📋 Signal Scorecards
Transparent signal quality metrics. Every signal we publish has a scorecard showing accuracy, sample size, calibration details, and methodology version history. We show what works and what doesn't.
Market Regime
68% 30dAccuracy = % of time the predicted regime matches the realized 4h price direction. Buy → positive return, Sell → negativ...
Composite Signal
63% 30dComposite score > +0.1 as buy signal, < -0.1 as sell signal. Measured against 1h forward returns....
Orderbook Imbalance
62% 30dSignal designed for 5m–1h horizons. Accuracy measured as hit rate when |imbalance| > 0.3 predicts next-bar direction....
Funding Rate
59% 30dMeasured as contrarian hit rate: extreme funding (|z| > 2) predicting mean-reversion within 4h....
Volatility Analysis
70% 30dVolatility regime classification accuracy: % of time high-vol label matches next-4h realized vol > 50th percentile....
Liquidation Cascades
56% 30dHit rate for liquidation cascade detection: % of detected cascades (>$1M in 5min) that led to >0.5% reversal within 1h....
Whale Flow
58% 30dLarge order detection (>$500K) directional accuracy. Measured as % of whale orders where price moves in the same directi...
Gamma Exposure (GEX)
60% 30dGEX pin accuracy: % of time that price stayed within the predicted GEX pin range (±1 ATR of max gamma strike) for 4h....
Cross-Asset Correlation
65% 30dRegime transition prediction: % of time that de-correlation events (coherence < 0.7) preceded regime changes within 4h....
Signal Safety (Coherence)
80% 30dData integrity validation accuracy: % of time that low coherence scores correctly identified data quality issues....
Why Scorecards?
Most signal providers show you the signal but hide the track record. Scorecards are our answer: every signal gets measured against real market outcomes, with sample sizes large enough to be statistically meaningful.
Reading the numbers:
- 65%+ — Strong signal (consistently profitable after costs)
- 55–65% — Moderate signal (profitable with good risk management)
- <55% — Weak/informational (use as confirmation, not primary signal)
All accuracy numbers are computed on out-of-sample data with no backfill bias. See Methodology for computation details.
Don't just stare at the dashboard. Automate it.
Every metric on this page is available via our sub-millisecond API.
Build trading bots, backtest strategies, and power AI agents with institutional-grade data.