Quote:
Originally Posted by Rusty Nails
Outstanding post Mpethy. I'm always amazed when I talk to someone and they don't know how to use HEM and/or don't bother spending a couple hours a month running some filters to review their play.
I'm interested in hearing your opinions on how sample size affects your stats. For example, in my review last month I ran some of the filters you mentioned when I discovered I was getting destroyed in my blind play this year. The problem was that there didn't seem to be a clear distinction in which hands were losers and which ones were winners when I looked at w/rs. KJs may be a big loser for me, but KTs was a winner. 66 was a loser but 55 was very profitable. I had to review the actual hands and try to adjust the wrs for luck factors. Thoughts?
You definitely have to control for variance when you are looking at small samples of hands.
Your process is sound. It is the same one I use when reviewing databases.
Just one point you might try: when you run these filters, select groups of hands to increase your sample size. In the examples you gave, being down with 66 and being up with 55 looks like variance. But what is your average of the two hands?
Same thing for KJ and KT; logically one explanation for being down with KJ and up with KT is that you might get attached a bit more to KJ and lose hands that average 1.8 streets of betting, whereas KT is easier to pitch when you are dominated, and might cost you only 1.4 streets. So if I am looking at win rates for those hands, the first thing I would do would be to look at my losses, and see if I could see such a pattern of calling two streets with KJ and calling less than that with KT.
More likely is that it is just variance, so, again, I would look at my average for those two hands to determine whether I was doing ok with them.
Here are some common groupings I use when filtering for calling raises from the blinds:
A9s-A2s; 54s-98s and 64s-T8s; J9s-KJs and T9s-QJs; ATs, AJs, AJo, AQs, AQo and KQo and KQs; 22-77; 88 and 99; TT and JJ.
When going through a group of hands, use the holecards report in HEM; this allows you to look at your win rate for each hand in the group (PT3 sorts this way automatically). In general, I am satisfied if the group has a positive win rate (the exact win rate that satisfies me varies by group).
I also ask:
Does my win rate correlate with hand strength within the group? How strong does the relationship appear?
Is there a logical explanation for my win rate not varying with hand strength, or is it likely variance?
Did I suffer an unreasonable number of coolers/set-ups/bad beats?
Did I NOT suffer any coolers/set-ups/bad beats?
What is my adjusted win rate if I control for the presence or absence of coolers/set-ups and bad beats by adding or subtracting an appropriate number?
For pocket pairs, I also check my flop a set % and adjust my win rate upward or downward by the average win for a set out of the blinds.