1. Welcome to the #1 Gambling Community with the best minds across the entire gambling spectrum. REGISTER NOW!
  2. Have a gambling question?

    Post it here and our gambling experts will answer it!
    Dismiss Notice
  3. Discussions in this section are assumed to be EV- as they are outside of the Advantage Play section. For EV+ discussions, please visit the Advantage Play section.
    Dismiss Notice

Baccarat In depth discussion of trends in Baccarat

Discussion in 'Baccarat Forum' started by Craps, Oct 4, 2018.

  1. Nathan Detroit

    Nathan Detroit Well-Known Member Founding Member

    Joined:
    Dec 25, 2014
    Likes:
    2,080
    Craps ,

    A very conservative Baccarat system . Thanks for having taken the time posting it .

    Well explained by the original creator of that system .



    ND
     
  2. kfmfe04

    kfmfe04 New Member

    Joined:
    Oct 10, 2018
    Likes:
    4
    Location:
    California
    My understanding of labouchère must be off or my understanding of rule 14. is off, because I can't reconcile rule 13. with 14.

    Under a standard cancellation system, I start with [1 1 1 1 1 1 1 1 1 1].

    If I lose when betting 2, it becomes [1 1 1 1 1 1 1 1 1 1 2].
    If I then win when betting 3, it becomes [1 1 1 1 1 1 1 1 1].

    Under 14., if losses do not affect the cancellation chart,
    If I lose when betting 2, it stays [1 1 1 1 1 1 1 1 1 1], so losses can't affect the original.
    If I win, I can't remove anything because the ends are 1's.

    Am I misunderstanding 14. or maybe I'm misunderstanding flat (which I take to be 1 unit)? Maybe you can clarify with a few examples of what you mean in 14.

    This is a rather messy system (14 rules is kind of crazy), but codeable, if you can elaborate on rule 14.

    Assuming many well-shuffled decks, I expect these rules to change the distribution of returns, but still yield a negative EV, with the EV increasingly negative as the average bet-size increases. Of course, on physical decks that have not been shuffled well or have been manipulated (eg a tampered ShuffleMaster), I would expect that you would capture some patterns with this kind of system (possibly yielding a positive EV).

    - Ken
     
    Last edited: Oct 10, 2018
  3. Craps

    Craps Well-Known Member

    Joined:
    Jul 19, 2018
    Likes:
    251
    Location:
    Las Vegas
     
  4. kfmfe04

    kfmfe04 New Member

    Joined:
    Oct 10, 2018
    Likes:
    4
    Location:
    California
    ok - it's getting late so I can't turn what you said into code yet (I'll try again tomorrow).

    for now, I'm trying to get one sample for each of the rules you've listed (there are actually more than 14, if you break them out). I've enumerated them from A-P and attempted to find cases of each (below is the output). I am matching for the end of the string (as those are the most recent results).

    In real life, it might be tough to follow these rules to the T - in a simulator, I have to sort them from the most stringent to the least stringent and go down the list until I find one that matches. A Rule 1 only seems to match in the very beginning. Once the string gets longer, one of the other rules will capture the trend/chop.

    I'll continue tomorrow when my mind is clear. gn. (btw, I have a running Monte Carlo simulator for standard progressions already - just expanding the code to fit this insane system you've listed).

    - Ken

    Code:
    PBPBPPPBPBBPBPBBP -> P Rule 12 - P Last (B Next)
    BBPPBPPBPBPPB -> O Rule 12 - B Last (P Next)
    PBPBPPPBPBP -> N Rule 12 - P Last (Wait)
    BBBBBBBPBPB -> M Rule 12 - B Last (Wait)
    BBBPBPPBPPPPBPPBBPBP -> L Rule 11, continue FLAT alternating until chop ends (B Next)
    PBPPPPBPBPBPBPPPBPB -> K Rule 11, continue FLAT alternating until chop ends (P Next)
    PBPBPBPPPPPPPB -> J Rule 9, if Win, SIT OUT next bet
    PPBBBBPPPPPPB -> I Rule 10, continue P streak, following B interruption
    BBPBPPPP -> H Rule 8 continue FLAT, follow P until streak ends
    PBPBBBBB -> G Rule 4 continue Flat, follow B until streak ends
    BBBPBPBPBPPP -> F Rule 7, pause after 3 P''s
    PBPPPBPBPBBB -> E Rule 3, after 3 B''s, expect a P
    BPBPBPBBPPBB -> D Rule 2, follow B
    BBBPBPPBPPPPBPP -> C Rule 6, follow P
    PBPPBPPBBBP -> B Rule 5, follow P
    BPB -> A Rule 1, follow B
     
  5. kfmfe04

    kfmfe04 New Member

    Joined:
    Oct 10, 2018
    Likes:
    4
    Location:
    California
    One other quick question before I go to sleep.

    If none of the rules match (it's unclear yet, whether this could even happen or not), sit out one round until something matches?
     
  6. Craps

    Craps Well-Known Member

    Joined:
    Jul 19, 2018
    Likes:
    251
    Location:
    Las Vegas

    Dear Ken,
    According to the author of this system, you write down your own Lab numbers. He uses 1111111111. Since there is no $1 bet in Baccarat, I personally prefer 111 or 121 or 11111. After you are done crossing out all the numbers in your progression, you are done with that shoe. The flat bet portion of the system are considered side bets and it is used to subsidized the progression if there are non 1's.
    Ex. After a bad run of losses the LAB might look like this 1111123456. So if you win a flat bet, the LAB will be 1111123455. If you lose a flat bet, do not add anything to the LAB. ln realty, after the LAB is done, you might not win 10 units.

    How do you play a a card like this B PPP BB PPPP B PPP BB PPPP B without getting killed?
     
    Last edited: Oct 10, 2018
  7. Craps

    Craps Well-Known Member

    Joined:
    Jul 19, 2018
    Likes:
    251
    Location:
    Las Vegas
    Don't over analyze. Its actually quite straight forward.

    When you see a B do the following:

    1. Look at preceding 2 hands. If it is single B and single P, play the chop.

    2, Look at the P run before the B. If it is 3 x or 4 x and the B before the said P run is single, bet P. If not, rule 1 bet B

    3, Look at the P run before a B. If it is 5 or more, bet P

    4. If the preceding 3 hands is BPB thus giving you BPBB sit out a hand cos this could be a reverse chop.

    That's it. Go through these 4 things before you make a bet after seeing B
     

  8. kfmfe04

    kfmfe04 New Member

    Joined:
    Oct 10, 2018
    Likes:
    4
    Location:
    California
    Hi Craps,

    ok - this is obviously straightforward to you because you've been doing it for a long time. You have to admit, even your list of 4 rules are missing a lot of details from the 14 rules.

    My standard play is just bet on P, every time. This would survive that series. I do not change from betting flat P/flat B unless I can find an edge. For example, I'd change bets if I were counting cards.

    Before I go further, I want to say thank you for posting the details for this system. It's worth studying a system like this to understand its benefits and pitfalls, before we try it out in a casino. We can also look for ways to make improvements (I have some ideas - they will be forthcoming in a later post).

    I have some preliminary results, but I'm still missing the following pieces (I will work on these later today):
    1. Institute the FLAT BET rules (in Rules 8 and 11)
    2. Institute the SIT OUT on WIN rule (in Rule 9)
    The results, so far, are interesting, but there have been no surprises (note: these are PRELIMINARY, I will post final results after I knock off the 2 details above). This is using 10k shoes of 8 decks, a $5k bankroll and $10 units. LAB starts with 5 1's.
    1. There's an impressive WinRate of ~93%
    2. In the 6.7% that you lose, there's about a 1% of a blowout of losing the entire bankroll of $5k (ie 5.7% that you will not lose something, but not everything).
    3. Expected Loss/Shoe is ~$10.82
    4. Expected Wagered/Shoe is $989.23
    5. The MC simulation shows a Luck/Alpha of -$2.24 or -0.04 (noise = no edge found)
    FACT: In summary, there's no alpha gained from doing these crazy patterns, as expected. If you do not care about the action and only care about the final distribution, I'm fairly certain we can find an equivalent system that's much easier to execute (fewer rules, simpler). It's not unlike someone throwing you a bunch of complicated arithmetic operations. If you do the math and find out that it's just +7. You can just do the +7 to get the same result. Note that this conclusion is based on a well-shuffled deck.

    OPINION: Most of these crazy rules are meant to obfuscate or cater to superstitious beliefs, including many Gambler's Fallacies. There are enough of them that for most people, it'd be difficult to follow them to a T. However, from a system seller's marketing, it seems like they are "adding value" by making systems more complex and intricate - the more rules they add, the more than can charge. I've noticed that my opinions are starting to converge to those of WOO (Michael Shackleford). I started off as a skeptic and after testing, I'm beginning to see what he sees.

    Ok. That's enough of a rant for now. I plan to finish out those last 2. rules. Then I'll post some readable results, including charts of distributions. So for anyone who would like to use this Failsafe system, at the very least, they'd know what to expect.

    - Ken
     
    Last edited: Oct 11, 2018
    Lani and Nathan Detroit like this.
  9. Craps

    Craps Well-Known Member

    Joined:
    Jul 19, 2018
    Likes:
    251
    Location:
    Las Vegas
    Ken, looks like you show great interest in this system. The author claimed that he only lost 2x in 70 shoes. He did not reveal his stop loss. I also found out that his "flat bet unit is 2" He would stop at +8 units with 1111111111 with all flat bets win/loss factored in.
    Anyway the adjustment that I would make is to rule #5 i.e. bet B after P. I would sit out and only bet with this exception i.e. if the preceding P run is 1 only.
    What do you think?
     
  10. kfmfe04

    kfmfe04 New Member

    Joined:
    Oct 10, 2018
    Likes:
    4
    Location:
    California
    Well, it's interesting because it's the first commercial Bac system I've tested.

    I'm in the process of making a complete survey of different systems to better understand how they should behave, so much of the engine/code was already written (including progressions, EZBac, etc...). As for real life, if I find that I like the resulting distribution of gains, I would replace this system with something much simpler that produces similar results.

    OPINION: I haven't had time to do the analysis yet, but I suspect that the feature with the biggest effect on the outcome is probably the Labouchere.
    I suspect the only result form the pattern rules is in complicating my code and adding to the confusion (neither one good). The only time I think pattern recognition could potentially produce some alpha is if it happens to positively correlate with some effective card counting (likely by accident). I mean it's easy to spew off rule upon rule upon rule to handle all the crazy corner cases, but it's really a pain to code and it's a pain (for most people) to follow.

    I've knocked off the SIT OUT on WIN for Rule 9. I have to implement the FLAT rules for 8 and 11 now...
     
    Last edited: Oct 11, 2018
  11. kfmfe04

    kfmfe04 New Member

    Joined:
    Oct 10, 2018
    Likes:
    4
    Location:
    California
    Oh, one other note.

    The more I do this, the more I understand why veterans ignore nearly everything except looking for a positive edge.

    Adding more rules will just redistribute the returns. That includes stopLoss and winGain rules - those I've added into my core engine early on. They act as ATTRACTORS in pulling the distribution to bunch up at the stopLoss and winGoal points.

    So the extremes don't get hit, but you end up with a higher frequency of samples bunched up at the stopLoss and winGoal. Percentage wins will also tend to drop (move towards 50%) as losses that hit the stopLoss don't have a chance to random walk back to a positive final number.
     
    Last edited: Oct 11, 2018
  12. kfmfe04

    kfmfe04 New Member

    Joined:
    Oct 10, 2018
    Likes:
    4
    Location:
    California
    That weird FLAT bet overlay is too much of a headache to code up (it's one-off code that'll never be used again, and yet, it messes up the main structure of the code). So I'm leaving it out - the effect will be minimal because those rules don't get hit often and when they do, having that FLAT would only leverage the bet up a little bit.

    The summary from one million shoes is below, using a bankroll of $1000 and units of $10. Labouchere[1 1 1 1 1].

    Notes:
    1. ~93% Percent Positive
    2. Expected Loss per Shoe: -6.69
    3. Most wins are 5-6%, completed within 5-15 Deals
    4. Losses are infrequent at 7%, but about 5% of the time (1 in 20), you lose the entire $1k bankroll. It takes a longer 31 draws to bankrupt.
    Below are screenshots of a batch run of 1mln shoes and a single run where the Labouchere progression completed in 17 deals for a 4% win. At the end of the single run is a dump of the remaining cards in the deck and a final card count.

    So this is a negative progression system (Labouchere), in the style of Martingales, but definitely less aggressive.
    But the distribution is similar - a lot of small wins with a few (1 in 20) losses that wipe you out.

    Years ago, when I first graduated from college and started writing simulators for a German hedge fund to trade futures, I was taught that the more rules a system has, the more closely it fits historical patterns, but the less useful it is for predicting future prices. This is especially true for financial markets because those price series are non-stationary (statistical figures like average, variance, etc... are not constant).

    But seeing a list of 14 rules for BAC strongly suggests to me that someone was looking at results, and kept adding rules on top of rules on top of rules to try to eliminate losses. Doing so will result in systems that are tightly optimized for your historical sample, but will not work against the greater sample space that is out there.

    I'm too tired to look at this any more for now, but I'll try, in the coming days, to run a plain Labouchere system and see how close it gets to these results. Perhaps a plain LAB plus a few simple rules will yield similar results.

    If anyone has questions, please feel free to bring them up.

    - Ken



    Screen Shot 2018-10-10 at 9.20.48 PM.png Screen Shot 2018-10-10 at 9.25.57 PM.png
     
  13. Craps

    Craps Well-Known Member

    Joined:
    Jul 19, 2018
    Likes:
    251
    Location:
    Las Vegas
    Wow! Very impressive breakdown of the system. Playing this kind of template system is not my cup of tea but good to know. I wonder how it would fare with just flat betting.
     
  14. kfmfe04

    kfmfe04 New Member

    Joined:
    Oct 10, 2018
    Likes:
    4
    Location:
    California
    Hi Craps,

    Thanks!

    Yes, that's a great follow-up question. In fact, just plain flat betting yields a somewhat "normal-like" curve with some skew to compensate for the negative EV. The weirdness of the shape is caused by matching all those pattern rules. Note I just used 10k shoes here... ...and I still hold the standard disclaimer: if real life shoes have weird distributions created by bad shuffles, the results will be different.

    When comparing the two, I can conclude two things:
    1. All those pattern matching rules don't do much to help you - if I did a distribution of flat betting without any pattern matching, it would look very much like the FLAT here.
    2. It is, in fact, Labouchere that is pushing the distribution to the ends - that's exactly what negative progression bets are designed to do. You cannot eliminate the negative EV, but you can squeeze the distribution so you have a greater percentage of winners. Also note that only essential difference between negative progressions, over the long run, is how quickly they push you to those ends.

    It is this squeezing of the distribution to yield results which show 93% chance of making money which makes these types of systems so marketable. Should you trade using systems like this in real-life? Well, this is what will happen if you trade a Failsafe Labouchere. You walk in with $1000 and you have a very good chance (~70%) of making $50 in about 14 deals and you have a small ~5% (1 in 20) chance of losing the whole $1000. If you visit Vegas only a few times a year, you might do this for fun.

    Here's the crazy thing - someone experienced with analyzing systems, like Mr. Shackleford, will easily tell you everything listed above (in less detail, but the essentials will be identical), without running ANY tests. Coming into this, I kept an open mind towards all methods and coded up the simulations anyways. After many runs, I can safely say I concur with WOO.

    This is what the scientific method about - don't take anyone's word for anything - go test it yourself and see if they match what people claim.

    - Ken
    (Graph comments: Flat is centered around $991 or a $9 expected loss per shoe. Labouchere has squeezed the distribution to 2 peaks: a mini 5% one at $0 where you've lost all your money, and a sharp peek centered around $1055.50, for a 5% gain)

    Screen Shot 2018-10-11 at 10.58.21 AM.png Screen Shot 2018-10-11 at 10.58.05 AM.png
     

  15. kfmfe04

    kfmfe04 New Member

    Joined:
    Oct 10, 2018
    Likes:
    4
    Location:
    California
    Here's an improved density plot with a better, non-default, x-axis scale.

    Screen Shot 2018-10-11 at 11.44.21 AM.png
     
  16. Lani

    Lani New Member

    Joined:
    Oct 15, 2018
    Likes:
    5
    Location:
    CA, USA
    Sure people can try to play with the trend. But according to my system-testing experience, there is no advantage over trend prediction, i.e. the odd is no better than prediction of single hand.
     
  17. Craps

    Craps Well-Known Member

    Joined:
    Jul 19, 2018
    Likes:
    251
    Location:
    Las Vegas
    Lani,
    Every Baccarat player is a trend player. Trends comes in all forms. Unless you play one hand and leaves you might not think trends. The moment you looked at the scoreboard, you are a trend player. You wait a hand to see what comes, you are a trend player.

    TRENDS DO NOT INCREASE YOUR CHANCE OF WINNING A CERTAIN HAND but, it MAY maximize the # of wins in a row if the trend holds true. It will also decrease your # of losses to a minimum i.e. one hand if the trend did not materialize. This is very helpful to those people who likes to chase their losses.
    A good example is to bet the FIELD bet 2,3,4,9,10,11,12 in Craps. Bet the field after a field roll only. This way, one can maximize the # of wins if the field runs. It will also stop one from chasing the field bet if a non field number runs.

    Can you see the benefit of following trends NOW?
     
  18. Nathan Detroit

    Nathan Detroit Well-Known Member Founding Member

    Joined:
    Dec 25, 2014
    Likes:
    2,080
    How about those ANTI TREND betting methods They do exist


    ND
     
  19. Lani

    Lani New Member

    Joined:
    Oct 15, 2018
    Likes:
    5
    Location:
    CA, USA
    I can understand what trend means. Now it comes to the problem how good you can predict the trend. Each game the house is not change. Either B or P comes still rules by the possibility. Say B comes then the trend remains, P comes the trend breaks. Casino using those boards and recording results for player is not helping your odds. The house is still on their side so they won't be afraid of any trend. They only concern when there is no people play. I discussed this before at BTC.

    If the trend works, it can be explained by some combinations of following cards may have advantage over B or P. For example, if following 6 to 10 cards have more 8 or 9, it will be good chance for P, because we know that if the first card for Player is 8 or 9, the odds would be on Player, if it is 0 or 1 or 2, the odds would be on Banker. Therefore, more 8 and 9 in the following cards will be good for Player. But the only issue is how to predict the combination? By counting cards? Maybe help. It is about randomness now.

    In short, I feel predicting trend is no better than predicting a single hand. Predicting the abundance of 8 and 9 or 0 and 1 may be a key.
     
  20. Craps

    Craps Well-Known Member

    Joined:
    Jul 19, 2018
    Likes:
    251
    Location:
    Las Vegas
    Lani,
    So, what IS the problem with following trends? I am not arguing that trend changes the odds. Don't you get a headache of WHAT TO BET if you don't make believe that there are trends. If you keep anticipating trends and one holds true, you will gain. How many times have you see people stayed out of long runs just because they were not on board when the trend begins?
    We will have to agree that trend HELPS in making a bet decision.
    It seems you have a problem with losses if you follow trends and have no problem with losses on wild guesses.
    Again trends does not mean following runs of B and P. it comes in all forms.
     

Share This Page