Review of Forester's device

[Bago]

Hi Forester,

I would be interested to make a review about your lattest device with audio feature.
I just want to test it, make a review with what i think is good, what can be improved etc.
I do not want to keep the device, after the testing done, i will ship it back to you.
Tell me how much you need to send the device.
I do not need any earpiece nor the LPS loop, i purchased two pairs when i had Stefano's tic tac toe phone, in case one broke.

Regards.

[Fatgambler]

Bago, first of all it is not that easy to win with a roulette computer, as you see on tv/stefano claims. In fact it takes couple of months to learn ff system well. What is said you need to have good experince before you can write a fair review. There are poeple outhere that have got his FFA system since day one, and they previously own FFZ. And these people that can do a fair review. If there was something bad with the system these poeple would already told him. I have known forester through a barely a year and trust me, if there is something that needs to improve, he will do it.

I'm a proud owner of ffz all i can say it truly delivers. If i was you Bago i would buy the FF(A/Z), you won't be disapointed.

[forester]

Yes Fatgambler,

I am working on improving program past few days.
As some already know there was only one version of FFA.
Now I converted program to be used with ZAP.

I start looking what can be done better and I found it.
It looks good but it needs more testing, few thousands of spins, I already did close to 1000. Soon I will write more about it.
For FFZ it is ok but problem is with FFA's design, now I can't just send you new chip because it can't be easy replaced. Option one is to send unit back and the other option is to buy programmer because PCB has connections for programmer and the chip can be in circuit programmed.

Advantage of having programmer is that you can see all details you played, you can see how system learned wheel, and you can readjust it easy and save games as many as you want.
I do not mind for code because I believe nobody smart will spend so much time to try to understand it, it is more then 100 calculations and few thousand commands, the program also works together with a program on the other memory chip. And anyway not many people can write firmware for microprocessor.

Bago, thanks for the offer, but I am not after reviews and I do not build systems for free.
Anyway I like more constructive complains then positive reviews.

I am not up to putting someone down or arguing. I was always presenting facts.
I see that you now understand much more and if now you go through arguments what Stefano presented you will see how much right I was.
From to be able to notice ball with FFZ, to purpose of single spin test.

I was only wrong 3 years ago when I thought it would be imposable to get enough accurate ball timing, since when I said IQE6 error correction brings ball timing close to accuracy of laser measurement I was correct.
FF system has it all, fast hardware for real time applications (not OS on mobile phone); error correction, great algorithms for calculation, no losing time in initiating audio player (FFA), and I still do not present it as a miracle.

If you want to experiment with FFA, few days ago I made an option to buy FFA without RF and HA. You still can plug in induction loop only it is not for team play.

[Bago]

Show me the scatter graph you get with your FFA with Stefano's 2 dvds that he gives with the phone. The first where he calculates offset and second one actual play, with 1 in 33 hit rate...

[Rollo]

the other option is to buy programmer because PCB has connections for programmer and the chip can be in circuit programmed.

Advantage of having programmer is that you can see all details you played, you can see how system learned wheel, and you can readjust it easy and save games as many as you want.

Is it possible to get such statistics out of you roulette computer using a chip-programmer? I suppose it is only "in laboratory" while having the programmer connected to the chip. But maybe even that could be made practical also in the casino? If you think data is helpful for the user, then maybe you could implement support for data storage and transfer-to-PC after the game? It could be helpful for observing user mistakes while learning, as well as learning characteristics of the wheel-ball-system. But probably it'd be too advanced for what the "market" is looking for...

You're not paranoid about the code, and for the right reasons since it is no easy task to reverse engineer a thing like this for anyone less than an experienced high paid specialist (although the puzzle might waste a few hundred hours of that guy Steven Humorouz time...) However, I think there are ways to make the code even physically (within reasons) inaccessible to programmers if you WOULD get paranoid about it...

[forester]

Yes it may be useful.
There is a bit of problem with FFA, because it doesn't start from any value but from players first clocking. If first few spins he makes similar mistakes then it may take for some time that system corrects it. Then prediction gradually may shift.
With programmer after one play I can set parameters so the system has close enough  parameters from the start.


Bago I did this today
First is prediction hits to rotor , next is prediction final result.

You can see how good needs to be prediction to rotor that after scatter we can get something. Even than in this particular case it is nothing special. 90% of hits are from (-18) to zero.

Prediction was accuracy 3 setting and system says 9.3 sec to the end, I clocked only ½ of rotor. Spins were not long enough for full rotor clocking.

For you is the best connect on msn and I will show you with camera how system predicts these spins.

What I am chasing now is that system at start and at the end of spins doesn't predict same. There is about 0-7 pockets difference it depends at which speed it catches the ball.
If I fix it I believe most of hits around -1 position would ovelap wit hits at -7.

[forester]

Code: [Select]

-18 xxx
-17 xxxx
-16 xxx
-15 xxxx
-14 xxxxxxx
-13 xxxxx
-12 xxxx
-11 xxxx
-10 xxxxx
-9 xxxx
-8 xxxxxxxx
-7 xxxxxxxxx
-6 xxxxx
-5 xxxxxxx
-4 xxxxxx
-3 xxx
-2 xx
-1 xx
0 xxx
1 xx
2 x
3
4 x
5 x
6 x
7
8 x
9
10
11 xx
12
13
14
15
16
17
18 xx

Code: [Select]

-18 x
-17 xx
-16 xx
-15 xx
-14 x
-13 xxxx
-12 xxx
-11 xxxx
-10 xx
-9 xxxx
-8 xxx
-7 xxx
-6 xxx
-5 xx
-4 xx
-3 xxxxx
-2 xxxxx
-1 xxxxxx
0 xxxxx
1 xxx
2
3 xxxxxx
4 x
5 xxxxx
6 xxx
7 xx
8 x
9 xx
10 xxxx
11 x
12
13 xxx
14 xx
15 x
16
17 xx
18 xxx

[Bago]

Your results are better than the 1 in 33 hit rate from Stefano against an old wheel spun consistantly at around 6sec/rev EACH SPIN.

[forester]

My point is that he doesn't have prediction –hits to rotor better then random.
Another point is that there is no way for computer to figure out is wheel tilted or not as he claims, based on (result –final number) calculation is completely different and because of high ball bouncing we can't compare results. The difference in calculation is huge. For example on tilted wheel the difference in between measured 1000 ms ball and 1200 ms ball is only rotor movement for 1.2 sec.
(That's why tilted wheel predictions is easier and doesn't require so much of precision)
While on leveled wheel it is rotor movement plus 37 pockets.
It is about 4 ms per pocket.


Stefano can't have 1:33 hit rate wit his computer.
Look his first 50 spins. From start he insists to have offset +11 even there wasn't highest build up there. On the next look to graph he looked area with minus to zero saying it has the lowest pick and said he should set it right opposite to that. His 100 spins were recoded earlier then his 50 spins for set up, and that is the only reason why he selected +11.
On video he says computer will automatically select different point. It would be wrong so why to claim that computer can do it if even at his demo it couldn't do it.
And finally if he claims +11 is jump then why first 50 spins ball doesn't hit rotor where is prediction or around that point and why at additional 100 spins it doesn't hit 11 pockets earlier.

On the graph if I played 7 pockets around zero point I would have 1:28 hit rate.

[forester]

I forgot to say that this was done with experimental program.
I am still working on it. Basically testing how much results can deviate if I run 100 spins then repeat them. For example if I have 50 spins all predicted zero.
If I repeat them it is ok if result deviates around zero but it is not ok if average deviation is around 5-6 pockets earlier or later.
So I am experimenting how to make it stable as much as it is possible.
I did run 2x 100 spins in next test and results pretty much overlap.
Also I tried to clock full rotor as much as it was possible. Based on 7 pockets play it would be 1:24 hit rate. With 200 spins the advantage is more visible.

Code: [Select]

-18 xxxx
-17 xxxxx
-16 xxxxxxxx
-15 xxx
-14 xx
-13 xxxx
-12 xxxx
-11 xxxxxx
-10 xxxx
-9 xxxxxxx
-8 xxxxxxxxx
-7 xxxx
-6 xxxxxxx
-5 xxxxxxxx
-4 xxx
-3 xxxxxxxx
-2 xxxxxxxxxxxxxx
-1 xxxxxxxx
0 xxxxxxxxxxxxx
1 xxxxxxxx
2 xxxxxxx
3 xxxxxxxxxx
4 xxxxx
5 xxxx
6 xxx
7 xxxxxx
8 xxxxxx
9 xxxxx
10 xx
11 xxx
12 xx
13 xxx
14 x
15 xxxxx
16 x
17 xxxx
18 xxxx