AI Accuracy Dashboard

Calibration Dashboard

AI accuracy

Brier score, Platt scaling parameters, and confidence calibration metrics for BWC-generated signals.

Brier Score

0.596

Poor

Platt Scaling

a=-0.17 b=-1.34

n=71

Graded Signals

490

45W / 445L / 0B · 9.2% WR

Avg Confidence

0.822

vs 9.2% actual win rate

Calibration table

Confidence Bin	Count	Predicted	Actual Frequency	Bias
0–20%	0	10%	—	—
20–40%	0	30%	—	—
40–60%	3	50%	0%	-50.0%
60–80%	37	70%	24%	-45.7%
80–100%	31	90%	10%	-80.3%

Outcome distribution

Failure Mode	Count	% Total	Avg R	Avg Brier
tp1 hit	45	9.2%	3.25R	0.016
tp2 hit	0	0.0%	0.00R	0.000
stop loss	187	38.2%	-1.00R	0.149
expired	203	41.4%	0.21R	0.068

Recent graded signals

Symbol	Side	Confidence	Calibrated	Outcome	Failure	R	Brier
ETHUSDT	LONG	—	—	LOSS	stop loss	-1.00R	—
DASHUSDT	LONG	—	—	LOSS	expired unfilled	-0.03R	—
VVVUSDT	LONG

Methodology

Brier score measures the mean squared difference between predicted probability and actual outcome (1 = win, 0 = loss). Scores range from 0 (perfect) to 1 (worst). Platt scaling fits a logistic regression to adjust model confidence toward empirical frequencies. The calibration table compares predicted vs actual win rates within each confidence bucket — a perfectly calibrated model would show actual frequency matching the predicted bin midpoint.