How many hands needed to identify a leak?

[Warik]

I've been slowly building my PT database and am noticing that I have a few starting hands that are losing money when they should probably be making money. KQo being the worst culprit.

Short term of course means nothing here, but when does one have enough hands to determine that hand X is a leak and needs to be plugged?

I have 3,000 hands in my DB and have had KQo 25 times for a total of 36.25BB.

Is it still to early to tell if it's bad play or bad flops? If I have enough info to determine that this hand is a leak, how should I proceed in plugging it?

This is the first time I'm attempting to change my playstyle to fix a problem, so bear with me. /images/graemlins/smile.gif

[pudley4]

Go through each of the KQ hands, especially the ones where you lost big.

Did you lose to a dominating hand? (AK or AQ, KK or QQ)

Did you get outdrawn on the river by a lesser hand?

Did you play the hand too fast or too slow?

Post the questionable ones here

[stoxtrader]

I think well over 500 hands with a particular holding is needed to even begin to get an idea about whether you play that hand correctly purely from the statistical results.

I agree that using the replay to feature to look at all those particular hands is a great idea, it'll take you about 5 minutes.

FWIW - I win .48BB/hand with KQo and lose .32 BB/hand with KQS (over 299 and 98 hands accordingly). MY play on the KQs hands has been fine, and I've simply (correctly) drawn to a number of flushes that haven't hit, and been sucked out on an extra maybe 3-4 times over what you would expect. i.e. say, on average, a 5 outer will beat your top pair say 5/46 or about 11% of the time on the river, I'd say I've been beat say 16% of the time - in small samples it is very very easy for this type of thing to skew the data.

best,
nick.