Comparative Strategy Analysis

[Mad Professor]

RANDOM outcomes determine the baseline against which any betting-strategy is first compared.

~However, if you run ANY, and I do mean A.N.Y. betting-strategy against enough random outcomes; it will come out in the negative with a LOSS. Period. Full stop.

If you don't believe that; then there's really no hope for you in a math-literate world. Go feed the unicorns.

~The idea behind dice-influencing is to create, through de-randomization, a situation where some outcomes will appear more than randomly expected...and some outcomes will appear less than randomly expected.

~If that difference-from-random is durable enough, in terms of manifesting that difference over a reasonably large sampling; then that offers up a profitably-exploitable situation.

~To profitably exploit that difference-from-random situation, a betting-strategy has to be crafted in such a way so as to create an optimal return-on-investment, while at the same time avoiding risk-of-ruin.

That is the cornerstone of advantage-play dice-influencing.

~Various software, such as BoneTracker w/DiceTool & EdgeCalculator has been developed to first DETERMINE edge and then IDENTIFY a given shooter's pattern-of-influence.

~The DiceTool/EdgeCalculator findings can then be used to craft a betting-strategy which optimizes that shooter's pattern-of-influence.

~Once crafted, that betting-strategy can be scripted into WinCraps, where it can be run against that player's historical roll-stats, or against an intentionally-skewed-to-SRR-rate Mersenne Twister RNG within WinCraps.

~War-gaming various betting-strategies through WinCraps can be quite enlightening. In fact, in-casino gathered roll-stats can be copy and pasted directly into it so that countless strategy variations and permutations can be hyper-tested, even against 'reshuffled' outcomes where the SRR-rate and skew-rate for certain numbers remain constant, but where the outcomes themselves are shuffled as to actual occurrence.

Ahigh, if you are still back in the 'trying-to-beat-random' stage in terms of crafting a betting-strategy that will beat random outcomes, as opposed to dice-influenced ones; then you have many frustrating days ahead of you.

On the other hand, if you believe that outcomes can be influenced, and you are looking to optimize their profitable-exploitation by first determining your pattern-of-influence and then crafting a betting-strategy for that; then you are on a righteous path.

In either case, I enjoy your input.


MP

[Ahigh]

My suggestion was only to make the point that for random data, the best bet is the one that loses the least amount of money per roll, and that's it (dp no odds) for a given table.

I think it's awesome to get a cost per roll for any strategy for a long-term view on how good the strategy is outside the domain of roll bias. You do need a simulator to do this for any sort of bets that change modes depending on what is going on. But I also absolutely believe that there are no positive expectation strategies given random data. The game is built that way, and I think we all know that (or we should anyway).

The strategy I gave has a maximum edge per bet at 1.67 with a 5% commission on the win. Each bet lasts until one of 9 outcomes appears (seven plus four, or seven plus ten). So it has a 9/36 of resolving meaning the bets last an average of 4 rolls. That's a 0.4175% edge per roll, and the strategy costs $0.21 per roll @ $50, $0.42 per roll @ $100, $1.04 per roll @ $250, $1.46 per roll @ $300 .. maxxing out at $4.20 per roll when you have $500 each on the four and ten. But the expectations for this simple strategy don't require a simulator to calculate those numbers, it's just pretty easy to do without a calculator.

All strategies effectively are going to go down with enough samples of random data over a long enough period of time. And I am not saying this because you guys don't already know it. I don't know MP's strategy, but I would expect it would have a lower cost than mine on average because I'm leaving money up there to try to win big. A conservative strategy that makes a low edge high likelihood of winning and regresses is a good low-edge strategy for the long run. I haven't even looked into his strategy, but between Super Rick, Bearish, and now this thread, I have seen that a lot of people are talking about it.

The only thing I can imagine that would make it more prone to losing would be:

1) The 5 and 9 bets
2) 4 and 10 bets for 6.66%, or 3.33% edges at $5 and $10.

I don't know if he even has #2, but the four and ten for $5 each loses as much as it does for $25 each around here.

[Buy The Four]

MP

You say that no one can win over the long-run on a random game? Is this because of the "house-edge?" I have a feeling that if casinos eliminated the house edge on craps and payed true odds for every bet on the layout, MANY players would still lose in the long-run because they are losing more bets than they are winning.

[Buy The Four]

I am game to play this strategy but in order for Method players to have a chance to succeed the rolls have to be taken from a continuous game....not spaced out like in the Testing Testing thread over on Diceinstitute. The Method's premise is that it can win over the long run on random outcomes. Method players play the table and not the shooter.

This thread is going to get very interesting. Especially since Heavy pointed out what most DI's won't admit (many of the DI's playing today do not have an advantage and are essentially random rollers) Furthermore, the notion that money can't be made on random rollers is controversial. I have The Method and have seen it do incredible things and have seen it belly up. This is going to get good!

[Mad Professor]

BuyTheFour, I agree with you wholeheartedly. Even given a non-House-Edge 50/50 game; most players would still lose...and many would lose badly. That's the nature of the gambler as opposed to the advantage-play dice-influencer.

Ahigh, I've provided plenty of links to the MP $204-Across two-hit Regression w/Passive-Aggressive Pressing betting-strategy (designed specifically for DI group-shoot situations); as well as links to the ~20,000 in-casino-collected dice-influenced rolls that are contained in the Testing, Testing...Formulized D-I Betting-Strategies thread.

Regardless of how many times I post these links, I obviously can't force you to read them if you choose not to.


MP

[Ahigh]

MP, I'm not ignoring your information on purpose, I just have limited time with all the other things I'm doing.

I will read and understand MP's betting strategy and code it up in the software I wrote that output the above image I included. I'm putting strategies into my software anyway, so the timing is good for me.

Basically my software is all automated and custom built. I have seen the software that you guys are talking about, but my goal is to have everything automated and tied together so that I can view the results interactively as I toss the dice. For me this is a hobby and not a way to quit my day job. So I have lower expectations for the results.

However, I don't believe that I need any of the other software and I enjoy the challenge of doing all of this stuff myself.

I also want to point out that I am still under fire from Wizard of Vegas as they I guess in the camp that absolutely despises anyone who believes that bias for a particular throw is worth any thought at all.

My position isn't to have a believe first, but rather to accept the possibility and then to arrive at a belief of my own based on my own personal work. When I came out to say that I intend to prove things one way or another on the dice institute forum, many people were very upset about this, and I think that was what led to RC getting all worked up and calling me names.

I still don't have enough roll data to know if my rolls are biased, but you're looking at my most recent 300 rolls recorded in the last month. That's still only an average of ten rolls a day because I've been working on my software more than recording rolls.

But now that my software is going, I can record rolls like crazy with the free time and things will start to get pretty obvious from the visualizations of all the stuff that the software is spitting out automatically as you are entering the data.

I will be making a you tube video soon to show how all of my stuff works, and I will share all the details. The software is not available for sale and I will not be selling any classes despite what anyone at Wizard of Vegas thinks. I am doing this for "shits and giggles" sort of like the die balance, except this is MUCH MUCH MUCH more expensive for me to do with a zero expected return on my time. Just for fun.

Of course I could win some money; but that's not the goal.

[Blackcloud]

UNHH!!Some body is more than a few bricks short

[Ahigh]

UNHH!!Some body is more than a few bricks short

Terse are we?

[heavy]

Of course I could win some money; but that's not the goal.

Whatever floats your boat. For some of us - that is the only goal.

Just out of curiosity - what is your goal? Inquiring minds and all that . . .

[Ahigh]

Here's the MP204 strategy on my last 300 recorded rolls compared to 32 across.

[Maddog]

Some post that were not pertinent to the previously outlined strategies were moved to a dedicated thread of their own.

These can be found here: ahigh - Information

[Ahigh]

Here's my lame low edge strategies on my own roll data.

Looks like MP's system loves my rolls.

I'll be coding up pass and dont with odds and lay odds here for comparisons.

Half of what I'm looking for personally with all this work is just to be familiar with the characteristic shapes of various strategies.



Sorry for going off on my life story to explain about my goals not being aligned with making tons of money on the game of craps and I appreciate the comments and splitting that off. If anybody else wants me to code up a strategy, let me know.

Also, my format for roll data is like this:

Code: Select all

ahigh/Volumes/Aaron/GoodShooter/perl/rdata% cat aaron_sr1_2012-Aug13-9:31pm.txt 
# Rolls
a 33`52`51`32`51`66`32`64`43`54`51`21`64`51`51`53`32`44`63`51`41`63`53`66`65`42`41`44`42`42
a 52`21`31`62`21`44`32`66`32`31`52`52`44`44`33`65`53`22`43`22`51`65`54`53`31`51`53`31`51`43
a 65`31`62`54`53`52`54`53`51`21`42`64`63`55`65`53`22`55`41`33`54`63`51`64`32`61`21`42`43`51
a 63`21`53`63`54`63`53`55`53`41

But I can try to work with other formats if somebody wants to give me more data.

I am absolutely having fun with this. So thanks for everyone's interest.

[Ahigh]

I hope she doesn't kill me, but here's my girl's rolls. Now keep in mind that she was rolling a whole shit-ton of sevens, and not exactly excited about it. But the interesting thing is that MP's betting strategy vs mine looks different on her roll. She does an ice-tong throw, and she doesn't play hardly at all. More of what you might expect from someone who is not practiced at throwing.

But I have a lot more of her samples from August since she likes to throw the dice.

I have tried to get her more interested in playing craps, but she is so busy with the kids and stuff, it's enough to not get her coming down on me for being so hard core. Anyway, plenty of data for her, just really heavy on the sevens she is!



Same thing with a $2000 starting bankroll:



Anyway, you can see from the graph she had a few 4's and 10's up front, and then more sevens came and rounded things out.

11: ------------------------------------------ 2
21: ---------------------------- 3
22: ---------------------------------- 4
31: --------------------------- 4
32: ------------------------------- 5
41: ---------------------------------- 5
33: ------------------------------------ 6
42: ---------------------------------- 6
51: -------------------------------------- 6
61: ----------------------------------- 7
52: ------------------------------------- 7
43: -------------------------------------- 7
53: -------------------------- 8
44: ---------------------------------------------------- 8
62: --------------------------- 8
63: ----------------------------- 9
54: ----------------------------- 9
55: ---------------------------- 10
64: -------------------------- 10
65: ----------------------- 11
66: ------------------------------------ 12

2) 21 3.65% - 2.78% = 0.87% (+5.00)----------- 2
3) 28 4.86% - 5.56% = -0.69% (-4.00)--------------- 3
4) 44 7.64% - 8.33% = -0.69% (-4.00)----------------------- 4
5) 65 11.28% - 11.11% = 0.17% (+1.00)---------------------------------- 5
6) 90 15.62% - 13.89% = 1.74% (+10.00)----------------------------------------------- 6
7) 110 19.10% - 16.67% = 2.43% (+14.00)---------------------------------------------------------- 7
8) 79 13.72% - 13.89% = -0.17% (-1.00)------------------------------------------ 8
9) 58 10.07% - 11.11% = -1.04% (-6.00)------------------------------- 9
10) 40 6.94% - 8.33% = -1.39% (-8.00)---------------------10
11) 23 3.99% - 5.56% = -1.56% (-9.00)------------11
12) 18 3.12% - 2.78% = 0.35% (+2.00)----------12

Total sevens 110 - Seven outs 81 (73.64%) - Seven winners 29 (26.36%)
Pairs 114 19.79% - 16.67% = 3.12% (+18.00 rolls)
Hards 75 13.02% - 11.11% = 1.91% (+11.00 rolls)
HiLos 39 6.77% - 5.56% = 1.22% (+7.00 rolls)
H2 21/6 ( 3.65% - 2.78% = +5.00%)
H4 17/7 ( 2.95% - 2.78% = +1.00%)
H6 18/5 ( 3.12% - 2.78% = +2.00%)
H8 26/8 ( 4.51% - 2.78% = +10.00%)
H10 14/5 ( 2.43% - 2.78% = -2.00%)
H12 18/5 ( 3.12% - 2.78% = +2.00%)

------------

[heavy]

AHigh - knowing the guys on the forum as well as I do - I think they would appreciate seeing $$$ plugged into your analysis. On fours and tens, could you you plug in a series starting with a $25 four and ten - running the following series: $10 - $25 - $50 - $100 - $250 - $500

Yeah, I know. That $10 place bet doesn't make sense from a mathematical point of view - but again, knowing the guys who haunt the form - most of them are $5 - $10 bettors and real world - that's where they'd start out on the four and ten. Your posts will interest them more if they can easily apply the info to their own situations.

[Ahigh]

This is starting to feel like work! But you know, neither my rolls or kelly's roll are really good on the 4 and 10 according to these results.

I added descriptive paramaters to the bet strategies that get parsed by the algorithm. no4 means buy the 10 only. s10 means start with 10 units. l100 means stop pressing at 100 (limit 100).

I also put some time towards come and dc with odds and lay odds, but I haven't made those strategies yet. Even just seeing the plot of the volatility for these strategies is something I've been interested in for a while. I already had an image of these charts in my mind before I started, but this has been a fun verification process for how these finance style charts look.

Also, VERY IMPORTANT, this is not a press all the way strategy.

Here is the press schedule for this chart: That's why the slopes down are not as harsh and the spikes up are not as sharp.

@buy_press_schedule = ( 10, 15, 20, 25, 30, 35, 40, 45, 50 )

[DeadCat]

MP

You say that no one can win over the long-run on a random game? Is this because of the "house-edge?" I have a feeling that if casinos eliminated the house edge on craps and payed true odds for every bet on the layout, MANY players would still lose in the long-run because they are losing more bets than they are winning.

Not MP, but I'll take a crack at answering this;

I agree. Even if casinos offered even money bets on coin flips I would not be at all surprised to see more losers than winners at the end of the day.
There is no trick by the casinos to that. I just think that a very large percentage of gamblers will play until they run out of available money, a problem (limited bankrolls) casinos don't have.

People can win over the long run in a negative expectation game, but it is just remarkable luck and probably some self-discipline (quitting when ahead) but the more they play the less likely that will be. Win or lose any bet and the casino takes some money from your bank roll, either when you lose it or by under-paying you on the winners.

Lyle Stuart had a bit to say about gamblers being more comfortable losing than winning. He likened their losing to a cathartic experience, cleansing them of their sins by losing their money. Why,he asked were planes full of returning junketeers chuckling over having "taken a bath...getting cleaned out...making their contribution." (?)

[Ahigh]

I had a bug in my buy 4/10 that was still pressing up to 500 not 100.

This is using random data and shows some better results for the buy 4/10 starting at 10 and not pressing beyond 100.

This has a press schedule as follows:

@buy_press_schedule = ( 5, 10, 25, 75, 225, 675, 1000 );

[heavy]

Okay, let's assume you're playing at the Santa Ana Star in Albuquerque - free buy and lay bets on the four and ten - zero house edge on the Field - and you buy in for $500 and bet nothing but those bets - say $18 six and eight and $20 in the Field. I agree that a large majority of players are going to bust out . . . OVER THE SHORT RUN . . . due to the inherent volatility of the game . . . even if you are playing with a zero house edge. Over the short run you are playing with a limited bankroll and the casino simply has more bullets than you do. Then the losing streak comes - and it will come - they will kick your dick in the dirt and stomp it. When your winning streak comes they'll smile and hand you chips - knowing that they'll get them back.

With that said . . . over the long run it's a zero sum game.

[Ahigh]

I can redo the above chart for many many more random outcomes and starting with a bigger bankroll. It might also be interesting to show the volatility of passline with no odds just to see the relative losses from the familiar $5 line @ $0.02 per roll on the long term basis.

But if you make the analogy of taking these bets to investing in the stock market, you want to catch things when a bunch of these numbers are "due" regardless of how you define "due" .. IE hunch, statistics, randomness, whatever. That is the elusive trick. Watch and then jump in and hope you have the bankroll to catch an updraft then get off when you get scared shitless to leave that much on the table.

And I want to point out again: do not underestimate the fun factor of this. You have to be ready to lose it all, but keep track of the edge and think about people who leave $25 up on a hardway. This is a less scary thing to do than to leave $20 on a hardway, or it should be in the long run. That's the argument. You are just skipping the first parlay from $2 on a hard 6 or hard 8. Those who say that's not your money are wrong because I take $20 hits down off the hardways all the time!

But if it makes you feel better, only try to hitch this ride by hitting a hard 6 or a hard 8 for $3, take five for you, and $25 for the four or for the ten. Paying a $3 vig to get $30 instead of $33 is still a high edge, but flat fee is a couple of bucks then you don't have to guess when to gun for a parlay on the four or ten, you can let the dice decide and just follow the program.

Let me suggest:

$3 hard 6/8 -> hit -> $5 to bankroll + $25 to buy the 4/10 -> press all the way, and you have the vig on the rail from the 5 you can put aside.

Always follow a hard 6 to the buy on the four and the hard 8 to the buy on the ten so you don't have to pick.

Let me name this strategy "A High Ride" if you will allow me to name it.

Since I haven't hit it, it would be great to get the story from the first person to do it.

Combined edge should still be relatively low since you take the big bucks to the lower edge bet to lower the total vig on the pay.

[heavy]

By the way, the graph avatar AHigh is using is completely appropriate. Hell, I was thinking about changing his username to Charts&GraphsHigh. Interesting stuff, really. With that said, I'll probably still preach that players who want to play the hardways limit themselves to around $20 a session and not bet a hardway until the table starts giving up hardways. Then parlay the first hit and press up from there. I really like taking $1 to $1000 and down on a hard six or eight. Okay, okay. Press it to $500 and pay me $500. I'd be pissed off big time if I missed a $4500 payout.