The Simulation Fallacy

[Scarface]

Fallacy - faulty reasoning;  misleading or unsound argument.  A failure in reasoning that leaves an argument invalid

We've all seen it here.  A winner will post their charts showing their winnings, and the naysayers come in to dismiss it.  They will say it was luck, or not enough spins.  100 spins is not enough!  500 spins is not enough!  Test with a relevant number of spins!  Your system will fail if simulated for 1 millions spins!

The simulation argument is a fallacy for several reasons.  The most obvious one is that no one will play continuously for 1 million spins.  You can take your profits at 30, 100, or 200 spins.  Our sessions may not all last the same amout of time.  When we choose to end a session is up to the player.

On the Turbo thread, some say that you will never encounter 37 spins in which there is no repeater.  This may be true, but it is not guaranteed.  Some commented that if you tested this for millions of spins, you will at some point see this happen.  While some may not see this in a lifetime, it is still possible.  But I think everyone can agree that it is a very rare event.  But here's the problem with the simulation argument.  A simulation will show that in millions of spins, I will eventually encounter that variance from hell that will wipe out my entire bankroll....but this is not true.  Your simulation ASSUMES I will continue to keep playing into this negative abyss until all my money is wiped out.  However, I bring 20% of my bankroll to the casino.  I know not all sessions are winning sessions.  If I encountered this rare event, the most I lose is 20% of total bankroll.  Your simulation will show far more losses than I actually got.  So, what are the odds that I see something so rare 5 times in a row that it would wipe it all out?  Probably similar to the odds of hitting the lottery 5 times in a row.

I played the stock market awhile, and looking back it really is like roulette in many ways.  When you buy and sell stock you have to pay a broker commision.  Its a small cost to play, which I think is alot like the house edge in roulette.  I did very well in a 1 year span, making several times my initial investment.  I made 100s of trades that year, and stayed in the market the whole time.  All trades were not always the same amount of money...never put all my eggs in one basket.  Not all trades were the same as far as how long I held the position....some trades may have lasted as short as an hour, while most lasted several days.  Some I even kept for several weeks.  This too is similar to roulette, since many of my roulettte sessions difffered in length of play.  I used careful money management while trading, using stop loss to limit losses. 

So whats stock trading have to do with simulations?  I made several times my investment that year, while the market was in a mild recession.  If I simply bought and held I would have surely lost money.....the market was down nearly 20% ytd.  If you were to look at a year to date stock market chart, you would see the end result as a loss.  This is similar to how people here will look at a simulation of 1 million spins.  There is your fallacy.  The fallacy lies in only looking at a chart`s end result.  The fallacy also lies in assuming a player will get the exact same loss as the simulation shows, using no money management or stop loss, and play until bankroll is wiped out.

[Wally Gator]

It's all a fallacy, right?  What if we added in the possibility that you are playing a biased wheel and win often and big.  One night, someone was tracking you instead of the wheel, then follows you home and robs you.  Can we really put any number of variables in to determine probability from an equation that is completely random?  So, when I hear the million spin argument, it completely misses the waitress who spilled a drink on you just as you were attempting to lay down big on that biased wheel.  Did the drink spill just on you or across the table?  Did they close the table for cleanup?  Change dealers?  It also misses the RNG machine that just shit the bed as you had 25 down on your 1/5/25 prog.  Add in any other random event you like to the mix.  There are millions of possibilities.

People believe what they will.  A few years back I thought I had the HG.  It lasted for over 3 years and I was able to pay 2 of my kids student loans off (over$200K combined).  Still have another one to go.   I was riding high.  It crashed and crashed big.  But, the loans were paid off.  I stopped playing for a few years.  Did I win?  I don't know yet, I'm not dead.  If I never play again, which is highly unlikely, I would say yes.  So, the loss was big, but the win was bigger over the long term.  It's all risk versus reward with some serious amount of greed in there.

I've been on this board for quite a few years and it's more of a bitching session now than a friendly forum where folks openly share their ideas for what they believe is or will be successful.  My hope is that everyone wins in the long term, regardless of what way they play.

I don't post much, but will end with this.  You only need to prove anything to yourself.  It doesn't matter what others think or do and it only matters when you have real skin in the game.  My experience has been that simulations are great, but that's all they are, simulations.  Nothing will prepare you for the real event, as there are events that will occur that you've not encountered before and variables you'll need to take into considerations that can't possibly be addressed through simulation.

[Scarface]

Wally, well said my friend.  Sounds like you had a great run!  There are many variables that simulations cannot account for.  One I think that has me helped many times is intuition.

I'm not saying running simulations are a bad thing when testing.  It's good to know what to expect.  Just saying that only looking at the very end result of 1 million spins is pointless. 

[junscissorhands]

Wally gator, finally some real talk.

[falkor2k15]

The permutation from hell can hit you any time. Doesn't matter how much time or how many spins you delay between sessions - your luck will run out. Therefore, simulating 1 mill spins is perfect to find out if your system is a winner. To be a winner you need edge that is at least the % of the house advantage; in the long run you would then be guaranteed to be in profit. If you don't have that edge then hiding in-between sessions is not going to help. Either that or you need some way to "recover" and turn a losing session into a winning session before the next permutation from hell comes. However, so far all attempts at avoiding variance or trying to "dodge" it somehow, has failed in my attempts by the hundreds.

[DoctorSudoku]

A few years back I thought I had the HG.  It lasted for over 3 years and I was able to pay 2 of my kids student loans off (over$200K combined). Still have another one to go.   I was riding high.  It crashed and crashed big.  But, the loans were paid off. 

Hi,
I vaguely remember that you had alluded to finding a consistently winning betting method on either this or another forum a few years ago. I am assuming that this is the method that you are referring to above.

Can you reveal if it was at baccarat or at roulette?

If it is too personal a question, then of course you don't have to reveal the answer. No problem.

[Scarface]

The permutation from hell can hit you any time.

I agree.  No doubt if you simulate long enough you will find a loser eventually.  But if you find what you're looking for, a bet that would last for 1 million spins, would you play it like the simulation? Maybe, you would have to use a progression from 1 unit to thousands!  Would you actually play like this?  Maybe a simulation of 1 billion spins will even show that to be a loser.

My stock market example shows a loss if you look at a year to date chart, but I had gains.  If you add up all my roulette sessions over the years, I'm sure there is a significant enough spins that the math guys would say I should be at a loss...but I'm ahead.  Not claiming to be a millionaire, far from it.  But I'm definitely in the plus, which is pretty awesome considering I was so naive those first few years....chasing losses, huge negative progressions.

I can't point to one single reason why.  There are things out of your control, like a negative variance from hell.  Nothing you can do about this.  But there's also things you can control.  I never bring my entire bankroll to the casino, only a percentage of it.  I will take profits early and quit.  I will change my strategy while playing.  I will accept a loss and move on, but will not risk my whole bankroll trying to catch up.  Discipline is the hardest part of this game

[Wally Gator]

Can you reveal if it was at baccarat or at roulette?

It was baccarat and was only when I attempted to transfer it to ECs in roulette that it failed.  I needed the break when it came, though, so I have no regrets.  If you are wondering if the zero played a factor, interestingly enough, it didn’t show during the sequence I played.  And, yes, that particular day I was overconfident, tired, hungry and should not have been playing.  Why don’t I just go back to playing it at bac? Probably will, but need to get in a better psychological space as I’m back at red and green chips, no orange and purple.  It was a very long few years and became nearly a J-O-B.  Not what I had originally intended and gives pause for continuing.

[Bigbroben]

The fallacy also lies in assuming a player will get the exact same loss as the simulation shows, using no money management or stop loss, and play until bankroll is wiped out.

It is very possible to integrate stop-losses and MM in simulations.  All the ones I do are tested with different stop-losses, goals, flat bets, progressions.

Say a curve showing the results of 10000 games with 1000 spins each.  Say this represents a millions spins from some wheel some guy named John Smith was considering jumping into.
Sure you can say: He'll play this way and might end up on a winning ride for months or years, or also the opposite.  He never knows where in that curve he'll land.
Would he visit the casino for years (or click for days), observing the spins, playing virtual, until he feels the ''bottom'' is reached?

Running simulations will simply help calculate the risks and find which ways are better than others, which stop-loss or goal brings the best benefits. 

Or course a surgeon will also run an afwul lot of ''simulations'' at school before to perform a real operation on anybody.
No matter how many sims were run, the real first time will always be a different experience than on dummies.

[Joe]

Hi Scarface,

I don't find your arguments against simulations very convincing. For a start, the objection that no one plays for a million spins seems irrelevant. Whether anyone actually does or not is beside the point because the purpose of a simulation (maybe not the only one, but one of the main ones) is to find out what happens in the long run. And supposing a system did actually make a good profit after a million spins; would anyone object that it was unrealistic? Probably not but then it would seem like a case of "heads I win tails you lose".  ;-).

Secondly, one person is very unlikely to play a million spins but what about 1000 people playing 1000 spins each? That would add up to a million spins and would be only a bit more difficult to simulate than one person playing 1 million spins and should give the same result, shouldn't it?

I do agree with you that things like intuition can't be simulated, but most systems are mechanical. If a system is mechanical it means that you always know what decision to make or action to take after every spin or bet, and the rules of the system have all been researched and worked out beforehand based on statistics (like Turbo's repeater system). But if you don't know how you're going to react (there are no rules) then aren't you just guessing and hoping for the best? If so, how can this give you an edge?

Of course a system doesn't have to be set in stone with completely rigid rules and no flexibility, but that doesn't mean it can't be simulated. You can simulate a set of guidelines using fuzzy logic, or incorporate a range of options or decisions to make based on certain criteria. Within this basic framework you're basically guessing or betting randomly. The coding would be more complex but it can certainly be done.

[Scarface]

Hi Joe,

My intentions with this post was not to totally discredit simulations altogether.  Some people, especially those new to the game, definitely need to test.  For example, the guy that claims you will never see more than 10 reds in a row, and uses a martingale progression.  A simple simulation of 200 spins will show him how common even bets go missing for 10 or more spins.  So yes, simulations are useful.

I'm more against the argument that some make against players winnings, using the simulation argument.  The argument basically goes like this:  if I run enough spins for enough trials your system will fail.  But simulation doesn't factor in alot of variables.  It can test the system, and rules only but leaves the overall strategy out of the equation.  It leaves the human factor out of the equation.

If using simulations to run millions of spins just to show that a certain system will fail if tested long enough is in my opinion a waste of time.  If you look, you will find it.  Look at bias wheel players.  They use chi square calculations to determine if a wheel has enough bias to give them an edge.  Of course, they can't realistically track over a million spins because this would take years!  So, lets say  an AP player tracked 10000 spins to collect their data.  If I ran enough simulations, I could surely find a 10000 spin simulation that would give them false readings....I could say, hey look my simulation shows you would have played this wheel and lost all your money.  If you run enough trials, you can find anything.  If looking for a loss you will find it.

[Tinsoldiers]

It can test the system, and rules only but leaves the overall strategy out of the equation.  It leaves the human factor out of the equation.

Very nice points.  The quote is what I hate to hear normally. Any strategy can be simulated as long as you don’t play what comes to your mind and the strategy is set out clearly at the start. Of course you can’t simulate conditions where you don’t like the dealer or the pit boss and moving on.

People typically attack with a million spin argument only when someone say they have a holy grail and it never loses. And trust me if someone had such a system then it is very easy to simulate. Not difficult at all.  You cited your investment in stocks, remember you chose to play the stocks you want and you had a winning strategy or luck was on your side.  Same goes for roulette, on the hypothetical scenario that you have the winning strategy or the likely scenario that you are riding your luck house edge will not catch you.

On the topic of simulation though I agree with you. You don’t need simulation of a million spins to prove your strategy works. As long as you can prove that for a finite amount of spins you will always end in the positive, it doesn’t matter how much spins you simulate.  But if you can’t prove above then it doesn’t matter even if you simulate a million spins and won, the first time you start to play for it the downturn could hit you on the face. 

[The General]

Fallacy - faulty reasoning;  misleading or unsound argument.  A failure in reasoning that leaves an argument invalid

We've all seen it here.  A winner will post their charts showing their winnings, and the naysayers come in to dismiss it.  They will say it was luck, or not enough spins.  100 spins is not enough!  500 spins is not enough!  Test with a relevant number of spins!  Your system will fail if simulated for 1 millions spins!

The simulation argument is a fallacy for several reasons.  The most obvious one is that no one will play continuously for 1 million spins.  You can take your profits at 30, 100, or 200 spins.  Our sessions may not all last the same amout of time.  When we choose to end a session is up to the player.

Why don't you test your system for only one spin? 

[nottophammer]

Why don't you test your system for only one spin? 

[Bigbroben]

Is that the guy who was asked if he had found Waldo?