Hi,

So I've noticed a pattern that I tend to run well in the first half of the month, and poorly for the second half. I'd like to work out what % there is that this would occur by chance. If there's not much chance, I have some posts in the Psychology forum to make.

Here is the data I have:

Thanks in advance, Punter

looks like a psychology problem to me or it could be the fact that you're running badly in the latter part of the month. Maybe the fact it's a new month causes people to be in line which is a style you exploit easily, and at the later part of the month people are less caring which could result in harder style to play against. I can definitely see the psychological side to your problem but am having a difficult time analyzing the probability question but it does look as if you're playing fewer hands in the latter part of the month.

No there is definitely a % that this could occur by chance. It depends. do you consider each month independent or the 2 halfs grouped as one?

In terms of hands, I've played a very similar number between the halves:

Assuming a break even player each 1/2 of the month there is a 50% chance you will be up and %50 chance you will be down with each 1/2 being independant of each other.

Now to your case: If I am reading your data right some months you are up early and down later so I assume you are talking about your total results.

If this is the case look at the 1st half and 2nd half of the month as independant events. I am assuming you play an equal number of hands each half of the month so the question will become where does 5760 and (261) over 125K hands (total hands/2) fit into your standard deviation.

Regaurdless of your concolusion about your performance I hope you make your post .

The right way to do this is a t-test. First, normalize by the # of hands to get $/hand. If we treat this as unpaired data (i.e. the 1st half of June is independent of the 2nd half of June) with unequal variance, then the t probability is 0.011. If we assume the data is paired (which I think is valid), then the t probability is 0.026.

Either way, unless your criterion is quite stringent, we have enough evidence to reject the null hypothesis that the disparity between the 1st half and 2nd half of the month is due to chance. Good luck figuring it out.

I would approach it this way.

You did better in the first half of 8 out of 12 months. Clearly that's well within the boundaries of normal variation, there's a 39% chance you will be this far or farther from 6.

Your absolute sizes in the first half of the month are 43% larger on average than those in the second half. This is presumably unrelated to your relative poker-playing ability in different halves of the month. Perhaps you play at higher limits or in livelier games in the first half of the month than the second, or maybe you play (that is, don't fold immediately) more hands.

Given that your play has more volatility in the first halves of the month, and assuming it's just chance that you happened to do better in 8 of the 12 first halves, the results are explained.

If the difference were something more likely to be the result of skill, say the results of two different players or your results at Hold'em and Seven Card Hi-Lo, then I wouldn't view the evidence so skeptically.

I'm not saying WHY the two halves of the month are different, but they clearly are. One thing to consider is the variance in $/hand. They're actually quite similar for the half1 and half2. The differential variance in $/half-month is due to the # of hands.

Clearly, the OP is more profitable in half1. The t-test is made for this, and it doesn't lie here.