)
• FWD WR ≥ 55% (current: no minimum enforced)
• FWD PF ≥ 1.2 (current: no minimum enforced)
• BT→FWD Decay ≤ 10% (current: some strategies show 30-50% decay)

Impact: This would dramatically reduce signal noise by disabling unproven strategies while preserving strategies with genuine edge.

Position Sizing Recommendations

TierAllocationCriteria
Core (60%)3-5% per pickPROVEN systems, FWD WR≥60%, PF≥1.5, ≥50 trades
Satellite (30%)1-2% per pickSANDBOX with FWD WR≥55%, PF≥1.2, ≥30 trades
Experimental (10%)0.5% per pickNew strategies, genetic mutations, <30 trades

Scoring Formula Restructure

CURRENT: 0.25×conf + 0.20×R:R + 0.25×fwd_wr + 0.15×fwd_pf + 0.10×trades + 0.05×agreement PROPOSED: 0.30×fwd_wr + 0.25×fwd_pf + 0.20×conf + 0.15×agreement + 0.10×trust_tier Key changes: - FWD WR increased 0.25→0.30 (most predictive metric) - R:R removed (redundant with PF, can be gamed by wide targets) - Trust tier added at 0.10 (rewards battle-tested systems) - Agreement increased 0.05→0.15 (cross-system consensus is strong signal)

Priority Action Items

Immediate (this week):
1. Implement hard gates in scanner: zero score if fwd_trades<30, fwd_wr<55%, fwd_pf<1.2, decay>15%
2. Cap daily signals at 20 via top-score selection (cuts ~90% noise)
3. Implement kill criteria for grade-F systems (see Section 0)
4. Add trailing stop at 70% R:R target — tighten SL to min(ATR×0.5×R:R, entry×0.005)

Short-term (2 weeks):
5. Switch to Grok normalized scoring (compare Portfolio G vs E vs A-D)
6. Implement tiered position sizing: position_size = min(0.05, (score×R:R) / sum_scores)
7. Add BTC buy-and-hold benchmark comparison to all portfolio sims
8. Adjust all PnL metrics -0.40% round-trip to account for exchange fees

Medium-term (1 month):
9. Walk-forward optimization for top 5 strategies
10. Add realized PnL tracking (currently all PnL is unrealized)
11. Build correlation check: prune_correlated_picks to avoid concentrated bets
12. Build a proper backtest of the scoring formula itself on historical picks

Confirmed Red Flags (3/3 AI Reviewers Agree)


• ~86%% of active picks come from Grade F systems (Codex finding) — the system is feeding mostly unvalidated noise into portfolios.
• Over-reliance on unvetted ML/genetic systems — high decay rates indicate overfitting. Most should be killed or sandboxed.
• Missing fees in PnL calculation — all metrics should subtract -0.40%% round-trip (0.10%% taker × 2 sides + ~0.05%% slippage × 2).
• No correlation checks — multiple picks on the same asset/direction from different systems create concentrated risk.
• Duplicate stacking — multiple systems bet same symbol+direction, treated as independent alpha when it's concentrated risk.
• Revival_Mutated_* strategies should be blocked from live capital entirely (paper-only until validated).
• Confidence is double-counted with FWD WR (they correlate, weighting both inflates scores).
• Data integrity: closed ≠ wins + losses in some systems — phantom/unresolved trades exist.

Projected impact of implementing filters + scoring: +15-25%% expectancy boost within 30 days.

Codex (GPT-5.3) Additional Recommendations

2-Stage Scoring (replaces single formula):
Stage 1 — Hard Pass/Fail Gate:
• fwd_trades ≥ 50 (strictest threshold across all reviewers)
• fwd_pf_net ≥ 1.15 (after fees)
• net expectancy > 0
• health = healthy
• NOT Revival_Mutated_*
• Grade A/B only (C on strict probation)

Stage 2 — Rank by uncertainty-adjusted net expectancy (not raw WR):
• Use expectancy = (WR × avg_win) − ((1−WR) × avg_loss) − round_trip_cost
• Penalize low sample size with uncertainty factor: expectancy × (1 − 1/sqrt(trades))
• One position per symbol+direction (no duplicate stacking)

Best portfolio method: D (High Agreement) + deduplication + vetting gate
Kill criteria: Probation at rolling-20 PF < 0.9; Kill at rolling-50 PF < 0.95 or decay > 15pp

10. Forex & Futures — Active Picks

Non-crypto picks from forex (7 strategies) and index futures (Connors RSI-2, TOM). These extend proven equity strategies to 24/5 markets.

Forex Picks
10
Futures Picks
1
Total Active
11
Unique Symbols
7
Strategies
6
BUY Signals
6
SELL Signals
5
Market Hours:
Forex: 24/5 — Sunday 5 PM ET to Friday 5 PM ET (continuous)
Futures (ES, NQ): Sunday 6 PM ET to Friday 5 PM ET (with daily halt 5-6 PM ET)

Key Strategies:
Connors RSI-2 on ES=F/NQ=F: Extends our most proven strategy (75.7% WR on SPY, p=6e-6) to futures. ES/NQ track SPY/QQQ with near-perfect correlation but trade extended hours.
Turn-of-Month on ES=F/NQ=F: Lakonishok & Smidt (1988) TOM effect applies equally to index futures.
Forex: 7 strategies including carry trade, mean reversion, JPY risk-off, DXY correlation, BB squeeze, session momentum, and DXY RSI fade.

Active Picks (11 total)

SymbolClassSignalEntryTPSL ConfR:RPnLHoldStrategyReason
USDCAD=X FOREX SELL 1.39411.3812261.399907 0.692.2 +0.00% 0d forex_zscore_200d_fade Z-score=1.88 above 200d SMA (1.38123), RSI=76. Backtest: 68.3% WR on 167 trades
USDCHF=X FOREX SELL 0.795910.7907890.79989 0.701.3 +0.00% 0d forex_zscore_200d_fade Z-score=2.05 above 200d SMA (0.79079), RSI=61. Backtest: 68.3% WR on 167 trades
AUDUSD=X FOREX BUY 0.7057660.7199130.697278 0.601.7 +0.00% 0d cta_golden_cross_200 Golden Cross: SMA50=0.71 > SMA200=0.68. ADX=38.4 (trending). Price=0.7058. Estab
USDCHF=X FOREX LONG 0.7910.849284040.7394268 0.951.2 +0.01% 0d regime_accumulation
USDCAD=X FOREX LONG 1.38931.475725021.33347932 0.951.1 +0.00% 0d regime_mild_bull
USDCAD=X FOREX SELL 1.393861.38689071.39804158 0.692.2 +0.00% 0d forex_zscore_200d_fade Z-score=1.86 above 200d SMA (1.38122), RSI=76. Backtest: 68.3% WR on 167 trades
USDCHF=X FOREX SELL 0.796080.790790.80006 0.711.3 -0.00% 0d forex_zscore_200d_fade Z-score=2.08 above 200d SMA (0.79079), RSI=61. Backtest: 68.3% WR on 167 trades
EURUSD=X FOREX BUY 1.1527381.1592751.14729 0.701.2 +0.00% 0d combined_confidence Combined Confidence: CS=0.70 (HIGH), TE=0.60 (WR=80.0%, 0 trades), MP=0.80; size
USDJPY=X FOREX SELL 160.298004159.739059160.7288579 0.681.3 +0.00% 0d combined_confidence Combined Confidence: CS=0.68 (MEDIUM), TE=0.57 (WR=78.6%, 0 trades), MP=0.79; si
CADJPY=X FOREX BUY 115.004997115.487121114.63335975 0.621.3 +0.00% 0d combined_confidence Combined Confidence: CS=0.62 (MEDIUM), TE=0.50 (WR=75.0%, 0 trades), MP=0.75; si
YM=F FUTURES BUY 0.7081650.711705830.70604051 0.751.6 -0.00% 0d futures_cross_asset_momentum Combined Confidence: CS=0.60 (MEDIUM), TE=0.47 (WR=73.6%, 0 trades), MP=0.74; si

Trading Blueprint v2.0 — Auto-generated from live audit data

Data freshness: updates every 15 minutes via GitHub Actions

Source: Audit Dashboard