Hybrid updates for Slingshot wheels

I’ve finished most of the testing for the Slingshot wheel mod. The wheel’s scatter is rated 63%, and predictions are 20 seconds before the ball falls. Considering the minimum scatter is usually 70% and average conditions allow getting predictions 7 seconds before the ball falls, it’s a very difficult wheel to test on. This is not even with consideration to how the rotor changes speed after no more bets is called.

It appears an important requirement to beat this wheel is targeting specific ball fall areas. There’s about an 80% accuracy increase when targeting a particular ball drop area, but only when it falls at a specific speed. This is about 15% of spins. For this particular wheel, the data suggests it’s possible only on one direction.

I haven’t yet assessed what’s possible with later predictions, but there isn’t much point as no more bets on this particular wheel is called about 18 seconds before the ball falls.

My experience with slingshot wheels in generally is that some can be beaten, and some appear impossible to beat. It depends on the individual wheel’s settings and properties. No wheel may be impossible to beat, but usually you just move on to easier wheels if one is too difficult. This particular wheel is mostly  a personal challenge and opportunity to improve algorithms. Although on this note, if profiting is your sole goal, remember to forget the difficult wheels and just focus on easy wheels.

To this point, the test wheel does appear possible to beat but with a comparative struggle. To get a better idea, more research is needed when I have time (probably within a week). Next I need to better assess the variation of rotor speed after no more bets is called.

I have about 6 hours of recorded video to test with, which is perfect for development because in one direction, the wheel appears almost impossible to beat with predictions this early. And in the other direction, there’s a significant edge with ball fall area targeting. If the wheel was too easy to beat, there wouldn’t be much to learn.