Positional Stat Question

[rippyshoes]

I have been reviewing PokerTracker and I have a weird question. I am a significant winner at 50max PLO in every position but the BB and the Sb, and I was wondering if this is similar to anyone else or if it might just be a leak of mine? The net amount is -255 in the bb and -121 in the sb, but the Diff w/o blind I'm winning. I noticed that I was playing too many hands for a while from the blinds and I'm beginning to work on fixing that but I'm not sure I'm just wondering if anyone else has something similar to this? Also this is over 11,000 hands so it is a reasonably large sample size.Thanks for your help.

[piiop]

[ QUOTE ]
but the Diff w/o blind I'm winning

[/ QUOTE ]

hmm that is strange.

[RoundTower]

I'm pretty sure a) almost everyone is like this b) 11k is not a good sample size.

[rippyshoes]

So at what point can I consider it a reasonable sample size? 50k? I always thought that > 10k was the start of a reasonable sample?

[Silent A]

Remember, most of your hands are not in the blinds. You need a significant sample for each position to make conclusions about your win rate in each position.

If you play full ring, 10K hands might only be a bit over 1K hands in each blind.

As far as your original question goes, being a significant loser in the BB/SB is standard because you have poor position and you have to post a bet with every hand. In the BB you're forced to bet 1/2 ptBB per hand. That's 50 ptBB/100 hands, a massive set back to over come. However, if you neglect the forced bets, this disadvantage "disappears" and now you have a huge advantage because of all the "free"/"discounted" flops you get to see.

[pete fabrizio]

[ QUOTE ]
So at what point can I consider it a reasonable sample size? 50k? I always thought that > 10k was the start of a reasonable sample?

[/ QUOTE ]

How big your sample needs to be to reach a certain level of precision obviously depends on the stat you are trying you examine. E.g., to get an accurate VPIP within 1% you probably only need a couple thousand hands. To get an accurate winrate within 1BB you may need a couple hundred thousand.

[Troll_Inc]

[ QUOTE ]
As far as your original question goes, being a significant loser in the BB/SB is standard because you have poor position and you have to post a bet with every hand. In the BB you're forced to bet 1/2 ptBB per hand. That's 50 ptBB/100 hands, a massive set back to over come. However, if you neglect the forced bets, this disadvantage "disappears" and now you have a huge advantage because of all the "free"/"discounted" flops you get to see.

[/ QUOTE ]

This is a very good clarification.

What I would add is that I think you should completely ignore the "Net Amount" for the blinds, and only consider the next column over, which is "Diff w/o Blind". Once you post the blind, it is not your money. However, don't fall into the trap of playing bad hands from the SB and getting yourself into trouble on the flop. Have a clear idea of what exact hand you want to hit and how you will play it at your table. I think a good basic strategy from the blinds is to limp when you have hands that can make the nuts on the flop. Then checkraise or bet out based on the type of table you are at.

[Troll_Inc]

[ QUOTE ]
How big your sample needs to be to reach a certain level of precision obviously depends on the stat you are trying you examine. E.g., to get an accurate VPIP within 1% you probably only need a couple thousand hands. To get an accurate winrate within 1BB you may need a couple hundred thousand.

[/ QUOTE ]

This is all true, but the general idea you read on here about 10k "not being enough hands" to know about winrate is preposterous.

Even with 100,000 hands, you'll never know your true winrate within 1 ptbb/100 because both you and your game conditions will have changed from the 1st hand until the 100,000th hand.

Conversely, if you are trying out a new strategy or a new game, etc., then comparing just 5,000 chunks of data can be very enlightening.