Conceptual question: systematic tree of betting patterns

[preparac]

Hi people

PT3 is really a great tool as it allows us not only to construct sophisticated stats but also it somehow forces us to understand systematically the flow of action throughout a hand. The first year or so of my PT3-self-schooling experience (with the periodic help of Kraada, Whiterider and the PT3-community – THANK YOU GUYS!) I spent trying to understand the basic nature of SQL i.e. PT3 and its applications to simple preflop queries and to a few isolated postflop problems. But now I feel the time has come for a new, more ambitious systematic approach which reflects the logical nature of THNL-Betting patterns and goes beyond isolated questions like for example how many times a player folds to a flopraise after cbeting and so on …

So I would like to construct stats which reflect systematicaly a tree of betting lines or patterns postflop – and that for every relevant flop situation (as for example: a player cbets in absolute position – what does this mean, how he reacts to resistance not only ? I do not plan to do this for any possible situation of course, as they do not happen often enough and you would need to have giant databases in order to get reliable data)

Instead of constructing blindly stats for special cases (a s I did until now) this time I want to get the concepts clear first. I want to start with HU pots on the flop (as betting patterns in multiway pots are much more complex). What I do not have clear in first place is which tendencies we should difrerenciate. So I would appreciate your help to develop this conceptional problems, guys.

I’m thinking about the following patterns

1. FwoA Folding whitout showing aggression = weak tight pattern
2. FwA Folding after showing aggression = bluffy/weak pattern (stabbing)
3. WTSDwoA whitout showing aggression = passiv pattern
3.a) L$SDwoA whitout showing aggression = passiv/weak hands pattern (calling stations)
3.b) W$SDwoA whitout showing aggression = passiv/strong hands pattern (pot control)
4. WTSDwA after showing aggression = agressiv pattern
4.a). L$SDwA after showing aggression = loose-agressiv/bluffy pattern (spewy)
4.b) W$SDwA after showing aggression = agressiv/valuebetting pattern
5. WWoSD = agressiv pattern

So if a player cbets the flop oop in and I want to know what this means, I would in a first step evaluate the percentage of the mainpatterns

1. FwoA % high = the Cbeter folds to a flop raise, or calls a flop raise and XF to a turn bet, or folds to a turn float … and so on. If this percentage is high, I shouldn’t give to much credit to this Cbet and consider to push this player out of this hand.
2. FwA % high = the Cbeter double or even triple barrels light and gives the hand up often, so I can call him down light or even try to bluff back.
3a) L$SDwoA % high = The Cbeter calls his hand down to light (so I do value bet him but do not try to bluff him)
3b) W$SDwoA % high = The Cbeter plays pot control with his strong/medium hands … not much I can do about it.
4a) L$SDwA% high = The Cbeter agressivly continues also with weak holding (so I call him down light)
4b) W$SDwA % high = The Cbeter valuebets agressivly his strong holdings (so I better fold to the double barrel…)
5) WWoSD % high = The Cbeter continues agressivly – (I probably should try to play back at him sometimes, especially when the FwA-% also is high)

I think, these pattern cover most of the standard straight-forward player types (not so much the trapy ones, but they are trickier to figure out anyway). So I would go on and construct the nominator to these 7 patterns accordingly. But what I do not have very clear is how the denominators should look like. Just all hands the player was cbetting oop? I think it could be sort of a challenge to process the 7 patterns altogether in the heat of the fight …

Anyway – I welcome your comments/considerations about this project. Do you see there any conceptual or practical flaw in my approach?

Thanx, prep.

[kraada]

First off the underlying goal is great. Not only am I totally on board with you there, but I've been tinkering in my spare time with similar ideas.

Personally, I've yet to come up with a coherent and consistent system that I feel is sufficiently complete. (If I'd come up with one, we'd have a new post in my Advanced HUD! :)

Looking over what you've got, I have a few questions to start: When you say having "shown aggression" what do you mean? Your example was a person who c-bet OOP - does the c-bet not count as aggression? Also they had a chance to cbet, does their preflop raise not count as aggression? I first need a grasp on how you are defining aggression before I can evaluate further.

(Note: If you are actually trying to talk about momentum changes, that's something else I've been meaning to look at - that is, if a person doesn't have the betting lead and takes it, how often are they really strong vs. how often if they have the betting lead they surrender it and in fact are really weak.)

[preparac]

With “showing aggression” I mean that [/b]starting with the original flop situation[/b] the player bets or raises at least once (i.e. he bets/raises again after having cbet).
As I stated in my post above I think it might be too ambitious to try to build a complete betting pattern/tendencies tree for all possible postflop situations. But I think it should be achievable for all relevant flop (and in a second step possibly turn) situations.

I would proceed in two steps:

First I want to define the relevant [/i]starting situations[/i]. I think we could start with all "primary" situations (i.e. first action of the player situations - you sure will find a better expression for this) as there are
in a PF raised pot:

donk opp, player: X
donk opp, player: B
cbet opp oop, player: X
cbet opp oop, player: B
cbet opp iap, player: X
cbet opp iap, player: B
float opp, player: X
float opp, player: B
face donk, player: C
face donk, player: R

and also the “secondary” situations (of course you can argue that not all of them are of the same importance and therefore necessary …)

donk opp, player: XC
donk opp, player: XR
donk opp, player: BC
donk opp, player: BR
cbet opp oop, player: XC
cbet opp oop, player: XR
cbet opp oop, player: BC
cbet opp oop, player: BR
cbet opp iap, player: BC
cbet opp iap, player: BR
float opp, player: BC
float opp, player: BR
face donk, player: RC
face donk, player: RR

(similar starting situations also apply for limped pots, and of course you could even make difference between PF raised and 3Bet pots …)

Second I want to apply youseful betting patterns to all of this situations.
I ask: Player had opportunity to donk in a PF raised pot. He didn’t. What this mean?. I already have a stat that tells me, this guy donkbets 10 % of the time when he could donk the flop, he checkraises the flop 6% of the time … , he folds 25% to a 3Bet, he has a WTSD of 40% … this indicates that .. a …a … timebank … ah what the heck, I want to have the following numbers:

Wenn this guy checks oop in a PF raised Flop he
1. FwoA (folds the hand whitout showing any aggression) = 50%
2. FwA (folds after showing some aggression) = 0%
3.a) L$SDwoA (goes to SD and looses whitout showing aggression) = 15%
3.b) W$SDwoA (goes to SD and wins whitout showing aggression) = 0%
4.a). L$SDwA (goes to SD and looses after showing some aggression) = 0%
4.b) W$SDwA (goes to SD after showing aggression) = 10%
5. WWoSD = 5%

Then I know: This guy probably check folds when I bet the flop, but I better check behind and see what he does on the turn, because he obv. never stabs at the pot and only puts money in the pot when he has a strong hand and he could be “trapping”.

And when I have a guy with stats like
1. FwoA = 50%
2. FwA = 20 %
3.a) L$SDwoA (goes to SD and looses whitout showing aggression) = 0 %
3.b) W$SDwoA (goes to SD and wins whitout showing aggression) = 0%
4.a). L$SDwA (goes to SD and looses after showing some aggression) = 20%
4.b) W$SDwA (goes to SD after showing aggression) = 5 %
5. WWoSD = 15%

… I can check behind I play back at him when he tries to stab at the pot ….

And when I have a guy with stats like
1. FwoA = 60%
2. FwA = 0 %
3.a) L$SDwoA (goes to SD and looses whitout showing aggression) = 25 %
3.b) W$SDwoA (goes to SD and wins whitout showing aggression) = 5%
4.a). L$SDwA (goes to SD and looses after showing some aggression) = 0%
4.b) W$SDwA (goes to SD after showing aggression) = 5 %
5. WWoSD = 5%

… I see that he is a passiv calling station and I can valuebet him thin in these spots

And when I have a guy with stats like
1. FwoA = 70%
2. FwA = 0 %
3.a) L$SDwoA (goes to SD and looses whitout showing aggression) = 5 %
3.b) W$SDwoA (goes to SD and wins whitout showing aggression) = 20%
4.a). L$SDwA (goes to SD and looses after showing some aggression) = 0%
4.b) W$SDwA (goes to SD after showing aggression) = 5 %
5. WWoSD = 0%

… this is a weak tight passiv player and I should push him out of the hand with any two cards but not try to valuebet him light when he calls any bet …

And so on. Similar (if not same identical) reasonings would apply for other starting situations, even if the percentages of the patterns would differ heavily of course…

[preparac]

I screwed up some percentages but I think you get the point anyway ...

[preparac]

ah yeah ... and I already constructed the super-weak-tight "... and surrender" pattern which is sort of a subpattern of pattern (1). "... and surrender " indicates all cases where the player folded to the very first bet (allways starting with the primary or secondary flop situation). I didn't include it in my pattern list, because in this case I think it's obvious that we shouldn't take all all starting situation cases, but only these where the playser actually was confronted with a single bet (or a raise after his starting-bet on the flop) later in the hand.

[preparac]

Looking over what you've got, I have a few questions to start: When you say having "shown aggression" what do you mean? Your example was a person who c-bet OOP - does the c-bet not count as aggression? Also they had a chance to cbet, does their preflop raise not count as aggression? I first need a grasp on how you are defining aggression before I can evaluate further.

I hope I answered your question with this. Short: I try to take into acount the preflop and flop action by defining a set of standard situations and then I want to know/sort out a players betting patterns/tendencies for the rest of the hand (F/T/R) starting from this point.

[kraada]

Ah I see, so the idea is to have these 7 options listed for a bunch of common spots? That's certainly doable, though it's obviously going to take a lot of stats.

One thing you might also want to take into consideration is that it is still going to take you time to look at these stats during play. If you're going to try and analyze 7 different stats each time you see a flop it's going to take you quite a while to get used to it and even then it's going to take time to do while playing - so it will limit the number of tables you can use this sort of strategy on effectively.

[preparac]

That's certainly doable, though it's obviously going to take a lot of stats.

I'm also worried about this, esp. since this 1600 columns limitation is already now a problem for me ...

One thing you might also want to take into consideration is that it is still going to take you time to look at these stats during play. If you're going to try and analyze 7 different stats each time you see a flop it's going to take you quite a while to get used to it and even then it's going to take time to do while playing - so it will limit the number of tables you can use this sort of strategy on effectively.

hm, yeah I also have my doubts about this. I was thinking a while about how to aggregate the betting patterns (agression/no aggression) and results (SD/no SD, win/loose) better, i.e. to construct appropriate denominators for them. This would be important also because sometimes it depends of the opponents if you have at all the opportunity to follow a pattern (... if they do not bet or raise, you cannot fold or call). But I couldn't figure out an easyer way which allows a meaningful interpretatation ... I'ts a huge difference if you go to SD because you bet your draws aggressively or if you you go there because you call villains down with your middle pairs ... Do you have any idea how to do this in a simpler way

[kraada]

I haven't come up with anything I've really liked yet, but the latest thing I was going to look into was how momentum and bet sizing changes affect hand strength, to see if there are some people who give off huge tells when they either bet specific sizes or when they take control of the betting in certain spots.

As to a general way to clump betting patterns together quickly and easily, I'm not sure. That's one of the reasons poker is really complicated, there are a whole heck of a lot of permutations.

[preparac]

Yes ok I agree ... beside of bet sizing we also should consider effective stack sizes, specially in tournaments (which we cannot because we do not now our opponents stack sizes i think?), table dynamics, ICM-impact and so on ... we will never have stats which reflect all this factors of course.

Beside of this: What you think about the seven basic betting patterns I'm thinking about ... Do they make sense in your eyes?

Your momentum change idea sounds intersting ... but also there there are so many different situations and dynamics going on ... i hardly can imagine that you can aggregate all these situations in a simpel way .. but I'm looking forward.