Engine backtest certificate

Cross-validated robustness metrics for the live signal-driven strategy. DSR > 0.7 is publishable; PBO < 0.5 means out-of-sample meets or beats median in-sample. Updated automatically when the engine universe rotates (S104.A signal-driven backtester).

Engine backtest certificate
Last regenerated · 2026-06-07 06:30 UTC

0.7754

DSR · deflated Sharpe

0.5050

PBO · prob backtest overfit

Walk-forward
-0.5396
Sharpe
0.37
Max drawdown
-37.1%
Total return
+29.1%
BTC buy & hold
+60.2%
Δ Alpha
-31.1%
Period
2022-04-29 → 2026-06-07 · n=1500
Strategy
Sentealpha signal-driven v2 (WEIGHT_MATRIX_V2 active) · pandas

Notes

factor_exposure skipped — df_factors not provided

IC by regime15h ago
Top signals ranked by |IC| within each market regime. F-66 IC = Spearman correlation between the signal and BTC d+H return. Values ≥ 0.5 are strong by Bailey & López de Prado (2014). Cell name "icir" preserved from the legacy F-66 schema; the metric is actually IC.

Bull rally

#SignalICnconf
1t1_halving_cycle−0.84736OK
2job_openings_per_unemployed−0.82117MEDIUM
3bullish_structure_4h+0.80836OK

Recovery

#SignalICnconf
1btc_power_law_ratio+0.46383OK
2wti_spot_close+0.43954OK
3agsi_eu_aggregate_storage_full_pct+0.43862OK

Bear decline

#SignalICnconf
1equity_breadth_pct+0.54960OK
2wti_spot_close−0.50259OK
3safe_haven_proxy_score−0.41960OK

Capitulation

#SignalICnconf
1m36_eua_carbon_yoy_pct+0.46162OK
2agsi_eu_aggregate_storage_full_pct+0.32890OK
3wti_brent_spread−0.28961OK

The certificate is research-grade. Past performance does not predict future returns. The full RESEARCH_DISCLAIMER is enforced as a universal footer on every page.

← Back to home