Not one bug was found over a 6 hour period, so everything went well.
We tested on the Mk7 Huxley wheel with Velstone, and ivorine ball. Predictions were roughly 8-9 seconds before the ball fall although we weren't focusing on how early predictions were - it was more about the ball scatter (what happens after the ball falls from the ball track).
Results were as below:
PART 1: With very basic settings, ie no dynamic adjustments for ball deceleration changes, no scatter management, and no target: very poor accuracy. It was difficult to ascertain if there was any edge at all. The equivalent settings of the lite and basic computer were used. To determine this, we observed the raw chart.
PART 2: Basic "correct revolution" auto targeting with adjustments for ball deceleration rate changes: The computer correctly noted the ball position at the end of the spin almost always within 1-3 pockets accuracy. When it was incorrect, the computer perfectly automatically adjusted. While the computer did everything perfectly, the ball scatter was "all over the place". The accuracy was roughly about 60% wins betting 18 numbers. To determine this, we observed the COM chart. So there was an edge, but only slight.
PART 3: Auto target for targeting diamond hits with adjustments for ball deceleration rate changes, and varying ball bounce for different rotor speeds and minimization of ball scatter: The computer correctly noted the ball position at the end of the spin almost always within 1-3 pockets. It did all the adjustments it needed perfectly. Even with the terrible scatter, the accuracy was approximately 90% wins covering 15 numbers which is roughly the same accuracy achieved in the last public demo. "Risk" was called on average about half of the spins.
NOTES ABOUT THE WHEEL:
Whenever the wheel is set up, you never know how the ball will fall until you actually test over 50 or so spins. Less than 1mm tilt makes a significant difference to not only where the ball falls, but also HOW it falls (including scatter, amount of floaters etc). This makes results different each time. Overall the ball scatter was abysmal, hence the poor performance of the basics settings. Observing the wheel with traditional methods would indicate the scatter was completely unmanageable.
The ball deceleration changes were considerable throughout the meeting. At one stage the ball was completing two fewer revolutions than normal from the point of prediction, and then soon changed back (from ball track grit). So that's about 5 revolutions when there would normally be 7 - so it's quite big difference.
The computer correctly noted the ball position at the end of the spin almost always within 1-3 pockets. When it was incorrect, the computer perfectly automatically adjusted
Overall the wheel rating was 9/10 which is probably the most difficult I've seen it. In such circumstances, it was not a surprise to see the basic algorithms perform poorly. Although it was great to see the advanced settings work perfectly.
While there were no problems/bugs, I did note one tweak that will slightly increase accuracy, and another that will make application even easier. These are being implemented now. At least one is a very minor tweak to the point where I could almost do it myself without the programmer (being done tonight). The other mod might take a bit longer, but it is not overly important anyway.
SHIPPING: I'm putting together packages for shipping. It takes me about 30 minutes to install each phone from scratch, so I wont get them all done overnight. And I have a lot of other work to do so I cant just do all shipping in one day. Shipping will be done progressively over weeks for those who are waiting. I'll contact you just before your parcel is shipped to confirm delivery address.
So essentially for the latest version, it's done. If there are bugs, 6 hours of testing didnt reveal any, in addition to other testing I've done. If you want to update your computer, follow the typical procedures and send your hardware back. Remember send only the phone itself - NO BATTERY because I cannot ship lithium batteries.
