E-16. Strategy 7 — BTC 5m statistical shock reversal (2026-07-15)
- Preregistered: D-019. 16 combinations, activity condition of 0.5~3 per day, train → top 3 to val.
- Data: train 253,525 bars / 880 days. Test and val did not need to be opened given the verdict.
- Activity result: 1,781~1,985 trades, 2.02~2.25 per day. The desired frequency was met exactly.
- Economics: even the best combination had Sharpe -12.34 / PF 0.27 / -80.95% / MDD -81.01%. All 16 failed Stage 1, so 0 went to val comparison.
- Cost breakdown (best combination tail 1%, volume 2x, distance 2 ATR, hold 6): baseline -80.95%, fees only -58.62%, fees + 1bp slippage -66.77%, no costs -6.75% (Sharpe -0.49, PF 0.95). The signal itself is negative even before costs, and high-turnover costs magnify it.
- Verdict: rejected (D-020). "Several times a day" can be produced, but OHLCV shock fading has no edge.
Source and migration evidence
Source: backtest/docs/experiment-log.md#E-16
Original SHA-256: 069c4aaaf53a20930ca8d94ba6365b4ce2b6bc560171d662a4c179ea55a6857c
Raw data and account information are not part of the public package. The original results do not replace statistical or scientific revalidation. The English text is a translation; numbers, tables and charts are carried over unchanged.