I have an idea which might help you test out new roulette computers, and to help calibrate the computer to win.
How about if you could take photos of the wheel while testing? I mean photos with exact time stamps. Afterwards, by looking at the photos and the times (in milliseconds) when they where taken, you could easily calculate the speed of the ball and the wheel and how their speeds changed over time. You could compare this with the prognosis your computer made. You might also make the assumption that the same ball/wheel system will behave similarly the next time you play on it, and use those exact speed measurements to calibrate you roulette computer specifically for the next session.
A problem might be that the house rules of most casinos do not allow photography. But maybe you could find a solution to that.
In order to do this, you would need a pocket computer which costs about 500 euros, and a webcam for less than 100 euros. And a simple program which I could provide for free, if you would be interested.
Now, I’ve suggested some “exotic” technical aids before, and you’ve never really been interested. You use to say that you see what happens and don’t need any exact measurements. But I think you might want to consider using some technology in order to exactly verify what is happening. It is only recently that cheap electronics allow for measuring roulette ball speed with millimeter and millisecond precision.
Thanks for suggestions.
The FF has no problem learning wheel while we play.
It takes only few spins for reasonable accuracy.
The FFA dos it even faster assuming that your first clocked spin doesn’t have greater clocking error then 50ms.
For example first prediction may be 31 if we repeat it. it may be 7 then 3 . If we clock it different place to see how the system will respond to different ball speed it may predict 7, but after few more spins it will predict at any place something as 3,26,3,0,35,26,26,3,15,3€¦etc.
First few spins the system gets averages, and predicts for any ball speed using average but after it becomes more precised learns and defines parameters for each ball timings.
90% inaccuracy of IQE6 comes from wheel imperfections or tilt.
For example the ball of 1000ms travels 6 rotations and 1050 may travel ¼ less.
But if wheel is tilted it may travel also 6 rotations and time for that maybe slightly longer which makes it even worst because the rotor will move more then expected.
Another problem is that ball with same speed may end up differently.
Still on reasonable wheel we should average to 80% hits within 18 pockets and 60% of those within 9, which makes it 50% to 9 pockets sector.
It is good enough, therefore if scatter is reasonable advantage can be recognized and used.