Check out our brand new Local Poker Communities! Get updates and interact with poker players in your area.
Visit the United States Poker Community | Visit the California Poker Community | Read more about the Launch of P5s Local

I need a statistics Expert!!!

Reply to Thread
  1. AawwNutz AawwNutz is offline

    Lewisville, TX
    United States

    Posts: 1,218
    Joined: Feb 05

    Aug 4th, 2007 (8/4/2007 1:12am)

    I need some help from a braniac - statistics.

    Here's the deal:

    I have a database of 40,000 hands and have pulled out 334 hands where I was allin for my tournament life. I have taken my hole cards versus my opponent(s) and computed my expected winning% as of the moment all the money went in.

    Much to my dismay, my projected winning % only came out to 42.05%. Obviously, I need to play better in allin situations (or adjust my short-stack strategy).

    I need help with what would be an expected normal deviation range from this 42.05% target. +/- 2%, 3%, 4%??? etc., etc... Should I expect to see something from 38-46%??? Please help me come up with a number.

    I'll let everyone know my actual results AFTER the discussion about normal ranges.

    As for all of you "sample size" nay-sayers, the +/- values accurately account for the limited sample size. If I had a million hands, my margin of error would be smaller than it would be for just these 334.

    Any help would be appreciated.
     
  2. AFink93 AFink93 is online now

    Norwich, CT
    United States

    Posts: 14,381
    Joined: Mar 06

    Aug 4th, 2007 (8/4/2007 1:26am) in reply to AawwNutz

    Its such a small sample to be statistically relevant I think. I think its possible to be 10% deviant either way in this sample.
  3. agriffrod agriffrod is offline 2010 WSOP #3 FT

    Posts: 7,044
    Joined: Dec 05

    Aug 4th, 2007 (8/4/2007 1:28am) in reply to AFink93

    anyone else think its ironic AFink commented?
     
  4. BerkeleyBoss BerkeleyBoss is offline

    Berkeley, CA
    United States

    Posts: 850
    Joined: May 06

    Aug 4th, 2007 (8/4/2007 2:40am) in reply to agriffrod

    i agree, with only 334 hands, you could easily be a victim or variance, and really be more like 50%

    also I think it might pay to seperate the times you're super short (shoving any 2), the times when you have about 10 bbs (only 1 move), and teh times you're deeper. You'd def need a HUGE sample size for this. But I just don't think you're going to get much meaningful info out of the stats now
     
  5. BarryGS BarryGS is offline

    Posts: 295
    Joined: Dec 05

    Aug 4th, 2007 (8/4/2007 3:45am) in reply to AawwNutz

    You would need the actual data to determine the exact confince interal. I would guess around plus minus 3% would be about 95% accurate. 334 hands would not be enough to be accurate but 334 all in are plenty.

    Given the Gap concept, I would expect most peoples expectation would be less than 50%. You more than make up for it with your fold equity, which I am sure you already know. Now calling all in for you life you be well over 50%, IMO.
  6. apestyles apestyles is offline Moderator

    Vancouver
    Canada

    Posts: 5,087
    Joined: Sep 05

    Aug 4th, 2007 (8/4/2007 3:55am) in reply to AawwNutz

    wow that high? U need to add the chips youve won when they fold to get a real picture..
     282 12 

    apestyles is an instructor at PocketFives Training . To get more of his advice and to watch his training videos, click here.

  7. Zota15 Zota15 is offline

    Orlando, FL
    United States

    Posts: 1,471
    Joined: Jan 06

    Aug 4th, 2007 (8/4/2007 4:01am) in reply to apestyles

    Yea Ape is right....if you arent getting your money in as a shortstack as an underdog more often than favorite...you are playing too tight.
  8. AawwNutz AawwNutz is offline

    Lewisville, TX
    United States

    Posts: 1,218
    Joined: Feb 05

    Aug 4th, 2007 (8/4/2007 10:45am) in reply to AawwNutz

    A couple of things:

    I'm not really worried about the actual meaning of 42.05% because that's my equity after the call of the allin. The number of times that I pushed allin and got them to fold pushes my number up to a very nice percent. In other words, the fold equity added to the hand equity is really nice.

    I'm more interested in the actual equity achieved. If I'm supposed to win 42.05% of those hands, what would be a reasonable range to expect.

    BTW, of my 40,000 hand database, I was allin and called for only 334 hands. That is the reason my sample is so small. It took me 1 year to get 40,000 hands, so I don't really have the option to use a much larger sample.

    Thanks for all of the feedback thus far.
     
    Thread Starter
  9. Frank1The1Tank Frank1The1Tank is offline

    Posts: 3,153
    Joined: Apr 07

    Aug 4th, 2007 (8/4/2007 11:05am) in reply to AawwNutz

    Also just a note, just because your behind slightly according to this sample, doesn't mean its not a +EV play. The odds of picking up the blinds are very good too, and usually for a short stack, this increases your stack by a meaningful amount. So the EV you're getting on this could be more like 60% which is of course, profitable in the long run.
     
  10. Enigma1157 Enigma1157 is offline

    Las Vegas, NV
    United States

    Posts: 1,003
    Joined: Nov 05

    Aug 4th, 2007 (8/4/2007 11:08am) in reply to AawwNutz

    What kind of hands are you shoving with to only win at 42% projected wins ? Typical shove hands of AA and KK and QQ would put you at 70% to 80%.....

    ..are you shoving with 5 - 8 or what ?

    ..also....stack sizes and button position and blind levels might be MORE important than the actual strength of your cards....so I don't think any kind of percentage you come up with is gonna have any real meaning......there are simply too many other important variables involved that you haven't quantified.....
  11. TY57RBA TY57RBA is offline

    United States

    Posts: 9
    Joined: Jul 06

    Aug 4th, 2007 (8/4/2007 11:22am) in reply to AawwNutz

    This isn't a huge stats problem but you're right to be wondering about variance. If you just run an average and standard deviation on your 334 hands the variance is going to be higher than if you do the same say for just big stack allins, or small stack allins, or early in tournament, or late in a tournament, and of course heads-up. The challenge is always how do you account for the variance in your data. The more variables you can control the easier it is to interpret your results. That said, there are sometimes really interesting stories buried in the variance; for example, if you find situations with high variance, say 2 standard deviations from the mean compared to situations with low variance, say less than 1 standard deviation from the mean, there's probably a lesson to be learned - assuming sample size is sufficient. You can even take it a step further and compare different means by calculating the standard error of the mean instead of the standard deviation. About 95 percent of the time your mean will fall within two standard errors so if you want to compare means, if they don't overlap at the 95 percent level they can be considered "statistically" significantly different. Again, sample size is important and there are things that you can do to assess sample size vs variance but that's probably not necessary. The stats I described above are fairly basic (you can find plug and play stuff on the net or a basic handheld calulator) and I'd hate to think that you need to go deeper, say with ANOVA or worse, because the kinds of differences you're looking for are likely going to be visible using the "intra-ocular" test: that is, the differences are so apparent that they hit you right between the eyes. And a disclaimer: I'm not a stats expert, but I use'm now and then, work with stats experts occasionally, and took the stuff in college long before most of you P5 members were born.
Reply to Thread

Similar Threads