Prediction drift measures how AI model picks change between the T-24h, T-6h, and T-1h collection windows before each NCAA Tournament game. A flip occurs when a model switches its predicted winner between windows. Confidence shifts track changes in stated confidence without a winner change.
Models with high flip rates in the T-24h to T-6h window tend to have lower overall accuracy. The most reliable predictors maintain stable picks at T-24h and T-6h, then update decisively at T-1h when late-breaking information like injury reports becomes available.
The drift chart visualizes each model's prediction stability across all games played. GPT-4o, Gemini 2.5, and Perplexity Sonar Pro each show distinct drift patterns based on their underlying source dependencies and reasoning approaches.
Flip Rate
The percentage of games where a model changed its predicted winner between collection windows. Lower flip rates generally correlate with higher accuracy.
Confidence Shift
A change of 10 or more percentage points in stated confidence between collection windows without changing the predicted winner.
Prediction Stability
The inverse of flip rate. Higher stability indicates more consistent predictions across collection windows.
PREDICTION DRIFT
How picks shift from T-24h to T-4h to T-1h · Flipped picks highlighted in orange
DOES CONFIDENCE INCREASE NEAR TIP-OFF?
We tracked whether each model's confidence language becomes more or less certain as the game approaches. High flip-rate models tend to start overconfident and then reverse - a pattern that erodes reliability.
Models with stable confidence trajectories adjust proportionally to new information rather than swinging based on framing.
GAME-BY-GAME DRIFT
LIVE PICKS
▶Live prediction events will appear here as they happen◆
▶Live prediction events will appear here as they happen◆
