S1TREND_UP
Momentum90
Long when the completed daily close is above the close 90 days earlier; otherwise flat.
BTC systematic research · Post-selection historical challenger
C4: a sparse regime router that beat the retained baseline on historical risk-adjusted performance
C4 combines three mutually exclusive long/flat sleeves: Momentum90 in TREND_UP, EMA 50/180 in TREND_DOWN, and an EMA200-filtered RSI2 reversal state in NORMAL_VOL__TRANSITION. Every other regime is flat.
Headline evidence
Performance is interesting. The label matters more.
The figures below are frozen historical research results, not a live or paper track record.
- +182.29%
- 1.100
- -25.25%
- 88
Historical challengerNot independent OOSNo execution authority
Strategy structure
Three sleeves, mutually exclusive by market regime
S2TREND_DOWN
EMA 50/180
Long when daily EMA50 is above EMA180; otherwise flat. TREND_DOWN does not imply a short position.
S3NORMAL_VOL__TRANSITION
RSI2 + EMA200
Long state enters above EMA200 with RSI2 below 10 and exits at or below EMA200 or when RSI2 rises above 70.
C4 is the union of S1, S2, and S3. Unexpected simultaneous sleeve eligibility fails closed. If none of the three frozen sleeve conditions produces long exposure, C4 remains flat.
Cost stress
The edge weakened as turnover became more expensive, but did not disappear
| Cost assumption | Total return | Sharpe | Drawdown note | Trades |
|---|---|---|---|---|
| 5bp / side | +182.29% | 1.100 | -25.25% | 88 |
| 10bp / side | +158.50% | — | — | 88 |
| 20bp / side | +116.72% | 0.850 | Worse than 5bp case | 88 |
At 20bp per transition side, C4 still exceeded the retained EMA 50/200 baseline on Sharpe in the historical comparison, but no longer dominated it on return, drawdown, and Calmar simultaneously.
Why it survived the audit
C4 diversified the historical edge across more trades and more than one sleeve
- 18.80%
- 47.50%
- 21.48%
- 176
Under the frozen 5bp concentration rule, C4 was not trade-concentrated and remained below the 50% year-concentration threshold. Its severe-cost case is less clean, which is why the next meaningful evidence must come from a non-overlapping forward period rather than additional historical rescue.
Limitations
What this page does not claim
- C4 was selected after inspecting earlier historical tournament results, so this is not independent validation.
- The 2022 to July 2026 sample is historical and the 2026 terminal period is partial.
- At 20bp per transition side, C4 becomes year-concentrated under the frozen >50% positive-log-growth rule.
- Bootstrap frequencies are retrospective path-rearrangement diagnostics, not forecast probabilities.
- No leverage, position sizing, stop-loss, take-profit, short exposure, or execution logic is included in these results.
Methods demonstrated
- 01Regime-conditioned strategy research
- 02Deterministic replay
- 03Transaction-cost stress
- 04Trade/year concentration diagnostics
- 05Leave-out robustness
- 06Block resampling
Current status
A challenger worth observing, not a strategy approved for trading
C1 and C4 were the two formal survivors of the frozen historical stress gate. C1 is the cleaner concentration candidate; C4 is the stronger performance candidate. The next research step is to freeze both definitions and observe genuinely new data without changing the historical rules.
BTCUSDTSystematic trading researchRegime conditioningMomentumDual EMARSI2Cost stressBootstrap