All About AI distinguishes predicting a winner from finding positive expected value. Using an Arsenal and Manchester City match, the video converts bookmaker odds into implied probabilities, removes the bookmaker margin and compares the normalized result with prices on Polymarket and Kalshi.
The worked example finds only small differences between the sources. The video stresses that fees, bid-and-ask prices and the quality of the chosen fair-value reference can erase a headline gap, so a model should monitor thresholds instead of treating every disagreement as a trade.
All About AI then asks Codex to sketch a system that retrieves live prediction-market prices, accepts bookmaker inputs, normalizes the data and flags outliers. The example returns no immediately executable edge, illustrating that the useful output of an automated system can be a decision to wait rather than a forced prediction.
Watch on YouTube
