Introduction

My problem with this post is that there is so much I want to say on the subject. I have decided that I am going to make this post more about thinking about opponents in general, rather than their specific tendencies, though discussing specific tendencies is unavoidable.

For some discussion of the math about using PT3 and HEM, go (re)read CotW 18: Getting the Most Out of PT3 and HEM. There was some good discussion in there about using individual stats, and how reliable they are in various size samples.

I. Describing Villains' Stats: The Field at NL $100

I have been playing around with the players tab in HEM recently, and I am on a mission to convince you guys that the field is not as bad as we think it is.

Here are the average stats for players in my DB since January 1:



I'll go into a little more detail later, but, as you can see, these stats aren't that bad. The average player leaks a little, but, basically, is getting a solid return on his entertainment dollar. It's not costing him that much to donk around online every now and again. He gets plenty of fun and excitement for his money.

Obviously, the regulars do a little better. If I remove the players I have fewer than 500 hands on, the NL $100 field looks like this:



I should note that, net of my winnings, this field of 89 regular players is losing money at the rate of about 2.5 big blinds per 100 hands. But, unsurprisingly, they are doing better than the field in general.

II. Describing Villains' Stats: The Field at NL $50

This is my largest and most accurate sample:



Again, net of my winnings, the field is losing money at about twice the rate it appears to be losing at.

Here are the regulars:



Net of my winnings, the regs in this sample are losing money.

III. Describing Villains' Stats: The Field at NL $25

I created this smallish sample expressly for the purpose of this post, so it is probably less accurate. If a $25 reg wants to run these reports and post the graphs as a response to this post, I would appreciate it.



And here are the regulars. because I don't play $25, and my sample was smaller, I cut the requirements for being a regular down to 200 hands.



Evaluating Villains' Stats: The uField

These stats tell a coherent story about the field playing micro stakes. It is a story of slightly increasing skill development.

At $25, the filed plays generally loosely and passively preflop, at 20/8. The field is pretty aggressive postflop, with an AF of 2.2, but makes far too many mistakes postflop, losing most showdowns and not winning enough hands when they see the flop.

In general, then, your average $25 table is loose passive preflop, and can be exploited by a solid tight/aggressive style of play. By far the biggest leak that the field at $25 has is limping too much; this leak can be exploited by isolation plays and value raises in position.

Postflop, we exploit the field at $25 by value betting our good hands. It is obvious from the field's stats that they go to showdown second best about 7% too often; so we don't get fancy against them, we just build big pots with, in general, a better hand than they have.

We should be careful making fancy plays against the $25 field; they can't fold, so we shouldn't be trying to push the average player off of the best hand in most situations. Pick your spots to bluff average players very carefully. Probably the most useful bluff at $25 is going to be the continuation bet IP after isolating a pre flop limper.

At $50, the field has tightened up a bit, but basically has the same leaks as the field at $25, just to a slightly lesser extent. Thus, the same general advice applies.

If you look at the stats for the field at $100, you will see that their stats tightened up and became a bit more aggressive than the $50 players; so the trend holds, but this is a uFR post, so I'm not going to spend any time analyzing them.

Evaluating Villain Stats: The Unknowns

The fact of the matter is, we don't play any hands against "the field," we play hands against individual players. A lot of them we are seeing for the first time. What assumptions can we make about these players, (other than that they are fish)?

Here are my thoughts, again, based on my analysis of the HEM players tab stats:

1. Unknown players are looser than average. This is a very clear trend. Based on a review of the NL $50 players that I have between 15 and 100 hands on, the average seems to be a vpip in excess of 33%.

2. Unknown players are either very passive or very aggressive preflop. I didn't see many players who played, for instance, 33/22. Usually, unknowns played a highly loose aggressive style such as 33/29, or a loose passive style such as 33/8. This evidences two completely different styles of villains, and you must quicky evaluate an unknown player and put him into one of these two categories as quickly as possible, as the way you should exploit them varies considerably. More on this later.

3. Unknown players are big losers. Moreover, they tend to lose their money quickly. Of the 13,400 NL $50 players on whom I have 15 or more hands, I have exactly 15 hands on 250 of them. Almost none of them are winners in the samples I have on them. This holds true for players on whom I have a few more hands--I have less than 100 hands on 80% of the players in my database.

4. There is a dramatic difference in the stats of the players on whom I have <100 hands and those on whom I have >100 hands. This truth is likely a result of Darwinism. Players who play 33/29 or 33/8 are going to lose a stack quickly. To survive, you need to have a little more discipline than this. Thus, even by 100 hands, stats become far more reasonable. Just glancing quickly over my database, it appears that the average stats on players I have more than 100 hands are are much more in line with regulars' stats than with the field's stats. So if you have 100 hands on a player, you should assume that he is at least somewhat reasonable. This is not to say that he will not have huge leaks--he will; but assuming that he knows at least a little something is a good idea.

VI. Evaluating Villains' Stats: A Useful Framework

In his book, "The Psychology of Poker," Alan Schoonmaker developed an excellent framework for evaluating villains' play by comparing it to the average play for that game. I'll explain and adapt it here.

Basically, Schoonmaker's idea is to rate players on a scale of 1 to 9 on tightness and on aggression, again, and importantly, compared to the field, not some arbitrary hypothetical ideal. Thus, using this framework, at NL $25, a player who played 20/8/2 would be rated a 5,5--completely average for the game. A tight aggressive player (a "stone killer," in Schoonmaker's words) would therefore be someone far tighter and more aggressive, say, a 13/10/4, and would be rated something like a 2,8 on the Schoonmaker rating system. A loose aggressive player, a "maniac," would wind up at about 36/22/3 or so, roughly a 7,8 in Schoonmaker's system.

I introduce the idea of Schoonmaker's framework not to suggest that you use it (although I have player notes on people that consist of nothing other than "7,8, lol"), but for two other reasons:

1. To emphasize the fact that we should be evaluating villains' stats in comparison to the field for some purposes; and,
2. So that we can accurately characterize and exploit our opponents and their motives for playing poker.

VII. Evaluating Villains' Stats: Exploiting Player Types

This section will also borrow heavily from Schoonmaker's book; all of the psychological insight is his, I have simply applied it to uFR player stats and tendencies.

1. The Average Player Playing 20/8 or 19/9. In general, this player is too loose and too passive preflop. He generally will not be making a lot of huge post flop mistakes; he is good enough to develop a decent sample on, to get away from obvious second best hands, and he does have somewhat reasonable starting requirements. I discussed these leaks and exploiting them above--basically, sit to his left and isolate him, tend to value town him and fold to his aggression unless you have a monster. Schoonmaker's key insight about loose passive players in general is that they are nonconfrontational in general, and that they are primarily playing to pass the time. They like the idea of winning, as you can tell from the fact that they have somewhat reasonable hand selection standards. But they are unwilling to put in the study necessary to become solid players (IRL passivity) and they are too nonconfrontational to be as aggressive as it takes to consistently play winning poker.

2. The Loose Passive Player Playing 34/4/1.5 This player is the extreme version of the average player and the average player's leaks. Everything I said (or paraphrased from Schoonmaker) about average players is true for these players, only to a much greater extent. Sit on their left, isolate ruthlessly, don't try to barrel them off their hands with air, but value town them relentlessly, and turbo muck any non-nutty hand if they raise.

To spot them, look for an unknown who posts a dead blind and then starts limping into the first few pots. If you limp even two of the three first hands you get dealt, I will assume until proven otherwise that you are loose passive.

3. The Loose Aggressive Player (Maniac) Playing 34/29/3. Loose aggressive maniacs are playing poker primarily for the opportunity to act out their aggressive and competitive impulses. They usually think they are excellent players and tend to ignore their big losses and remember their big wins. Their egos are wrapped up in dominating the table. As Schoonmaker points out, in live games, the maniacs tend to be loud, obnoxious, flashy and literally throwing their chips into the pot. In an online game, they are harder to spot, but I look for guys who sit at the table, post a dead blind, and quickly start raising their first few hands.

I don't usually care whether I have position on them or not, because I can as easily check/call 3 streets as I can call 3 streets. the way I exploit these guys is to have a better hand and then just let them barrel into me. It is high variance, as they can flop well-disguised two pair hands and the like (and, contrary to popular wisdom, 33/29's are allowed to flop sets, btw), but I think it is the most effective way to exploit these players. I would hazard a fairly informed guess that my aggression factor in hands against maniacs is <1.

4. Tight Passive Player Playing 10/4 There aren't actually that many of these guys playing full stacked; most of the players playing this tight are shortstackers. The general tendency for tight passive players is to be extremely risk averse. They are not at the table to gamble even a little bit. They want to win at a slow steady rate, and minimize their losses. This is why playing short stacked appeals to these players of this personality type; it is an additional risk mitigation measure. Online, players of this type are as interested in grinding out player points as they are in winning money at the table. You can think of rocks and shortstackers as the online equivalent of the old retired fogeys in the casinos that read the newspaper at the poker table and fold, fold, fold, until they pick up aces.

Obviously you can't gamble with rocks, you need to avoid them until you have a hand that matches up well with their range. For short stack rocks, this means waiting until you are ahead of their range; for full stack rocks, you need to play only hands that are ahead of their range or well-disguised hands that can flop well and stack their range reliably.

Conclusion

Everything old is new again. In my opinion, Schoonmaker's book was groundbreaking when it came out, and I have read it several times. But I have not seen the concepts discussed on 2+2 recently, and, imo, this is a significant leak in uFR. So I wanted to use this post as a way of reintroducing his psychological insights. HEM gave me the capability of describing his player types in a fairly statistically rigorous way. I left out of the post a lot of the number crunching I did to apply stats to the player types he describes in general terms. Basically, I look at this post as a refinement of Schoonmaker's work that is only possible because of the software we have available that he didn't.

If there is a takeaway from this post, it is the same takeaway as you should get from Schhonmaker's work: Don't assume that most villains you are in a hand against are playing to win. They are not. Most of them play poker primarily for other reasons, and are, therefore, exploitable. Their stats tell us not only how they play, but why they are playing, and their motives are every bit as exploitable as the playing style that results from their motives for playing.

1rd

first, tl dr.

This is so long, I am sure it is outdated as the games are progressing at the speed of light. Good try though, Matt.


I thought I would add that your last paragraph is soooo true. On frequent occassions, I have been known to chase improbable draws against regulars that will tilt if I hit. Losing sessions usually, but the loss is more than made up for by the hilarity of the post hand chat. There are a few on 2+2 that I 3Bet PF ANY time I get position on them.

It's fun.

I totally buy it. The majority of my thinking these days is not "how do I stack this fish?" but rather "how do I out level this guy?"

Or slightly shrinking rake contribution??


Only read part 1 and have to leave work now... will get to the rest tomorrow. Thanks mpethy.

out of my top 50 winning regs in my DB at NL50 with decent hand samples, most are short stackers.

mine are mostly rocks. but there are winners in my db with all different styles.

I concur. At the 200 and 400 level it amazes me how many 11/9 regulars are successful. Of course it is ALL relative. It takes these guys twice/four times as many hands to win the same that a kush/ricky/lacie takes......but these NITs are successful nonetheless.

These guys get ripped all the time at the SS and MS forums for being so NITty, "how can anyone pay them off?" Well, because they aren't always playing against the regs.

very nice post

Great post mpethy. I was however hoping on some in depth discussion in this thread on how different stats that we have available to us on our huds relate to one another. As a mass tabler my game relies fairly heavily on stats - I.e. I'm a hudbot.

There are a lot of extra insights that can be acquired when say taking stat x and stat y and drawing conclusion z in regards to villain's play.

A basic but good example of this is AF relative to Vpip. Its pretty common to see a 9/7 player with an AF > 5 and its important for newer players not to rate this player as overly aggressive.

Comparing ATS to PFR can give a very good idea on how positionally aware a villain is. When the gap is large you can deduce that they are obviously raising more in late position but also that they are playing much tighter UTG. There is a big difference imo between a 12/9 with an ATS of 15 and a 12/9 with ATS of 40.

I could probably think of a bunch more examples and hope others chime in and give some more ideas on this too but I'm posting from my phone right now and my thumbs are just so damn big.

Excellent, thank you for your work.

Thank you M.

Interesting. I read Schoonmaker 2-3Yrs ago and have been trying to find the time to get back to it. Maybe this will spur me on.

Great work OP, gonna read it several times until the next cotw.

[QUOTE=mpethybridge;11689874]Introduction

there is so much I want to say on the subject.

Go for it! The interest is here!