Craps Systems That Don't Work

The other night I watched some craps players betting heavily on the field and getting a lucky streak. One player went from $300 in his rail to more than $5k. And that's when he started yelling "bet the field! The experts say it's a bad bet but they're wrong!"

And about a half-hour later he was back to less than $1,000 which is when he left.

This reminded me about some other failed strategies:

ONLY BET RED ON THE COMEOUT

I watched a player at Caesars buy in with $100k and he used this strategy. He bet a 3 way 7 on each shooter's come out roll. And for each shooter who failed to roll a winner 7, he'd double the bet on the next shooter.

So for the first shooter it was 3 X $25. For the second shooter it was 3 X $50... and so forth.

This was his only bet. The problem? No one threw a winner 7 and the guy lost all $100k.

BET $100 ON 12 FOR 30 ROLLS IN A ROW

The strategy was simple. On 30 rolls of the dice bet a black chip on boxcars. The player figured he'd hit at least one. No one threw a 12. He lost $3,000.

After he lost $3,000 several shooters threw midnight but there was no money left to bet.

BET ON SIX

I'm not sure how much money this player had but his strategy was simple. Bet $100 on the hard 6 and $120 on the place 6 for each shooter.

The problem: no shooter rolled any 6 before his bankroll was gone.

I probably shouldn't be wasting my time on here, but what betting system in a negative sum game possibly has a winning strategy?

Maybe your talking advantage play or cheating?

You're correct. There is no betting system that works in the negative expectation game of craps.

But the schemes are interesting.

Quote: AlanMendelson

You're correct. There is no betting system that works in the negative expectation game of craps.

But the schemes are interesting.

You know you bring up a good point. I've seen so many people obsessed with betting systems yet its the one thing that I still struggle today with understanding peoples motivations towards them. I feel this information could be of value to me as a game designer but I am also just curious as to why? Is it because people want to think they can disprove math as if they are smarter or know better, or is it negative reinforcement that makes people want to commit to these betting systems? I have often though that people prefer the martingale because most of the time they will walk away a winner or even, but that one time they are going to get a bad run and loss everything, so they are a martingale player because they haven't lost it all ye but its just a matter or time. But all those are just my assumptions and I would love your input since its probably difficult to find these religious betting system players to interview with.

I dont follow any betting scheme myself, but I've heard them all:

Aces follow aces, midnight follows midnight, second roll yo, second roll horn... etc.

You could easily market some craps side bets using these. There's probably a market for them.

I can't get any traction on field bets unless I'm gonna DI some field numbers myself. I always have to make up my losses from when other people were shooting. I'd want to run a progression on the Buy 4 & 10 too.

I see why these ‘systems’ failed.

Quote: AlanMendelson

...The problem? No one threw a ...
-snip-
... No one threw a ...
-snip-
... The problem: no shooter rolled any ...

..................


Jean Le Rond D'alembert was a brilliant French mathematician circa 1759 who erred in writing that the probability of coin landing on heads increased for every time that it came up tails previously - because of this error his reputation suffered - although he is still regarded as having achieved greatness in his field

a betting system - the Alembert - a version of the martingale - is named after him

if a brilliant mathematician could falsely believe this - it's easy to see how more average people could become susceptible to the lure of betting systems



*

If there's no winning streak, there's no winning streak.

"To beat a random game you have to be good at being random." -Chapz

Quote: AlanMendelson

You're correct. There is no betting system that works in the negative expectation game of craps.

But the schemes are interesting.

About 7 or so years ago, after reading thru yet another thread on this forum about winning systems for Craps, I decided to apply some of the skills acquired from my day job (in this case AI and Machine Learning) to the problem of coming up with the optimal system for Craps. As the saying goes


I decided to use one of my favorite ML techniques: a Genetic Algorithm (GA). For those not familiar these, it's basically a "survival of the fittest" approach. You create a population of a thousand or so bots that are all trying to solve the same problem. The trick is that they each have a slightly different approach to the problem where that approach is controlled by the bot's "chromosome". Each bot gets graded on how well they did. You then kill off the 90% or so that scored lowest and let the remaining 10% "breed" a new generation of bots. Keep repeating with each generation getting better and better until you're satisfied with the results. Very Darwinian.

The nice thing about using a GA is that pretty much any behavior can be included in a bot's DNA. Pressing bets, hit-n-run, rat-holing..... all covered. I tried to include everything forum members had advocated. About the only things not included was DI and staying well hydrated. The chromosomes are just strings of 0s and 1s so you just use a RNG to create a range of random behaviors for the 1st generation of bots.

Once I had my bots, I hooked them up to a simulator. Each generation had 10000 bots playing 100 sessions. A bot started a session with $500 and played until

• it went broke,
• it decided to quit, or
• the session ended after ~150 passes

Each bot's score was their average bankroll at the end of each session. Variance was used to break ties if two bots had the same score (i.e., lower variance was considered better). I set the system up so it would run for a maximum of 100 generations or until the best score for each generation was no longer improving (which ever came first). Since running a GA can take a while, I hit the RUN button and then went off to lunch.

When I got back from lunch I had my "best of breed" bot. All I had to do is examine it's digital chromosome to find out what it's strategy was for playing Craps. It turned out that the optimal strategy it used was

So there you have it. No point in any further argument as advanced technology has given us the definitive answer as to what the best strategy is for playing Craps.

Sorry you wasted your time. D’Alembert works for me

Just lost a total of 20 PL odds points in 199 rolls on WinCraps.

Quote: ChumpChange

Just lost a total of 20 PL odds points in 199 rolls on WinCraps.

It happens.

There's a section in 'Midgley's 7,500 Dice Rolls', where the PL loses 14 times in a row, over about 100 actual dice rolls.

I got a new strategy and WinCraps let me win $15K on it in 300 rolls; that'd be $1500 at a $15 table. It depends on winning 7 or more PL in a row, and 7 or more odds points in a row.

Quote: TumblingBones

... snip ....
When I got back from lunch I had my "best of breed" bot. All I had to do is examine it's digital chromosome to find out what it's strategy was for playing Craps. It turned out that the optimal strategy it used was

Optimal Strategy

that it never made a single bet!

Playing this PL & Odds progression is one long losing streak until a couple of short winning streaks comes along. I'd better go back to betting on the DP with triple odds and switching sides after 2 odds losses.

Trying a DP with 3X odds and switching to PB 6 & 8 on 4 DP losses or an odds loss. WinCraps keeps sevening-out right after I place my place bets. I need to hit the place bets 4 times to come out ahead for the hand.