Excession - Comment on your Articles

[nofoldemholdem]

Excession,
First off, I want to commend you on writing some great articles. I read through them and I got alot out of it.

However, I must say that I found 1 misconception, and 1 flaw worth mentioning.

The first misconception is your description of PostFlop Aggression. This number is calculated based on (Bet% + Raise%)/Call%. It does not use fold or check numbers in the calculation. Several times you mention that Folding would affect PostFlop Aggression, and it doesn't.

The 2nd flaw is a little more serious. I'm refering to the parts where you talk about which player types have been making money, and which types have been losing money. Your data seems to show Aggression as the common factor for winning players, more so than Tightness. However, with a sample size of only 20 hands, or 50 hands, the aggression you are seeing may be a reflection of the cards the player was dealt, rather than the natural aggression of the player.

Take 50 hands for example. Say a player has 28% VPIP and plays 14 hands. Of these 14 hands, he raises 3 of them, giving him a PFR% of 6, which you consider to be Passive PreFlop. This player ends up having negative BB/100 at the end of the 50 hands. Now take this exact same player, and give him pocket aces for one of his hands. He raises this hand PreFlop, putting his total to 4 hands raised, and PFR% of 8, making him Aggressive PreFlop. He ends up winning the a pot with AA, and he ends up having positive BB/100 at the end of 50 hands because of that pot. Now, did this guy win because he is an aggressive player? Or was he an aggressive player because he got 1 good hand? I'd say he ended up with +BB/100 because he got a good hand, and that good hand also made him appear to be aggressive. This is the problem with such a small sample size. 1 good hand can sway everything. And since you are categorizing things by Aggression, the flaw is that it's much easier to be Aggressive when you have good cards.

The above example was for PreFlop Aggression, however things are even worse for PostFlop, since there are even fewer Bets, Raises, and Calls at this stage. Of the 14 flops seen above, say the player bets on 3 of them, raises on 1 of them, calls on 3 of them, and folds on the rest. His PostFlop Aggression would be 4/3 = 1.33, which is considered Passive. Now lets say one of those hands he called on, he actually flopped 2 pair instead, and raised. His PostFlop Aggression now becomes 5/2 = 2.5, very aggressive! The player wins the 2 pair hand, and ends up winning for the session. Again, did he win because he was aggressive? Or was he aggressive because he had a good hand?

With such a small sample size, 1 action can change whether a player is passive or aggressive. And it's very likely that the aggressive players were winning players, because they were getting better cards. My examples were based on your data for 50 hands... at 20 hands, I can only imagine how much worse it gets.

That's my 2 cents.. I'd like to hear some discussion.

[excession]

[nofoldemholdem]

[excession]

You could categorise on the basis of a high went to river but low WtSD% - and call him the 'chaser' - the problem is that you would need an extra category and a lot of hands played to be certain of this player type and I'm not sure it would alter your best play.

He will be classed as a 'fish' by default. So you would bet into him. This would be the right thing to do - you might be doing it because you expect him to fold, but it's still the correct thing to do if you knew he was in fact a chaser...

When you say Loose Aggressive or Tight Passive I'm not sure what you mean exactly.

None of the passive post-flop players are winning types (OK individual players who can read well or are slowplayers in a garden of maniacs may be winning players, but as a strategy post flop passivity sucks).

In contrast almost all the post-flop agressives are winning types - ABC,TPA, TAA, sLPA, sLAA - anyone less than 35% Vp$iP with a decent PFA can expect to win a bit..

Even LAA, LPA do ok unless out and out maniacs.

By contrast the post-flop, passives have a dreadful time - CS, TPP, TAP, sLAP, sLPP, LAP, LPP - these guys lose...TPP's lose less but they still don't outrun the rake..

Now what you are saying is that in small sample sizes this is self-selected for strong hands- get strong hand/good flop-bet or raise more - win - be shown as aggressive and winning - get slightly weaker hands/worse flops call more (as you point out folding doesn't actually affect the stats)- be shown as more passive

I do recognise this as true to some extent (which is why I do point it out in 2a). I'm all ears as to how to try to take it into account other than noticing that as you filter by more and more hands played the winners stay winners ad the losers stay losers and the nos don't move much - the only exception is maniacs who move from losers to winners between 20 and 50 hands - but that's because a far more powerful self-selection is going on - they self-select for luck - they double up or bust - the ones left at 50 are lucky and rich and a small fraction of those that started out..

I think if you are going to use GT+ at all for folks who have played less than 200+ hands you have to make some assumptions and ignore issues like this to some extent.

The remarkable thing for regular users (as AP10 who play 15K hands/week says in the curent results bankroll thread) is just how good at predicting actions GT+ is. I find that pre-flop stats settle down after a few as 20 hands and that the post-flop ones are pretty solid from 40+. For every player with 35 hands played and a Vp$iP of 40%+ who is just hitting great starting hands, there are ten who are just loose. The evidence that the issue doesn't invalidate the stats isn't theoretical but empirical - it really works to predict outcomes very quickly..

It reminds me of a long debate I had when playing cricket once near a swarm of midges - I reckoned that maybe the wall of air moving in front of a cricket bat would move the midges out of the way if you swung it at them - they would disperse rather than be swatted. ...after 15 minutes on and off discussion one of the guys took the bat, waved it through the cloud of midges and we all saw the midge blood on it for ourselves...

Empirical testing - the core of scientific knowledge..

[nofoldemholdem]

[excession]

[APerfect10]

I agree with Excessions statements and results. My DB is huge . Here is my data for TAA & TPA in the same format as above...

Hands (Players) PTBB/100

TPP (Rock)
20 hands (1388 players) -1.12
50 hands (616 players) -0.33
75 hands (378 players) -0.06
100 hands (255 players) -0.57
150 hands (143 players) -1.46
200 hands (98 players) -0.91
500 hands (22 players) -3.19

TPA (Mouse icon)
20 hands (2196 players) +4.02
50 hands (1744 players) +3.93
75 hands (1395 players) +3.91
100 hands (1130 players) +3.81
150 hands (818 players) +3.62
200 hands (598 players) +3.82
500 hands (172 players) +4.85

This should clear up the sample size issue as both of our databases are essentially identical for our purposes.

Are you basing these assumptions on your data, or as general hold'em wisdom?

Clearly being more aggressive post flop is A LOT more profitable than being passive. Both pure data and general hold'em wisdom point to the same conclusion. To believe otherwise is ignorance...

[APerfect10]

To even go further, being more aggressive pre flop is also more profitable than being passive. Comparing a TAA vs TPA where both play similar games; however, the TAA raises more frequently pre flop.

TAA (Eagle icon)
20 hands (468 players) +6.85
50 hands (328 players) +6.37
75 hands (248 players) +5.72
100 hands (192 players) +6.93
150 hands (130 players) +5.84
200 hands (83 players) +5.79
500 hands (15 players) +1.75

[APerfect10]

Excession, I've always wondered how you determined the passive/aggressive pre flop line of 8%. Was this just the default? Is this standard? I am interested in learning how this number was derived.

It would be interesting to analyze my database and find the optimal PFR%. This may be rather tricky to do considering this is only one of many factors that plays a part in the winning process but I think it would certainly be possible. Maybe there isnt a clear cut and dry number but I may try to find it