BTC Strict Candidate Review

Only strict, low-frequency BTC/ETH stat-arb candidates are shown here. The previous optimistic minute-level BTC results have been removed from the review.

Version 2026-05-10 07:45 UTC. If this page appears blank, open with ?v=202605100745 or hard refresh.

Main window: 2024-05-08 09:00 -> 2026-05-08 08:00 BTC strict audit: 12h / 1d / 2d BTC fees tested: 5bp / 10bp / 20bp BTC candidate grid: 11,664 strict variants

Strict-Candidate Summary

Best Candidate Sharpe
1.87
2-Day 21d Slow, 5bp fee
Best Low-Turnover
22.5
Trades/year for Daily 5d Low-Turnover
Best Return
+125%
Daily 7d Aggressive, 5bp fee
Grid Size
11,664
Strict variants screened

BTC Audit Result

The earlier high-Sharpe BTC V1-2 results are removed from this page because they used sample-in statistics and optimistic execution accounting. This page keeps only strict candidates with shifted signal statistics, two-leg BTC/ETH PnL, and gross-leg fees.

Strict Model

Candidate selection prioritizes daily or slower trading, fee robustness, drawdown, rolling Sharpe, rolling return, and annual trade count. The final choice should be one of the candidates below, then paper trading can be launched.

BTC Launch Candidate Shortlist

BTC strict launch candidate shortlist curves

Rolling 90D Return

BTC candidate rolling 90-day returns

Rolling 180D Sharpe

BTC candidate rolling 180-day Sharpe ratios

Trade Frequency

BTC candidate trades per year and days per trade

BTC Variant Quality Map

BTC strict variants Sharpe versus drawdown scatter

BTC Candidate Metrics

Candidate Core Params Return / CAGR Risk Trading Load Read
Daily 7d Aggressive 1d, w=7d, z=1.2/0.2, hold>=1d, fee=5bp, lev=1.0 +125.43% / 50.10% Sharpe 1.70, Sortino 2.39, MaxDD -11.03%, Calmar 4.54 115 trades, 57.5/year, 6.4d/trade, exposure 77.9%, avg hold 9.8d Best overall strict candidate; still needs walk-forward because turnover is moderate.
Daily 7d Fee-Stress 1d, w=7d, z=1.2/0.2, hold>=1d, fee=10bp, lev=1.0 +109.10% / 44.57% Sharpe 1.55, Sortino 2.20, MaxDD -11.93%, Calmar 3.74 115 trades, 57.5/year, 6.4d/trade, exposure 77.9%, avg hold 9.8d Same signal under higher cost; still strong enough to keep.
Daily 5d Low-Turnover 1d, w=5d, z=1.0/0.1, persistence=3, hold>=3d, fee=5bp, lev=1.0 +89.10% / 37.48% Sharpe 1.42, Sortino 1.84, MaxDD -11.99%, Calmar 3.13 45 trades, 22.5/year, 16.2d/trade, exposure 59.2%, avg hold 18.8d Best operational candidate if we want lower trading frequency.
Daily 5d 20bp Stress 1d, w=5d, z=1.0/0.1, persistence=3, hold>=3d, fee=20bp, lev=1.0 +72.75% / 31.41% Sharpe 1.23, Sortino 1.62, MaxDD -13.02%, Calmar 2.41 45 trades, 22.5/year, 16.2d/trade, exposure 59.2%, avg hold 18.8d Useful stress case; survives expensive execution better than high-turnover variants.
2-Day 21d Slow 2d, w=21d, z=2.0/0.5, hold>=3d, fee=5bp, lev=1.0 +63.31% / 27.73% Sharpe 1.87, Sortino 2.05, MaxDD -11.65%, Calmar 2.38 50 trades, 24.9/year, 14.6d/trade, exposure 43.6%, avg hold 12.8d Lowest-frequency promising variant; lower return but attractive Sharpe and exposure.

BTC Best By Resolution

BTC strict launch candidates best by resolution

Launch Readiness

If selecting for lower operational risk, choose Daily 5d Low-Turnover or 2-Day 21d Slow first. If selecting for higher expected return, choose Daily 7d Aggressive and size it smaller.

Before Paper

The remaining launch blocker is not the old data-leakage issue; it is final implementation parity: exact exchange symbols, order sizing, hedge leg accounting, and slippage/fee settings must match this strict model.