Even though the tennis world has moved on to Wimbledon, we thought we would share how our French Open predictions performed. The chart below compares our predictions against the actual outcomes.
- Predicted Round-by-Round Win % (Red): How our model assessed the chances of predicting match outcomes based on our original predictions. For example, if Nadal had a 98% chance of winning in the first round, this method would capture that as a 98% chance of correctly predicting the outcome. Based on this, the overall percentage generally drops round to round as the likelihood for any favorite continually drops.
- Prediction vs. Actual (Blue): How our model fared against the actual match outcomes based on our original predictions.
- Updated Prediction vs. Actual (Green): How our model fared against the actual match outcomes based on updating our predictions after each round.
In a well-tuned model with enough data points, you would expect the ‘Predicted Round-by-Round Win % (Red)’ vs. ‘Prediction vs. Actual (Blue)’ to be close to each other. And the ‘Predicted Round-by-Round Win % (Red)’ and ‘Prediction vs. Actual (Blue)’ result in an overall record of 83.4-43.6 (65.7%) and 83-44 (65.4%), respectively. The ability to predict each match after each previous round is complete is even higher at 96-31 (75.6% correct). This shows that the model fared well and provides confidence leading into our Wimbledon predictions.