Questions on NL bet sizing

2007-03-05T07:57:00Z

So, much to the chagrin of my at present roomate, terrencechan (because I constantly hound him with questions that don't make sense or he doesn't know the answer to), I've been pouring through the pages of Mathematics of Poker that deal with NL bet sizing (pg. 241, pg. 272-278). Since the poisoning my impressionable brain with dastardly game theoretical ideas I've been zigging and zagging trying to apply principles I've weaned from T and now the book despite, you know, the fact that I'm stupid. I even started playing HU to make these easier. However, I am playing cash games (and buying in full), so I'm constantly dealing with stack sizes larger than perhaps kept in mind during the writing of these chapters. So please, help me :D

From reading and reading and guessing and guessing, I've come to some (I hope accurate) conclusions and started working it into my game, which I'll detail afterward. However, I do have some questions that perhaps some of you smart folk who have me on friends list can shed some light on or debate with me (and if any of you want to further a discussion via e-mail or IM, let me know).

1)
On p.274 and p.275 it's discussed that if you cannot bet a geometric growth of the pot between 33% and 100% of the pot over three streets, you can add a fourth street (assuming your opponent chooses to raise). However, there is some overlap:

A full pot size bet over three streets is a perfect geometric growth when the effective stack sizes are 13 times the pot size (for example, a stack of 130 and a pot of 10). So, we move to a fourth street -- starting with about 65%, since over four streets betting 65% of the pot will move a stack of roughly 13.5 pots allin. Does this mean we start there and move back up to a full pot size bet, which over four streets will get allin a stack equal to 40 pots? Do we ever move to five streets? Is there a point where we give up with this and just bet a certain percentage? Is it then we bet 41.4%?

2)
On pg. 274 and 275 it's said that on static type boards (KK4 rainbow is given as an example, and earlier when static board is defined K82 rainbow is also given as an example) a bet of 33% to 100% of the pot is sufficient, hoping for a three-street geometric growth.

So I wonder.. does the "maximum" (a full pot bet) increase on drawy (K82 with a flush draw) boards? What about on very drawy boards? (789 with a flush draw, or three of one suit). Does it increase to 125%? 150%? 200%?

3)

On p.275 adjusting the geometric growth for dynamic boards is discussed (mediocre dynamic boards, such as K72 with a flush draw). An example is given that the geometric growth of the pot is 50%, so one should increase it to 75%. Does this apply down the board? If the geometric growth of the pot is 40%, should one bet 60%? Does 80% become 120%? Does 100% become 150%? I'm assuming the idea here is to neutralize the significance of the river bet, due to our opponent being able to have clear folds with many of their flush draws. What if, due to large stack sizes, we were forced to move our bet into the four street realm and now, despite increasing our bet, the river bet is still a full bet (because we neutralized the last bet, the fourth bet). Is this ok? I mean, there's not really a question here, I'm just confused in general on the application of increasing the bet size due to the drawiness of a board.

4)

On p.275 it says that on some very drawy boards, say 9c8c7h, you may want to assume for your geometric growth of the pot there are only two streets. I'm inferring this is because you want to get all-in on the turn -- a perfectly reasonable hope, I might add. What if you can't? (Even if, as per question 2, we allow betting 200% the pot, two streets of betting will only get you all-in if your stack is 12x the pot -- it's very common in cash game situations to have more than that). Do you move to three streets? If so, there is an overlap between static and very dynamic boards when the stack size is 13x the pot: the stack sizes are too large for a two street progression (as even betting 200% of the pot will not move the player all-in), however a three street progression of 100% of the pot will. The player on the static board is not forced to add a street, as he can get all-in without overbetting the pot.

I suppose this is easily reconciled by assuming one is allowed to bet slightly more than 200% of the pot on a very drawy board -- whatever percentage alleviates the overlap (while keeping question 3 in mind as well). But still, as you can see, this whole situation confuses me.

---

Well, I can't remember if there were anymore questions specifically related to bet sizing. Typing all of this out has sort of maxed my abilities, as it were. However, I do have two more questions somewhat related i'll throw in:

A) p.272 discusses the idea of an auto-bet for the pre-flop raiser. I inferred the idea of the raiser having a much stronger range than the caller was not totally applicable in HU situations, as it feels as though my hand ranges are somewhat close between my calling and my raising hands (this could, perhaps, be a flaw in my preflop hand ranges, of course, since i'm no smarty pants and i almost picked them out of a hat). Also, simply because of the "folk concept" of pot control my gut tells me checking should be my friend when raising from the button. Thoughts? What about raising from out of position (as a re-raise pre-flop, or raising a limper).

B) This covers blocking bets (referred to in the book as "preemptive bets" in the appendix to Ch. 15 starting on pg. 171 and on pg.277. On applying pre-emptive bets, though, I wonder the following:

-Should pre-emptive bets be made on the turn as well, or only the river? What about the flop?
-If I was the aggressor on the previous street, should a pre-emptive bet be an option? Or should one only use the pre-emptive bet when donk-betting?
-How should one approach appropriately sizing a pre-emptive bet keeping the stacks in mind (since part of the strength of a pre-emptive bet is 3-betting all-in with your strongest hands).

---

So, while I'm sure my brain is teething with more questions regarding the whole NL bet sizing idea, these are what I've written down thus far. I said earlier I'd discuss how I'm implementing it, so I'll go into that now:

Obviously, it's unreasonable to expect that one can look at every flop situation, extrapolate the optimal percentage of the pot to bet and why, then calculate your flop bet without being a little generous with the numbers. I mean, I'm a decently smart boy arithmetically, but I sure as shit couldn't tell you exactly what 61.19% of 60 is in an acceptable amount of time to use at the poker table, let alone figure that percentage. So, I tried the following (so far)...

I took 50%, 60%, 67%, 75%, 85%, and 100% as my starting percentages. I say starting percentages because, as you may guess from the questions above, it may be appropriate to consistantly add more percentages (perhaps above 100%) to the repertoire. I picked these because they provide for a pretty decent spread along the 50%-100% line, and they're all somewhat easy to calculate (at least, for my feeble brain... furthermore by "somewhat easy" I mean, while it's no walk in the park to try and quickly figure 2/3 or 3/4 of a large or tricky number whilst feeling the pressure of a poker match, one can find an exact or acceptably close enough estimate with a little bit of practice). Next, I deduced approximate stack sizes (referred to by multiples of the pot -- so if a pot were 10, a stack size of 130 would be 13, where if the pot doubled the stack size would be 6.5) each percentage would get in over three streets. The exact (I use this term loosely, because I figured these stack sizes by assuming a pot size of $12 [common to 1-2, what I presently play] and rounding my bets up or down to the nearest dollar... then I cut the stack size off to two decimal points) stack sizes are as follows:

50% / 3.5
60% / 4.83
67% / 5.91
75% / 7.41
85% / 9.34
100% / 13

Once again, the number on the right is multiples of the pot. So, if the pot is 12 and effective stack size is 42, betting 50% of the pot is nice and perfect.

The next step, I'd assume, is to process the above chart into a sort of range. At present, I'm not sure exactly how to weight or utilize the range (should the exact amount be the middle of the range, or should the range be weighted to one side? if the last bet does not quite put the person all-in, does one make the extra large bet on the river or leave the money leftover?) but it can't matter too much. So, after establishing and memorizing the range, one can utilize as follows:

It's the flop action. One takes the opponents stack, divides the size of the pot. Compares the result to the chart, chooses appropriate percentage of pot to bet, then calculates that percentage of pot, types in bet, hits bet. Tedious, yes, especially when dealing with decisions on whether or not to bet and the general pressure of playing, but easily doable without taking too long making decisions (of course, one would have to get seriously good at the arithmetic [perhaps a large amount of practice/good memorization, as the same numbers repeatedly pop up] if one wanted to multi-table!).

I use this approach with any pre-flop action (or, if checkraising, the continuation bet). If I raise to 6 and my opponent calls, the pot is 12. When he checks the flop and our effective stacks are 156, I consult the chart and see his stack is 13x the pot (I should really develop a variable to refer to stack/pot, you know, like "M" in the Harrington books, so I don't have to consistantly refer to it as "stack is 13x the pot"... then again, M is one's stack/the pot, except pre-flop, so maybe I should use that... sigh it's too late). and I can bet 100%. Similarly, if my opponent raised (to 6) and I re-raised (to 18) and he called, the pot is 36. While our starting stacks are the same as last hand, the effective stack on the flop is slightly less (144) and the pot is significantly larger. The stack/pot drops to to 4 (and now I bet 50 or 60% of the pot).

Similarly, I use the same approach when check-raising. If my opponent raises (to 6) and I call, then check, and they bet (12), I assume the pot is 12 [the preflop action] + 12 [his bet] + 12 [the call I make before I raise his bet], totalling 36. If our stacks started the same as the last two examples it will once again result in 144/36 (which is 4) and I'll bet 50 or 60% of the pot. Let's say 50%, so that's 18. So, I'd type 40 in the little box and hit "bet".

Simple huh?

Not really :-) but hey! It's all in the name of winning, right?

I have a headache. I don't want to talk anymore, out of fear I'll continue to make less and less sense.

Non-poker poker quote of the day

2006-11-14T21:03:29Z

"Everyone wants to be right, but no one stops to consider if their idea of right is right"
-FM Alexander

What makes a fish?

2006-11-14T00:16:22Z

I've been datamining quite a bit recently, and feel that probably the most useful and reliable way to use this information is in game selection. However, this leads one to wonder, what stats constitute a fish?

My old school method of game selection utilized one item: percentage of players seeing the flop. This method, however, is significantly less useful when playing shorthanded games compared to when playing fullhanded games. Later, playing short or heads up, I came up with an even greater method of finding the fish: short stacks. If someone didn't have the max buy-in or more, I assumed they sucked. This method is almost always correct, but obviously would leave me oblivious to people that suck and have lots of money (the best kind).

A friend tipped me off to a much better way than to try and find all the kinds of fish: try and find all the sharks, and just avoid them. Now I do realize I am playing 1-2NL, and obviously if anyone scares me at this limit I should throw away any hopes of being a winning 5-10NL player, but I'm more after this to maximize my winrate than to actually try and avoid anyone/any table that might turn me into a losing player (something I rather doubt could happen at this limit). So, a better question than one provided earlier: what makes a shark?

Easiest one: winnings. This would be even easier to spot as a shark probably plays often, and someone who plays often will have more hands in my database, and a stat like winnings would be more dependable. Then there's the triumverate of stats: VPIP, PFR, AGG (voluntarily put $ in preflop %, preflop raise %, aggression factor). Plenty of fish are fish because they're too tight as opposed to too loose. Plenty are fish because they raise too much as opposed to too little. Plenty are just fucking manaics, and plenty are just passive sheep. What statistics should one look for?

I realize it's somewhat silly to alienate anyone who may read this blog by posting stupid shit and then just posting nothing for weeks and then pose a discussion oriented question but hey, some of you were suckers enough to add me to your LJ friends list and you'll see it anyway. My opinion is one should really just look for passive people because they're the easiest to identify -- low PFR, low AGG. One may miss out on those who leak cash due to being maniacs, but there aren't that many dedicated, constant maniacs out there anyway.

siyokoy

Questions on NL bet sizing

Non-poker poker quote of the day

What makes a fish?