Understanding AI Confidence Scores in Trading

What Are Confidence Scores?

A confidence score represents how certain our AI is about a prediction.

Interpreting Scores

80%+ (High Confidence)

• •Strong model agreement
• •Clear historical patterns
• •Significant detected edge
• •Consider larger position size

70-80% (Good Confidence)

• •Solid signal
• •Standard position sizing
• •Most of our surfaced signals

65-70% (Moderate Confidence)

• •Smaller edge detected
• •Consider smaller size
• •May require additional research

Below 65%

• •Not surfaced to users
• •Insufficient confidence
• •Too close to random chance

How Confidence Relates to Sizing

Kelly Criterion suggests sizing based on edge and confidence:

Higher confidence → Larger position (within limits)

Lower confidence → Smaller position

Important Notes

• •High confidence ≠ guaranteed win
• •Even 80% confidence means 20% chance of loss
• •Always use proper bankroll management
• •Diversify across multiple signals