Identifying changes in play statistically NLTRN

[TNixon]

Before I take on the monumental task of going over a few hundred hand histories to examine my play, I'm trying to approach things from a high level and see if there's any way I can statistically determine if I'm simply playing worse.

There are a couple fairly simple metrics that I've run.

The first one is an analysis of all-in situations. What I'm calculating is basically EV/chip, or how much of each chip in the pot I expect to win based on my hands compared to my opponent's hands at the time of the all-in.

The thought behind this first statistic is that if I were making much worse calls, or pushing in worse spots, my Ev/chip should go down when I compare one set of games (when I was winning) to the other (evil nasty losing streak).

The second is a similar analysis of pots that go to showdown. I'm doing a similar calculation on the money that goes in on each street, using the equity calculations for our actual hands compared at that time. For example, if I start with AhAc, and my opponent starts with JhJc, then I get 81.8% credit for any chips that go in preflop. If we don't get all-in preflop, and the flop is Jd 8s 2s, then I only get 8.6% credit for any chips that go in on the flop.

If I'm just generally putting money in bad in situations where neither one of is is all-in, then I should see this metric (again measured in EV/chip) go down.

I actually expected to see both of these numbers drop at least a little bit, due to tilt and frustration calls throughout my bad streak, but they both actually increased slightly, so if I have made negative changes to my game, it's not in an area that can be measured by one of these two metrics.

Are there any other metrics I can look at to try to get some idea of what I should be looking for when I'm digging through hand histories?

Another obvious possibility is that I might be folding away many more chips than I was previously, but I haven't yet figured out a meaningful way to try to measure that.

[teteatot]

I've heard people talk about going about it this way a lot, I think there's even a PT addon that does it for you, PokerEV or something like that.

I think it's incorrect though, since we put villains on ranges and not specific hands. For example, using that method, if you lose a match due to getting it in pf with KK vs AA, it will be a hugely negative bump, but it's silly to think you played poorly.

Basically, ev/chip thinking chalks variance up to primarily how the community cards fall, and I disagree with that, I think bad runs primarily come from being in an inordinate amount of nasty spots where villain has the top of his range.

[TNixon]

Given enough hands, though, it should all average out, shouldn't it?

I'm not necessarily trying to figure out if I'm playing *poorly* using these numbers directly. I'm trying to figure out if I played significantly *differently* in two time periods, and if so, where those differences might be, so I know what to look for when going over a few hundred matches.

Using just these two metrics (which admittedly is a far cry from being comprehensive), it looks like there are few differences at all between the two periods. The ev/chip numbers are very similar, and even things like average chips risked per hand are very close.

[soop]

How about looking at hero calls? That was what I found was doing me in when I moved up in stakes.

Check to see how often you called bets on the river and what you win percentage was.

[LordMushroom2]

You seem too obsessed with math, TNixon. I am like you. I like cold hard mathematical numbers. But when we do calculations and learn from them, the gain we get from those also have a cost, which is what we didn´t learn from playing because we chose to do math instead.

There are definately times when choosing to do math instead of playing is better, and most players do too little math. But you seem to do too much. I do the optimal amount, of course. [img]/images/graemlins/wink.gif[/img]

This is meant as friendly advice, not as an insult or something.