Article sur le continuation Bet

    • gugusg
      gugusg
      Bronze
      Inscrit le 17.03.2007 Messages : 2.825
      Voici un article plutot interressant sur le continuation bet trouvé sur 2+2. Rien de révolutionnaire mais des graphiques plutot interressant ;)


      Ah, the continuation bet, my old friend. I’ve got two strong beliefs about the continuation bet:

      1. It’s one of the easiest and most effective moves you can make in no-limit hold’em
      2. It’s one of the most overused moves in today’s no-limit hold’em games

      Yes, yes, I know that it applies to other forms of poker as well, but no-limit hold’em is the most popular game today and the one where people talk about continuation bets the most. Plus it’s the game that I’ve got data on.

      A continuation bet (or c-bet) is when you make a bet on the flop, turn, or river when you were the last one to make a bet or raise on the previous street. Most of the time, a c-bet refers to a bet on the flop made by a pre-flop raiser. You raise pre-flop and get one or more callers. Then on the flop you bet again – no matter if the flop hits you or not. The bet is simply the continuation of the strength you represented pre-flop. It is a bluff of sorts, but it is a very effective bluff since the same type of bet will be made with a made hand.

      Part of its logic is that the pre-flop raiser is more likely to have a stronger hand than a mere caller. Therefore on the flop he’s the one more likely to be ahead. When both hands miss the flop (which is a common occurrence) the pre-flop caller often gives up when the pre-flop raiser makes an aggressive move. In fact, for situations like a pre-flop raise being called by the big blind, Chen and Ankenman state in The Mathematics of Poker that the pre-flop raiser should bet 100% of the time when checked to. The difference in the strength distributions between the two hands is so great that an autobet on the flop is called for. Notice that this effectively always puts the pre-flop raiser out of position – the caller can check and decide his action after the continuation bet.

      A continuation bet needs to get everyone to fold more than one time in three (33%) to be profitable when your c-bet is half the size of the pot. One time in three you profit a whole pot. Two times out of three you lose an additional half pot. That breaks even. If your c-bets are 2/3 of the pot, then you need to win 40%; you need 50% success for a full-pot bet. These percentages assume that your hand has no value and that if you are called or raised you will concede the hand. This also assumes that the chips all have the same value, like in a cash game or early in a tournament. If you’re later on in a tournament then your bubble factor is going to be higher and you’ll need to win more frequently. This is due to the non-linearity of chip values; the chips you are risking are worth more than the ones you hope to gain. For more information on bubble factors, see my previous articles stored at http://base.google.com/base/search?authorid=1448569&hl=en&gl=US

      But how often do c-bets work in practice? Well, this wouldn’t be a Tysen Streib article without some cool numbers and graphs, so let’s dive in! There are many factors that affect how likely a continuation bet succeeds. Some of them I can’t quantify, like your reputation. If your opponents always see you putting in a c-bet whenever you can, they will start calling more often. Other factors I will attempt to quantify below, by analyzing a large database of online hand histories.


      Number of Opponents and Their Skill Level

      The more opponents you have on the flop, the less likely your c-bet will work. Not rocket science. More experienced opponents will also call c-bets more often.



      This graph shows how likely c-bets worked as a function of the number of opponents on the flop. There’s a line for a collection of $10 tournaments as well as one for tournaments with a buy-in of $100 or more. The thickness of the line represents the standard error in the numbers; I have fewer data points with 5 or more opponents, so the error is higher.

      I have a couple of caveats to this and subsequent graphs. I’m defining a success as “what is the chance a bet on the flop takes down the pot?” If anyone calls or raises, then it’s a failure. These aren’t strictly continuation bets – there is no requirement for a pre-flop raise nor for the bettor to be the one that raised pre-flop. This was purely for sample size reasons. If I had restricted my analysis to pure c-bets then I wouldn’t have enough data points for my further subdivisions below. So the absolute value of success won’t be exactly the same as with pure c-bets, but my purpose here is to show what the factors are that influence the chance of success.

      The $10 tournaments are observed hand histories – there is no common player throughout. The $100+ tournaments are all from my own histories, however, I eliminated every hand where I personally saw the flop. Therefore these statistics should be free of any bias introduced by the presence of a common player. My sample size of the higher-level tournaments is much smaller, so I won’t be using them in the further breakdowns below.


      Pairs and Flushes on the Flop

      Continuation bets are more likely to succeed when the board is paired or if there is no flush draw (a rainbow flop – all three cards are different suits). This is because other hands are more likely to miss these flops and not contest the bet.



      Again, the thickness of the bars represents the error; my sample has only 92 flops of trips where a bet was made on the flop. Fewer samples means more error.

      Although I’m not showing it here, the chance of success in higher-level tournaments is much more similar across board types. I would suspect that higher-level players will contest a paired flop more often since it’s such a common ploy to bet at a paired pot in order to pick it up. It used to be that people would bet at paired flops because that was a great opportunity for free money. Now that everyone knows that, it’s no longer a free lunch.


      Connectivity of the Flop

      The more connected a flop is, the more likely someone will have a draw and call that continuation bet. I divided all the unpaired flops into five categories of connectivity:

      • 0-Gap flops have all three cards in a row (654 or JT9)
      • 1-Gap flops have one missing card (532 or QJ9)
      • 2-Gap flops have two missing cards (AKT, 975, or 632)
      • A Mixed flop has two touching cards and one disjoint card (QJ3 or K76)
      • Unconnected flops are all others



      I would have expected Mixed flops to be a little closer to 2-Gaps than Unconnected. I guess I shouldn’t fear straight draws on those kinds of flops as much as I do.


      High Cards on the Flop

      I’ve always noticed that my c-bets work more often when there’s an ace on the flop. That’s why when I raise pocket nines pre-flop, I always put in a continuation bet when the flop comes up something like AQ6. The pre-flop raiser is much more likely to have an ace than a caller.



      This is a chart that shows success as a function of the highest card on the flop (unpaired flops only). The smaller success on the lower flops is probably due to the increased connectivity rather than the lack of high cards. Notice that ace- and king-high flops separate themselves from the rest of the pack. I’m surprised to see that king-high is essentially just as good as ace-high. Nice info.


      Bet Size

      I’m sure it will come as a shock to everyone that bigger bets are more successful at picking up pots.



      It does appear that there are diminishing returns for bet sizes above half the pot. Perhaps Harrington was onto something when he suggested that half the pot was the ideal size for a continuation bet.


      In Conclusion

      I hope you’ve learned something from this study. If you aren’t a regular user of continuation bets, I hope you start. If you’re a habitual user, I hope you learn to back off in certain situations. A good example of when not to continue bet is when you have a fairly good draw, especially if you are in position. Checking behind to see a free card will give you more information, help keep the pot small, may induce a bluff by your opponent, and eliminates the chance of a check-raise. In addition, by occasionally not c-betting, you give more credibility to the times where you do bet. This helps preserve fold equity on future hands.

      In addition, low-level players love to chase draws. When the board is draw-heavy and you totally miss, it may be better to simply give up the hand, especially if you have more than one opponent. Higher level players are more suspicious of c-bets in all circumstances and it becomes more of a test of psychology. There are some people who will call continuation bets with anything (especially in position) and then see if the pre-flop raiser gives it up on the turn. If the raiser fires a second shot on the turn, they often give up if they still have nothing. If the raiser checks, they’ll bet to pick up the pot. If you suspect you are up against this kind of opponent, you will have to bluff a second time on the turn more often when you miss. You should also check-raise the turn when you have a good hand more often than you would against other opponents. It all comes down to knowing your opposition and playing against their tactics.
  • 7 réponses
    • domib34
      domib34
      Bronze
      Inscrit le 30.01.2007 Messages : 3.827
      Qu'est ce qu'on peut en conclure ??

      Le CB a plus de chances de réussi si :

      - Petites limites
      - Pas de tirages sur le flop
      - Scare cards présentes
      - Taille bet > 60% du pot
    • gugusg
      gugusg
      Bronze
      Inscrit le 17.03.2007 Messages : 2.825
      la taille perfecte serait entre 50 et 70%, au dela c'est moins rentable on dirait.

      moi je fait tjrs à 70% le CB, ca combine une bonne fold equity (50% ca fait pas enorme je trouve) et ca permet de bien faire grossir le pot si j'ai touché :tongue:
    • gugusg
      gugusg
      Bronze
      Inscrit le 17.03.2007 Messages : 2.825
      ce qui est étonnant aussi c'est que la carte A au flop fait que le conti bet reussi alors que moi je trouvais pas, vu que les vilains suivent souvent avec des ace rags et qu'ils couchent jamais un conti bet avec ca ;(
    • Budbox
      Budbox
      Bronze
      Inscrit le 29.01.2007 Messages : 1.055
      Merci pour l'article.

      En même temps le type n'a pas vocation à tout savoir, donc c'est possible que des choses qui fonctionne mieu chez lui ne marchera pas chez nous ( je parle pour le Ace carte haute )
    • Destolon
      Destolon
      Bronze
      Inscrit le 29.04.2006 Messages : 4.204
      ouais sympa, mais en gros il donne de belles graphiques et stats pour ce qui tout a fait clair et pas vraiment neuf :)

    • cassoul56
      cassoul56
      Bronze
      Inscrit le 02.06.2007 Messages : 440
      Sympa les graphes mais bon sa aide pas beaucoup on va dire.
      Et merci quand meme pour l'article.
    • Siliciom
      Siliciom
      Black
      Inscrit le 03.02.2007 Messages : 5.258
      Message original de gugusg
      ce qui est étonnant aussi c'est que la carte A au flop fait que le conti bet reussi alors que moi je trouvais pas, vu que les vilains suivent souvent avec des ace rags et qu'ils couchent jamais un conti bet avec ca ;(
      J'ai remarqué que contre certains joueurs particulièrement loose préflop et weak postflop, l'as au flop leur fait constament peur, car la plupart du temps ils n'en auront pas vu leur loosiness. C'est aussi facile de savoir quand ils ont l'as ou pas, si fold => pas d'as (ceci dit on s'en fou), si raise => as :P

      Enfin c'est des reads que j'avais eu sur certains joueur, et cela devenait vraiment très facile de jouer un flop avec un as contre eux (en tout cas de prédire +ou- leurs mains).

      Merci pour l'article sinon, même si ça n'apprend peut-être rien de nouveau, c'est toujours intéressant d'avoir une explication faite par une autre personne.