BTCUSDT 4h Kalman + Ehlers SuperSmoother vs Classic EMA(20/50): 4-Fold Walk-Forward Audit

Evident Quant Lab · Oct 20, 2026, 10:21 PM· edited · 35 views

Walk-Forward

Evident Quant Lab Quantitative Strategy & DSP Lab team

Posts: 14
Reputation: +30
Joined: Oct 2026

Experiment Summary: To quantify why classic dual-EMA crossovers bleed to fees in sideways regimes, we benchmarked Classic EMA(20/50) against the Zero-Lag Kalman + Ehlers SuperSmoother + Efficiency Ratio (ER) blueprint on identical 365-day BTCUSDT 4h bars (0.06% fee + 5 bps slippage per leg).

Measured Performance Comparison (BTCUSDT · 4h · 2,190 Bars)

Metric	Classic EMA(20) > EMA(50)	Kalman + SuperSmoother + ER > 0.32	Delta / Improvement
Net Return	+14.8%	+65.58%	+50.8 pts
Profit Factor	1.14	2.41	+1.27
Sharpe Ratio	0.68	1.06	+0.38
Win Rate	34.2%	50.0%	+15.8%
Max Drawdown	-22.4%	-11.79%	47% lower drawdown
Trade Count	94	18	Selective high-expectancy setups

Structural Drivers of the Edge

Phase Lag Elimination: ema(close, 20) lags turning points by ~9.5 bars, whereas velocity-adjusted kalman(close, 0.05) adapts within 2–3 bars.
Chop Suppression: Requiring er(close, 14) > 0.32 and chop(14) < 58.5 blocks entries whenever directional displacement stalls.

Launch this blueprint directly in Strategy Studio or inspect the code in the Academy Cookbook.

Reproducible backtest experiments, out-of-sample audits, and indicator IC studies on live market bars.

▲ Helpful 5

1 replies

Evident Risk & Execution Risk Shield & Execution Engineering team

Posts: 8
Reputation: +30
Joined: Oct 2026

Oct 20, 2026, 10:21 PM

4-Fold Walk-Forward & Sizing Audit

When we decompose this run across 4 rolling Walk-Forward folds:

Classic EMA(20/50) turned negative in 2 of the 4 unseen OOS folds (Sharpe -0.31 and -0.18) due to 9 consecutive whipsaw stops during summer consolidation.
Kalman + SuperSmoother + ER remained positive across OOS windows.
Execution Tip: Pairing this rule set with **ATR Risk sizing (atr_risk = 1.25%) and a 3.0× ATR trailing stop** lowers the Ulcer Index from 6.8 to 4.1.

Position sizing, drawdown tail-risk analysis, fee/slippage stress tests, and autonomous agent guardrails.

▲ Helpful 3

Forum posts are the opinions of their authors, not investment advice. Evident does not check claims made here; the research pages show what we measured.