One third of the fish and maniacs are winning -- some PT stats

[Blarg]

This gives me hope for the future of online poker!

I did some classifying of players, in accord with standards I note below, which someone recommended a few months back here on the 2+2 forums, and came up with some interesting figures. NOTE: THIS WAS DONE ONLY FOR PLAYERS WHO I HAVE RECORDED 100+ HANDS ON. THIS MEANS MANY PLAYERS HAVE BEEN LEFT OUT. It is also devoutly to be wished that this filtering out of players who have not played 100 hands or more will help a bit in avoiding prematurely pegging players who have nutty 10 or 20 hand swings as the wrong type of player. You have to draw the line somewhere, so that's where I picked.

This sample goes from .50/$1 to $3/6 on Party and Empire, but I kept 6-max out of it because applying the same standards for 6-max and regular games wouldn't work. The figures are taken from Pokertracker's Summary tab.

I was very curious about figures like these when I first started out playing hold'em a few months ago, and I remain so, so I provided them here for people who might be curious about the kind of figures a roughly 50k played-hands database, all low-level, limit games, might yield.

TOTAL PLAYERS: 8,591
TOTAL HANDS PLAYED: 438,346
SESSIONS: 11,970
HOURS: 7,349.20
BB/100 (3.05)
VPIP: 31.91
PFR: 5.42
W$SD: 46.94
Winners: 3612 or 40.14%
Losers: 5,387 or 59.86%
HANDS DEALT: 47,979
TOTAL RAKE: 36,585.75

Okay the above includes everyone not in my 6-max games.

Below are the players broken into categories.

FISH
Total Players: 1,496
Hands: 61,437
Hours: 1,082.57
BB/100 (10.43)
VPIP: 60.21
PFR: 6.51
W$SD: 42.66
Winners: 511 or 32.71%
Losers: 1,051 or 67.29%

NOTE ON FISH: I also included those who I manually noted as fish while playing. Almost all of these were not just fish according to numerical measurements, but people of such outstandingly bad play that I put them on my buddy list. I DID NOT label someone a fish because they beat me -- I want useful info in Pokertracker, not just anger and bad beat stories.

==================================================

MANIACS

Total Players: 21
Hands: 3,138
Hours: 55.28
BB/100 (10.27)
VPIP: 61.85
PFR: 15.87
W$SD: 39.02
Winners: 8 or 33.33%
Losers: 16 or 66.67%
================================================== =
LAGs

Total Players: 36
Hands: 5,446
Hours: 91.55
BB/100 3.58
VPIP: 37.75
PFR: 12.78
W$SD: 48.17
Winners: 25 or 64.10%
Losers: 14 or 35.90%

================================================== ==

TP(tight-passive)

Total Players: 255
Hands: 92,545
Hours: 1,548.51
BB/100 0.79
VPIP: 16.32
PFR: 3.79
W$SD: 55.74
Winners: 155 or 48.29%
Losers: 166 or 51.71%


==========================================

LP(loose-passive)

Total Players: 162
Hands: 24,643
Hours: 416.18
BB/100 1.72
VPIP: 37.13
PFR: 2.82
W$SD: 51.17
Winners: 102 or 53.68%
Losers: 88 or 46.32%

=================================================

TAG(tight-aggressive)

Total Players: 164
Hands: 31,264
Hours: 519.00
BB/100 3.04
VPIP: 18.28
PFR: 8.30
W$SD: 56.44
Winners: 114 or 57.29%
Losers: 85 or 42.71%
=================================================

So, a third of the fish are winning, and so are a third of the maniacs. The tight-passives are losing, but not by much, in terms of how many players are winning vs. losing, but winning ever so slightly in terms of bets won. Figuring the high rakes on low-limit games, they're doing better than these figures show, and in higher limit games where the rake becomes trivial, it's possible these figures would be different. Loose-passives are winning slightly more often as tight-passive, but at about 2 and a half times the rate TP's are.

Loose aggressives are winning 2/3 of the time, and at a very solid 3.58 BB/100 rate. Aggression clearly beats being either a loose or tight passive, it would seem.

Tight-aggressives, though, are actually winning players dramatically less often than loose-aggressives, but still win more often than either TP's or LP's. Their BB/100 won of 3.05 confirms the value of aggression over passivity, but is still this type of play at these levels is being rewarded less than loose aggression.

It looks like both looseness and aggressiveness are rewarded, and loose-aggressiveness rewarded most of all, at these levels.

This is only a roughly 47k played-hands database, with 438k hands in total. It will be interesting to see whether bigger databases echo these figures, and how much they differ from the databases of people at higher levels, where it is likely both looseness and tightness -- indeed errors of any and every kind -- will be punished more thoroughly than they are at lower limits.

[DocMartin]

Interesting stuff. Looks like there will be plenty of $ for the taking. What are your PT requirements to be considered a maniac (I'm looking to add that icon to the filter).

[ctv1116]

This is why your analysis is flawed: if people on this board say you need 50K hands to know your win rate, then what makes you think 100 hands will determine whether a player is a "winner" or a "loser"? Here's a simple exercise: Let's say that every player in your database is a -1BB/100 player with 15BB/100 standard deviation. Let all of them play 100 hands. How many of these players will be up for their 100 hands? The answer is about 40%, when in reality ALL of the players are by definition losing players.

[lefty rosen]

Maniacs can win if they slow down on the double bet round. Also many moves that would be considered maniac moves are winning plays.......

[daryn]

the "fish" are not winning.. they just think they are. and you do too.

[Cosimo]

Assume 1000 people that all play the same. Have each play 100 hands. The resulting data is 100,000 hands, enough to establish a winrate. The key here is all hundred play the same. The problem, though, is accurately establishing play patterns after just 100 hands. It could have been a strong run of cards for a tight-aggressive; or some loose-passives on tilt; etc.

So I'm going to agree that the analysis is flawed. I don't think that a consistent LAG could be a winner, especially not for 3BB/100h or more.

But on the other hand... put a LAG on a table full of weak-tight-passive players, and I think the LAG could take all their money. That could be what is inflating the LAG's numbers.

[Greg J]

Awesome post Blarg! Very informative. Thanks for taking the time!

EDIT: I think there might be a few methodological issues though, esp re: your sample size for maniacs, as well as the discreet categorization. Still, good thought provoking stuff.

[Blarg]

[ QUOTE ]
What are your PT requirements to be considered a maniac

[/ QUOTE ]

Woops OMG. I forgot to paste in my swiped criteria!

Jeez what a dope. Here we go.

[ QUOTE ]
LAG: >30 vp$ip >10 pfr
TAG: <25 vp$ip >6 pfr
LP: >30 vp$ip <6 pfr
TP: <25 vp$ip <6 pfr

... more than 50% vp$ip is a Fish
... and if the same person also raises preflop >15% he or she gets the Maniac icon.


[/ QUOTE ]

There we go. Sorry for the (my) confusion!

[dink]

I think the important thing you have highlighted here,

is that some of the fish win some of the time

even if they are losing players overall.

And this my friends is what brings em back time and time again. selective memory is a wonderful thing for us but terrible for the little fishes.

so I say good on you little fishes with your 33% winning sessions, keep on winning (a little bit)

Dink

[Blarg]

[ QUOTE ]
This is why your analysis is flawed

[/ QUOTE ]

It's not so much an analysis as some Pokertracker figures. I admit that the classification has a certain arbitrariness going in. You have to have a cut-off somewhere, and I chose 100 hands. Many people choose quite a bit less before they classify and assign icons in Pokertracker.

Don't take it as more than it's meant to be. It's interesting in the aggregate, the 400k-plus hand sample size. I did not carefully examine all 8000 players.