Our poker group has recently started tracking results of our regular games. I am curious to do some analysis of the results to support my thesis around who are good/ bad players, vs. who are plain running lucky/ unlucky.

For example, Player A has won total $2000 this month with 8 wins and 2 losses; while Player B has won total $2500, with 6 wins and 4 losses. Who is "better", and how do I quantify this? Would this be StDev? But StDev wouldnt differentiate between absolute winners and losers.

Furthermore, how do I adjust for the fact that some of us have played 10 times this month, while some others might have played just 3 times?

Thanks
Ankur

Sounds like the beginning of the end for your game. Trust me, the bad players have no clue how much they are losing and aren't going to like it when they find out, nor are they going to like others knowing.

As for the "who is better" argument, get back to me when you have 1,000 hours on each player. Confidence intervals are so wide at ten sessions that the stats are meaningless. Generally speaking though, the measurement of "how good" someone is in live poker is their winrate per hour.

For (a lot) more discussion on winrates, check out the winrates bankroll and finances thread in LLSNL.

I agree with G's "The beginning of the end" sentiment. Poker is one of those things where 90% of the players think they are above average. Clearly defining "losers" is a bad idea, esp when the point of a home game for many people isn't actually to make money, it's to play poker. (Even if they tell temselves otherwise.)

Thanks. Agree it's a bad idea to track scores publicly. One of the players decided unilaterally to maintain an excel and told everyone about it. Now everyone keeps asking him who is the biggest winner etc. I too feel this will slowly kill the game, or make it less fun at a minimum.

Well, if you bring up the "need 1,000 hours to tell the underlying stats from the effects of luck," you might get the bad players to keep going for a while hoping they're just on the downside of variance.

Mostly though, I'd tell that guy to knock it the **** off. Explain to him that he's tapping the tank, and get him to announce that it's too much work, and everyone who's interested can just keep their own stats. You can send him to this thread if he thinks you're just being paranoid.

You need a lot of hours to track yes. But in the mean time, calculate how much they win per game on average. But as others haave mentioned, its the beginning of the end to track results. Very bad idea to track.

Yes, bad idea.


There may be some assumptions, but if you had 10,000 hours on all the players, and there were no result skewing hands (like winning a BBJ), you could determine the player you thought was better. It might be different from the player I thought was better.

Player A has a lower win rate and lower variance. Probably won more sessions, was able to count on winning at least a little most weeks, maybe never got busted out early.

Player B has a higher win rate and higher variance. Probably won more overall chips, even though they might have had less winning nights, and possibly busted out a few times.

Without knowing complete results, most casual players would prefer to be Player A and think Player B was a donkey who occasionally got lucky.

Knowing the complete results, most players would prefer to be Player B. Assuming they aren't on a limited bankroll and can afford the bad nights.

Hi, revisiting this again. We now have around 20 months of stats, so around 200 games or over 1000 hours (not on each player, just total time the game ran). We track only final scores per game (not number of hours in each game).

Some of the metrics being used to rank include net wins, number of games played etc. Assuming winrate per hour is not possible to track, what other metrics would you recommend?

You haven’t clearly defined the parameters of what you’re measuring to determine who is “better.” You need to do that first before you decide what variables are pertinent. I would consider tracking hours and gross +/- to determine an hourly win rate and use that as the measure.

Regardless, you may want to consider adding intangibles to keep the fish in the game. Maybe a note about who is improving the most or who brings the most fun to the game. Maybe bluff frequencies or something. You want the fish to still feel like they’re contributing something to the game other than their money.

Was going to type my old response to op, until I saw this was bumped.

Still agree that the losing players aren't going to be around much, unless it's the equivalent of penny stakes and just a good night out having fun.

Winrate per session/night would be 2nd best, assuming there aren't major outliers of session times.

Again, I think the veteran players would continue to preach what a bad idea this is for the health of the game.