Suitable Sample Sizes for Stats

    • WhiteWalker
      WhiteWalker
      Bronze
      Joined: 06.06.2015 Posts: 464
      :milord: Seeking Some Wisdom
      I am just wondering what you guys judge to be a reasonable judge of stats for players. I understand vpip/pfr/aggro can have a fair idea by ~50 hands. What I am really wanting to know is how you judge stats such as 4bet OP/ 4bet IP, steal, fold to steal, win at showdown, went to showdown etc. These stats are useful for river decisions and bluff catching but I do not want to focus on them. I do not currently use these stats but would like to consider there use for the future when I have a suitable sample size on the player pool here at the micros.

      What d'ya think?
  • 3 replies
    • Kyyberi
      Kyyberi
      Coach
      Coach
      Joined: 09.07.2010 Posts: 10,510
      At least in PT4 you can check the sample size for the stat by just hovering the mouse over it. Overall hands isn't goodish indicator cos someone with same hand numbers might have a lot less sample size in that one stat.

      Won@showdown for an example depends greatly on player's VPIP and postflop fishiness. Someone with 1K hands might have gone to showdown 200 times and someone else with 1K hands might have gone only 15. So always check the actual sample size.

      No one can tell where the borderline for "reliable" sample size is. If someone has folded to steal 66% it's a bit different if it's 2/3, 45/68 or 244/366. That's where you have to estimate the probabilities. Stats don't give you straight answers, they just give you information. But poker is a game of imperfect information, so you have to use your poker skills to use that information.

      If someone wins at showdown 25% (2/8), the stat just tells you that the opponent has won twice out of 8 possibilities. It doesn't tell you pot sizes, number of players in the showdown or the actual situations (bad calls/bluffs or setups). It tells you that it is probable that the opponent goes to showdowns with wider range. When the sample size goes up and the percentage stays low, it tells you that it is more and more probable.
    • Kyyberi
      Kyyberi
      Coach
      Coach
      Joined: 09.07.2010 Posts: 10,510
      And I would like to add that when people say that "under X hands the information is useless" that is nonsense. Even one hand gives you more information than 0 hands. If an opponent raises 3 out of 3 hands preflop, it is a bit more probable that he is not a nit. If someone folds 3 out of 3 hands, it is a bit more probable that he is not a monkey maniac. Even that tiny bit of probability can be used to increase your EV.
    • WhiteWalker
      WhiteWalker
      Bronze
      Joined: 06.06.2015 Posts: 464
      Cheers for the response kyyberi, its good to have a rough idea and I am beginning to attack weaker opponents who are too lose in EP when I am in LP much more thanks to the use of stats and my noticing of that, the stats back up what I feel is rite but it is good to have stats as proof of my reasonings.