I know this is a small sample but is this too odd?

[Pil Sung Do]

I'm having a bad run of cards, so I check my poker tracker stats.

In my last 1061 hands, I have received 136 T's. This is an average of 12.8% of all my hands.

During the same time frame, I have received KK, QQ and JJ a total of 7 times. (JJ never!)

Please help get over my paranoia

[CMonkey]

On average, one would expect to receive each pocket pair about 4-5 times over that number of hands. What do you mean by T's? (you received starting cards that had a ten in it?)

[Pil Sung Do]

What do you mean by T's? (you received starting cards that had a ten in it?)

Correct. The three hands that were TT, i counted the T twice.

[CMonkey]

Eh, you're getting somewhat shorted on starting cards but not ridiculously so. The small sample size probably has a lot to do with it.

[kelvin474]

You'd expect a Ten 1/13 of your cards if each card were equiprobable.

you have a ten 0.06409 of the time (136/(2*1061)).
The expected ten-frequency 1/13 is 0.07692.

The normal approximation to the binomial is reasonable here, so the actual ten-frequency is normal with

mean .07692
std. dev 0.0058

You can say that under the hypothesis that you are getting T's at a rate of 1/13, you are 2.2 standard deviations below the expected number. There is a 2.8% chance of this result or farther from expected in this size sample.

The thing about the pocket pairs: its true your getting those hands less often than average.

However, consider this point on each of the tests: The fact that you decided to ask now biases the test. If i plan to play 100,000 hands, there is a much better chance than 2.78% that i will be able to find a streak of 1,000 hands where I was 2.2 SD to the left on # of tens received.

The probability of pocket Jacks is 1/220 on any deal if each subset of 2 from a 52-card deck is equally likely. The chance of avoiding them for 1000 hands is about 1%.

However, if you play 1000 hands, you will almost def. be able to look at the results and say "oh my god i cant believe i never got hand X, or i only got it once or something". So, it sucks that its a good hand like JJ you havent got but you can't say much unless you assigned a high prior probability of cheating. Its especially difficult when you take into account the biased nature of the test.

Try this: play 1000 hands and predetermine what you're going to check for instead of picking out what looks odd after the sample. You may be able to reduce the sample size required for the same power/false alarm rate test using the sequential probability ratio test:

http://www.agrsci.dk/plb/bembi/africa/sampling/samp_spr.html

You could pipe your data from a PT database into a program that calculates and updates the P(too many or too few whatever event your measuring).

[Pil Sung Do]

My knowledge of statistics is not very good, so a lot of that went over my head.

I do agree that you can take any group of data and point out "flaws", "inconsistencies" and such, and that was one of the reasons I asked here.

Thank you for your answer, I think that you're telling me that I'm getting the short end, but not so terribly so that red flags should go up (just a bad run of luck).

I hate to say it this way, but this was all brought about because I made a small cashout and then watched as the cards turned liquid nitrogen on me (-197C).

[pokerrookie]

Good thing you are playing with Kems.