Day 27: Selective Participation

Testing whether selective trading based on funding regime or volatility improves outcomes.

Published

Mar 8, 2026

Summary

Result: NOT DEPLOYABLE — Marginal improvement only.

Strategy	Expected PnL	vs Baseline	Participation
Always trade	7.93 bps	—	100%
Low volatility only	8.68 bps	+0.75	81%
High funding only	7.98 bps	+0.05	11%
Extreme regimes	7.94 bps	+0.01	52%
Low funding only	7.93 bps	+0.00	45%
Regime-momentum aligned	7.93 bps	+0.00	26%
High volatility only	3.44 bps	-4.49	4%

Verdict: NOT_DEPLOYABLE — The improvement (+0.75 bps) doesn’t meet the +1 bps threshold.

The Question

Should we trade all the time, or only when conditions are favorable?

Hypotheses: 1. Funding regime: Trade only when funding is extreme (high cost to hold = better risk/reward) 2. Volatility regime: Trade only in low-volatility periods (less slippage) 3. Momentum alignment: Trade only when momentum aligns with funding direction

Methodology

Funding Regime Classification

Using percentile-based classification on Binance funding rates: - High: > 75th percentile (> 0.01%) - Neutral: 25th-75th percentile - Low: < 25th percentile (< 0.002%)

Distribution: - Low: 53.6% - Neutral: 41.1% - High: 5.3%

Volatility Regime

Using 5-minute volatility ratio (current 5m vol / 20-period vol): - High volatility: ratio > 1.2 - Low volatility: ratio < 0.8 - Normal: 0.8 - 1.2

Walk-Forward Backtest

Data: 5-minute candles, Jan 2022 – Feb 2026
Quote distance: 10 bps (from Day 25)
Test years: 2023, 2024, 2025

Results

Funding Regime Strategies

None of the funding-based strategies showed meaningful improvement: - High funding only: +0.05 bps (11% participation) - Low funding only: +0.00 bps (45% participation) - Extreme regimes: +0.01 bps (52% participation)

Conclusion: Funding rate at 8-hour intervals is too coarse a signal for 5-minute trading.

Volatility Strategies

Low volatility: +0.75 bps improvement - Participation: 81% - Net improvement: modest

High volatility: -4.49 bps (terrible) - Participation: 4% - Key insight: High volatility periods are poison for maker orders

Momentum Alignment

Regime-momentum alignment: +0.00 bps (no improvement)

Key Insights

Funding regime doesn’t help: 8-hour funding intervals are too coarse for 5-minute trading decisions
Low volatility is slightly better: +0.75 bps improvement, but requires 81% participation - marginal edge
Avoid high volatility: -4.49 bps is a massive penalty. Could use this as a filter (skip top 4% most volatile periods)
Selective participation is not worth it: The improvement is too small relative to complexity

Recommendation

Keep trading always. The marginal gains from filtering don’t justify: - Reduced capital efficiency - Implementation complexity - Regime misclassification risk

However, could add a simple rule: skip periods when 5-minute volatility > 1.5x 20-period average (avoids the -4.49 bps penalty).

Files

research.py — Full backtest code
backtest_results.csv — Year-by-year results
regime_stats.json — Funding regime statistics
verdict.json — Decision record