@ turbo

[falkor2k15]

The way to test it properly would be like this :

1. Wait until a number has repeated and bet on it.
2. When you make the bet as instructed in (1) place another bet on a single number selected at random.
3. Continue to place more bets on numbers which repeat, and everytime you do this select another number at random and bet on it too.
4. When you get a win, stop.
5. Repeat steps 1-4 several dozen times (at least), then compare data from both groups.

So you have a "control group" (betting random numbers) which is betting in tandem with the "treatment group" (betting repeats). The only difference between the 2 systems is the numbers you're betting on (the amount of numbers you bet on is the same in both). You could also use a progression if you like, but it must be exactly the same for both systems.

This is a basic requirement for good experimental design; everything must stay the same for both "experiments" except for the one attribute of interest (in this case the bet selection). The hypothesis is that betting on repeats are better, but better than what? In this case the experiment is trying to find out whether it's better than betting randomly. So if the results show that you win more often when betting on repeats you can be sure that it's actually betting on the repeats which makes the difference and not something else like the amount of numbers or the money management, because these are the same in the control group. But if there is no significant difference it shows picking the repeats has no effect; you might as well pick any random numbers.

All this might seem obvious but hardly anyone seems to do a proper test like this for bet selections. 

My guess is that Turbo hasn't done a test like this for his system, and that it's only the progression which is doing the work.

You haven't understood, but neither did I till I ran hundreds of tests over the course of 3 years. Repeats and uniques are artificial constructs - can be one or the other depending on the viewpoint of random past spins. Therefore, you aren't predicting a unique or repeat, you are hoping to win on a 2/37 chance bet from 2 different numbers that you believe have different properties when this is all simply an illusion: each number is created equally for each independent trial.

Marking relationships between numbers based on uniques and repeats simply indicates how long we are likely to wait for a win if we keep betting in a series of increasing odds with lower payouts. It doesn't help us predict anything on a spin-by-spin or even game-by-game basis, and it doesn't help us escape the house limits, and it doesn't help us profit nor escape break even. So instead of calculating all probabilities (with proportionate payout odds) on paper the uniques/repeats act as rough indicator for your break even game.

What's the probability of 2 dozens in 2 spins? I can either calculate it or run a series of simulations and let the uniques/repeats tell us. That's how we can calculate cycle lengths or cycles defined by orders without using a tree diagram - but no prediction and no profit - just stats about a break even game.

[wiggy]

My guess is that Turbo hasn't done a test like this for his system, and that it's only the progression which is doing the work.

I was testing using some Wiesbaden spins the other day and I came across a group of 45 numbers that had 18 repeats before the first 3peater. The problem I have with Turbo's method is how you would handle that in a live Casino environment and especially using a progression where you may be placing different chips on different numbers. If it's not practical, it's not playable for a lot of different reasons and you at least have to give MrJ credit for pointing this out numerous times over the years. What works on the kitchen table / sims counts for nothing if you can't play it for real.

[nottophammer]

I was testing using some Wiesbaden spins the other day and I came across a group of 45 numbers that had 18 repeats before the first 3peater. The problem I have with Turbo's method is how you would handle that in a live Casino environment and especially using a progression where you may be placing different chips on different numbers. If it's not practical, it's not playable for a lot of different reasons and you at least have to give MrJ credit for pointing this out numerous times over the years. What works on the kitchen table / sims counts for nothing if you can't play it for real.

Wiggy last week was in Grosvenor Luton, on touchscreen playing the wheel in Blackpool, betting the 4 hottest with different values piece of piss.


Only short set as had to get to airport to do a pick up

[falkor2k15]

I think there's only one way unique & repeat indicators might possibly help us:

If you play only 4 games in your lifetime and you wanted to win all of them. You could sit around waiting for 20 virtual losses on betting numbers (1s to becomes 2s) and then start betting with a progression. You would then be pretty much guaranteed to get a win within the next 5 spins, but it would involve waiting your whole life for it to happen.

But if you refer to my quadruplets topic you will see that you can arrive at a similar extreme situation by only having to wait for 8 cycles:
1 3 6 2 7 8 9 4 5.... now bet all and guaranteed a win.

You aren't winning through prediction but by waiting for an extreme situation that is a bit more predictable than X amounts of reds in a row to be followed by an (independent) black.

And you don't win on events that are more extreme than other less extreme events - you win (over a series of spins) because you have encountered the most extreme event.

It's still an incredible amount of waiting time - though not as much as 20 numbers bets. The cycles are helping you keep track of all different combined bets - real or virtual - of different ratios/payout odds.

If I get a win on order 1 I can carry over the last 8 uniques to the next game and be in the same situation still:
1 3 6 2 7 8 9 4 5.... 1

3 6 2 7 8 9 4 5 1...
I can't do that after 22 reds in a row followed by a black.

So that helps to maintain an extreme situation and reduce waiting time between games.

And since the pigeons are not equal with quadruplets, if we begin a new cycle defined by options 4-10:
6...

We are then already at another extreme situation where we could bet options 1-5 (covering 6) and guarantee a win on a unique (mostly) instead of a repeat and within the house limits.

So that's the next direction I'll be taking my testing and simulations.

Incidentally, Dyksexlic said we should have less pigeonholes than pigeons, such as SAME or DIFFERENT for a repeat, but I already tested and this doesn't help.

[wiggy]

Wiggy last week was in Grosvenor Luton, on touchscreen playing the wheel in Blackpool, betting the 4 hottest with different values piece of piss.


Only short set as had to get to airport to do a pick up

How's it going Notto? Let's hope we win tonight against Columbia. 

Yeah, I would have won with your numbers as well, so it's all good. I guess I am making a point that roulette can be a cruel game and kick you in the nuts when you least expect it!  If you had 18 doubles, then you can't play them all and one day you are going to pick the wrong ones and keep picking the wrong ones and an aggressive progression is going to land someone in a whole heap of trouble. Some old posters like 'Number Six' were wise suggesting using a divisor for hot numbers. Anyway, It's all there in the archives.

[Joe]

@CoderJoe....that's a nice idea, but how long would you need to test to arrive at a conclusion. For example, if you were just testing 1/2/3 numbers, the variance could make both ways seem like winners even though at least one of them obviously isn't.

Good question, but a proper statistical tests are powerful so you don't need many trials. 100 will be plenty. I'm not talking about a 100 spins, but 100 sessions, where the first win ends a session. So suppose you get some spins and start betting at the first repeat :

  6
30
23
14
23         first repeat here #23. Place 1 unit on #23 and another unit on a randomly selected number. Say this is #3.

  9  loss for both bets
18 loss for both bets
10 loss for both bets
19 loss for both bets
16 loss for both bets
  6  Another repeat, so put an additional chip on #6 and select another random number, say this is #14. Continue getting spins with both systems betting 2 numbers each (#23 & #6 for the repeats system, and #3 & #14 for the random system.

31 loss for both
  5 loss for both
23 WIN for the repeat system.

So this is one trial. But there are some options here about what data to record. You could choose to count this as +1 for the repeat system because it got a win before the random system. Now reset the spins and start tracking again until either system wins, then mark either a W or L for each system. After the 100 trials you would then calculate the proportion of wins for each system. At this stage you would do a statistical test (I won't go into details here) which tests for statistical significance, which basically tells you whether the result is due to chance or whether there is a real difference.

Alternatively, after the first win you could count the number of spins it took to get that win for the winning system, and then carry on getting spins until the other system won too, then record the number of spins it took for the losing system to also win. e.g. in the example the repeat system won after 9 spins, so record a 9 for it and continue getting spins. Say the random system won 4 spins later, you would then record a 13 for it. At the end of the 100 trials you have a list of pairs of numbers

9,  13
12, 5,
7, 9,
....
....
etc,

These numbers represent the number of spins it took for a win for each of the systems. If the repeat system has any merit then the average waiting time for a win should be smaller than the waiting time for the random system, and again you would use a statistical test which will tell you whether the results you got are likely due to chance, or not.

[Joe]

You haven't understood, but neither did I till I ran hundreds of tests over the course of 3 years. Repeats and uniques are artificial constructs - can be one or the other depending on the viewpoint of random past spins. Therefore, you aren't predicting a unique or repeat, you are hoping to win on a 2/37 chance bet from 2 different numbers that you believe have different properties when this is all simply an illusion: each number is created equally for each independent trial.

hi falkor, what haven't I understood and what does this have to do with the test I suggested? Your post is confusing.

The point of the test is to find out whether playing repeaters is better than betting randomly. You can just follow the logic of independent trials and say it's a waste of time or you can believe Turbo who also makes apparently persuasive arguments, or you can choose not to believe anyone and test for yourself (preferred). But you have to test properly using the correct methodology otherwise the results can be misleading.

link:s://:.dummies.com/education/science/designing-experiments-using-the-scientific-method/

You say you've done 100's of tests but where are the results and how did you do them?

[wiggy]

Thanks very much for the reply and explanation CoderJoe 
I am looking into betting hot numbers again, but I just want to concentrate on a few numbers using the last dozen or so spins. That second repeat seems to come in waves using a 12 spin rolling basis. I will do some tests based on what you have shown. Thanks.

[cht]

Save you the trouble. It won't work. Everybody knows that.
Just like this test and many many others.
link:s://:.rouletteforum.cc/index.php?topic=20411.msg204296#msg204296

[Scarface]

Coderjoe, thanks for your reply!  Comparing repeaters vs random data would be a good way to approach.

I do think it can be compared to other bet selections too, like singles.  Although difficult and time consuming.  Let's say a single number goes missing for 360 spins in x amount of sessions played...that would be similar to a dozen numbers missing for 30 spins...both are down 360 units.  Hard to imagine that playing 2s becoming 3s in repeaters can have that type of negative variance.  Maybe it can, but the problem would be it's very time consuming to test 1000 sessions to find out 

Something I have tested many times on paper.  Tracking numbers as they hit.  So far, I've always seen the "first place" repeater jump ahead of "2nd place repeaters" by at least 2 everytime.  For example, a number may repeat for the 5th time, while the next hottest repeater is on its 3rd repeat. 

[Scarface]

Falkner, your tests are not applicable to what we are talking about.  You seem to be stuck on testing dozens.

Try this.  Track all numbers until you get first repeat.  Now play that repeat.  Add new repeats as they appear.  When one of your numbers hit, record the data.  Now play the hit number...this number is the first one that repeated a 2nd time.  Add any new 2nd time repeaters until you get a hit, then record data.

Basically, you want to treat each one like a new session.  1 repeaters becoming 2 is one session.  2 repeaters becoming 3 is another session.  Etc, etc. 

From my testing, there is usually 1 number that repeats at least 9 times before all other numbers have hit. 

Now, look at the relationships between the 8 sessions.  Are there opportunities in there to profit?  I would say yes, with right money management.  This can win to the unequal distribution of numbers....one will eventually outshine the rest.  The only way it could lose is if everything falls equally each and every time (all 2s fall before 3s, then all 3s fall before 4s, etc)...this will not happen

[nottophammer]

CoderJoe
Mortagon untill he stopped posting


Basic but it's same on Random. org



Only 100 games but it'll still be same at 200 games, 15 point something for spins 11-40 and 60 spins 29 to 30 non-hit.
Now if this 30 spins gives 15 non-hit then there's  15 repeats coming, coming when is the ?, but each non-hit has its own average to hit plus a known maximum over time.

Go on then Taotie mock cuming

[cht]

Save you the trouble. It won't work. Everybody knows that.
Just like this test and many many others.
link:s://:.rouletteforum.cc/index.php?topic=20411.msg204296#msg204296

Just got back, played 3 quick games.

You either get it or you don't.

You play hotties shown by the marque you surely lose.

Some unknown guy played those hotties made money now on the way home.
I know it sounds insane.

[falkor2k15]

hi falkor, what haven't I understood and what does this have to do with the test I suggested? Your post is confusing.

The point of the test is to find out whether playing repeaters is better than betting randomly. You can just follow the logic of independent trials and say it's a waste of time or you can believe Turbo who also makes apparently persuasive arguments, or you can choose not to believe anyone and test for yourself (preferred). But you have to test properly using the correct methodology otherwise the results can be misleading.

link:s://:.dummies.com/education/science/designing-experiments-using-the-scientific-method/

You say you've done 100's of tests but where are the results and how did you do them?

If you understand what repeaters are then you know they're random. Of course, like you, I never knew what to believe and followed gurus. Nobody ever explained repeats to me the way I came to understand them and have described to you for your benefit. After hundreds of tests the results were anything but misleading - they were always break even. Each test I run over 1-2 million spins resulting in break even and no edge. And we are talking a whole repertoire of tests carried out almost every day for 3 years. If you can suggest one I can show you the results. The most elaborate test I ever did was suggested by MoneyT based on 4 streams, including positions, bet alternatively and including a hedge bet.  Of course it broke even just like the other tests I carried out - 99% of them were based on uniques and repeats and multiple streams. I pushed Notto for instructions on his method, but he will not reveal otherwise I could show that to be break even too.  However, nobody wants their false reality to be shattered; they prefer to live in a dreamworld alongside religion and the like.

[Bigbroben]

One thing I found to be worth adding to any system when playing straight-ups ( could apply to any other inside bet  perhaps), is to remove the oldest bet on a win.
I think Vaddi also had mentioned something in this sense. 

Ex:
Nrs: 15,29,30,1,0,8,18,0 (new nr).
Remove bet on 15.



It will make sure the nrs to play with do not climb as quickly as it would.  Advantage when the repeating gap is short: greater profit.  If the gap is wide: would have lost anyway, so smaller losses.
Many times the nr that hits was the one that would have been removed next, so you then conclude the betted nrs were just enough.
On a regular trot game, you might end up at spin 37 and still only be at 10-12 nrs alive.  On a good game, 8nrs alive and 3 or 4-peaters still hitting.
Not a HG, but effective most of the times, especially to catch hot nrs and to remove dead ones. 
I find it is also good to decide when to remove a previously hot nr.  If it is the next in line to be removed, then be it.

Take it as an extension to a system.