re: PT stats, how many hands needed to see patterns?

[questions]

I've had PT since last year, and was curious to know how many hands I need in my database before I can start seeing how my stats compare with a perfectly random distribution? Specifically, I have 5000 hands in my database on FTP and while my pocket eights have won every single time of the 12 times I've had them, my kings have won 65% of the time. That seems odd.

[PBJaxx]

Um...A LOT more than 5k hands. At this rate, you might have an idea in 20 years.

[questions]

Okay. So I should get back to you?

[oracle3001]

Different stats require different numbers of hands before they converge, but you are going to need a lot more than 5k. For example, I think I read that about 100k hands is about what you need to get a fairly accurate idea of your real win rate.

[tagtastic]

50-100k hands to get a general idea on how you're playing (ie: "how am I doing with suited connectors" or "am I showing a profit cold calling preflop on the button").

Many, many more to get meaningful data on specific situations (ie: "do I show a profit calling a raise with 56s in the big blind"). Like 500k+ hands easily, probably more.

In NLHE the big pots can swing the stats for specific situations way out of whack, so it takes a massive sample size before the numbers for stuff like specific hands starts to be meaningful. If you can look through your data and ignore some of the rare large pots, you could get a rough idea with less hands, but it'll be just that: a rough, inaccurate estimate.

For limit the sample sizes for different situations are probably smaller, but more general stats (like winrate) probably require a larger sample.

[MicroBob]

[ QUOTE ]
my pocket eights have won every single time of the 12 times I've had them, my kings have won 65% of the time.

[/ QUOTE ]


for 5,000 hands this isn't odd at all.

It would be more odd if you did not have anything looking like this in such a short sample-size.


Do you seriously think you found something here?
FTP is setting up so that only you are magically winning a ton more with 88 than you are with KK?
Or they set things up so that everyone on the site wins more with 88 than KK? Because somehow this will yield extra profits for FTP?

Even if they were going to rig the deck to supposedly generate more profits do you really think they would just make it so obvious for 88 to always be the nuts and KK to get cracked a ton...and that nobody had noticed something like this before you.


You need to be asking what the point of all this would be in the first place.

The shorter answer is at the beginning of my post thought. No, this is not a big deal in the least.

[drunknpokerfreak]

50-100,000 hands to get close. 200-300,000 to get real close.

[DMoogle]

It depends SOOOOOO much on which stat you're looking at. For example, you're probably be within 1% of your true VP$IP over 10k hands, but if you want to look at statistics for individual hole cards (e.g. A3o, 67s, etc.) then you'd probably need a database of literally millions of hands.

[MicroBob]

I think you're within 1% of your true VPIP in WAY less than 10k hands.

For most of my opponents, they hit a fairly reasonable VPIP after only 200 hands with them and it very rarely deviates by more than a couple points after that from my limited observation.

If I have 1000 hands on someone then the VPIP I have on them is pretty much going to be their true VPIP.

Most of the time you don't have to worry about it.
If a player is 18/12 after 200 hands then he is probably going to be 18/12 after 1000 hands and 18/12 after 10,000 hands (unless you get some hands with him with many more or fewer players than the first 200 hds).

This is a stat that seems to converge extremely quickly.

[questions]

[ QUOTE ]
Do you seriously think you found something here?
FTP is setting up so that only you are magically winning a ton more with 88 than you are with KK?
Or they set things up so that everyone on the site wins more with 88 than KK? Because somehow this will yield extra profits for FTP?

[/ QUOTE ]

Of COURSE that's what I think! Silly. [img]/images/graemlins/laugh.gif[/img]

Seriously, to someone like myself who is neither a rocket scientist nor a calculus professor, such stats look odd. Thus, my question.