I've completed testing of a new setting for the Uber and Hybrid V3. It has been run on my pc so the timings wont be as accurate as the phones, but the difference isn't overly significant. The purpose of the setting is to better deal with very difficult wheels, in particularly where the scatter is quite bad. For the testing I used a dvd where the velstone mk7 gives virtually random results when using typical roulette computer algorithms (X ball revs until drop, like VB). Of course you aim to avoid such wheels, but it isn't always possible if you are unable to travel.
The results over the same 100 spins were:
1. Basic settings only (no target etc): break even (no significant profit)
2. Target with standard tuning (not dynamic): +24 units
3. New algorithm (not dynamic tuning): +145 units
#1 was pretty much a flat-line chart with no discernable peaks. #2 had marginally clearer data. #3 was much clearer. There was an enormous difference. Keep in mind this is one of the kind of wheels you would normally not bother with because the accuracy is normally not worth your time compared to other wheels.
While this difference may not be as dramatic with different spins or a different wheel calibration, the results are certainly promising. You may know even rotation of the wheel of 20 degrees can be the difference between a wheel being very easy or very difficult to beat.
The current uber version already has part of this implemented, but the option to use it is removed from the menu system because it is too impractical to use. To make it practical requires about 8mb storage on the phone (which wont be a problem), and a bit more coding. Once the new programmer is more familiar with the software we'll get it all done, then I can test more.
How it works is basically better dynamic modelling of the relationship between the variables, in particularly detected ball speed, and the variations in ball deceleration.
Regarding uber progress, for now the new programmer is still learning what the code all does. The code is a lot more complex than when I last changed programmers so its taking longer than anticipated.
