Previous article was showing effectives of FFA computer in attempt to predict where the ball will hit rotor. Then results were simulated to show what would be final outcome after 10,000 spins. On this graph is displayed the FFA’s real performance on the same wheel.

The FFA was set to get the best accuracy which means that the prediction was set to be later in time. Averaging to 6.8 seconds before then the ball drops.

After playing 5 spins (only) to let the system learn the wheel parameters, I decided to set the offset to +16 pockets. The graphs shows exact final results where the ball was stopping in relation to predicted number. Everything marked with green color is where the advantage is. After analyzing data if someone was playing according to this prediction and covering 11 pockets from -5 to +5 from prediction, he would play with advantage of 53%.

It is average of 4.24 hits per 100 spins for each pocket in green area. It means that for playing $100 per spin in average the player would profit $53 each spin. If he was playing a bit more to the right and covering 6 pockets the advantage would be slightly better 58% because the area marked with yellow color would be avoided. This is a typical problem when playing in real environment. Sometimes players decision where to play is not the best or he wouldn’t be able to cover all numbers as he wants to.

53% and 58% are real advantage figures, after the system overcome negative 2.7% house advantage.

To achieve such advantage of course that we need accurate prediction.

During the test I took notes of only first 30 spins to see how well the system predicts ball failing point.