A quick poker tracker question....

[RustedCorpse]

Just a quick question:

What is the min sample size where you would start evaluating leaks in your game?

[Webster]

a few thousand - however it depends on what you are looking at. Position? Pre flop?

The more you look at the numbers the greater understanding you will have of your game.

HOWEVER - I think it's main use is as offense and not defense. I like knowing my opponents.

[Tigerscott]

I would guess at around 2000 hands. If your a consistant player you can start around 1000.

[DeeJ]

It is very good for showing loose opponents, this is a quick win.

For identifying leaks this is harder and requires more hands. For example, you can play 1000 hands and not get certain suited connectors in position any times at all. You can easily only get some pairs less than 3 times over 1000 hands. Basically you pick up specific pairs (eg KK) once every 200 hands or so.

You may be able to spot leaks with offsuited cards eg AK AQ with only 1000 hands tho as you are getting these 1% of the time and will likely have double figure samples to look at.

[dfscott]

Wow- I'm surprised at the responses to this question. I didn't count any of my statistics as valid until I accumulated well over 5000 hands. Even now, with well over 10,000 hands, I'm looking at overall performance, such as BB/100 or BB/hour. I can't look at how valuable specific hands (such as JJ or KTs are), since I've only gotten them less then 50 times (except for maybe the offsuit hands), hardly enough to be statistically signficant. And trying to evaluate hands played from a particular position is completely undoable with this small amount of data.

I agree that opponent evaluation is the best immeidate benefit. Being able to export player statistics into the notes and using them during actual play is invaluable.

[steamboatin]

At the risk of being nosy, how many hours a day do you guys spend playing poker? How many hours does it take to play 2,000 hands?

[BRMan]

My first post so I apologize if this is a silly question but this is something that I have been wondering regarding this number of hand histories in PT question.

Almost every post I've seen mentions that you need at least 10-15K hands in PT before you can start to see some trends. At the same time, many people say they use PT to get a read on other players. I've got 16K hand histories in PT but the most hands I've played with any one particular player is around 500 hands and most of them are well below 500 hands. In 16K hands, I've only got about 30 people that I've seen for more than 200 hands.

My question is, if I can't trust my stats until I've got 10-15K hands in PT, how can I trust the stats that I have on a guy that I've only seen for 100 hands?

[dfscott]

I'm married with two small children and a full-time job, so I don't play for long periods of time like some people. I play about 4-5 nights a week after the kids go to bed, usually for 2-3 hours at a time. At the .05/.10 limit where I started, I was averaging just over 50 hands/hour (this seems pretty typical). After a few weeks, I moved up to 2 and finally 3 tables/hour (I still go back and forth between 2 and 3, depending on how tough the game is), so I end up seeing between 100 and 150 hands/hour of play.

You can do the math, but for reference, I hit 10,000 hands after about 2 months of play.

[dfscott]

The stuff you're looking for in other players is much less refined that what you're looking for in yourself. You don't care if he's profitable playing AJo or if folding too much on the river. All you care about is how many pots he gets into and how often he raises. You only need about 50-100 hands to tell how many pots he gets in and how agressive they are (that's only about an hour or so at the table with someone). That way, when he bets, you can decide if it's worth chasing your flush draw or if your TPTK is still good.

For example, here's a guy from my database:

Hands: 108
VP$IP: 89.91
VP$IP from SB: 100.00
Agression Factor: 0.19
Pre-flop raise: 0.93

Ok, so what can we tell about this guy? Well, he plays just about any two cards and hardly ever raises. This guy is a dream -- he'll chase anything with junk. If he's in, you wanna be in because he's throwing money away.

Interestingly, his BB/100 hands is 38.24. However, this isn't really significant over this many hands and it's actually a great thing. He's winning enough to think that this strategy will work long-term, when you and I know it won't. There is always the possibility that he's a great player and just got an incredible run of cards, but a quick look at some of his hands confirms that's not the case. Looking at one of his games, I see that he called a pre-flop UTG raiser from the button with T2o.

Here's another player:

Hands: 131
VP$IP: 12.98
VP$IP from SB: 18.75
Agression Factor: 1.35
Pre-flop raise: 3.05

If this guy's in for a raise, I'm folding without a great hand. He plays very few hands and plays them hard, implying that he only plays premium hands. A quick look at the few hands he's played (only 12) shows that he's played AA 3 times. The hands he shows down with are flushes, straights, trips, etc. If you end up at the river with this guy, you better have the goods.

Notice that you can get this same information about yourself with only 100 hands, but you already know that you're tight and agressive, right? (At least you should be!) The data you're interested in about yourself are the more subtle items that require a much bigger sample size.

Hope this answers your question.

[O Doyle Rules]

[ QUOTE ]


My question is, if I can't trust my stats until I've got 10-15K hands in PT, how can I trust the stats that I have on a guy that I've only seen for 100 hands?

[/ QUOTE ]

You can't. See my post concerning Session win percentage and you can see a big swing in win rate from a block of 96k hands to another block of 204k hands. In my opinion, even 15k hands are pretty meaningless in the grand scheme of things.

[BRMan]

That is very helpful, dfscott. It's funny how something seems kinda obvious once someone explains it to you. /images/graemlins/smirk.gif

I have had hot streaks where the cards were falling and my VP$IP got up over 30 but those hot streaks have never lasted more than 100 to 150 hands or so. So generally, it may be safe to assume (at least until he shows you differently) that someone with a VP$IP that high over 200 hands is pretty loose. I think I'll just use a filter of 250 hands when exporting player notes in PT.

Thanks for the help.

[BRMan]

Thanks for the reply, O Doyle.

Having played only 16K hands and having done okay (2.22 BB/100 hands), I'm hoping that my 16K hands actually do mean something ... /images/graemlins/grin.gif ... but from what I've seen, I could see how it might not.

I think for me, the important point is that I don't need quite the level of detail to get a decent read of other players that I need to get a decent read of my play.

[dfscott]

[ QUOTE ]
That is very helpful, dfscott. It's funny how something seems kinda obvious once someone explains it to you. /images/graemlins/smirk.gif

[/ QUOTE ]

No problem. PT can be very overwhelming with the amount of data it contains. It takes a while to learn where the important stuff is. I got lots of help in the 2+2 forums.

[ QUOTE ]
I have had hot streaks where the cards were falling and my VP$IP got up over 30 but those hot streaks have never lasted more than 100 to 150 hands or so. So generally, it may be safe to assume (at least until he shows you differently) that someone with a VP$IP that high over 200 hands is pretty loose. I think I'll just use a filter of 250 hands when exporting player notes in PT.

Thanks for the help.

[/ QUOTE ]

250 is a lot of hands to play with someone. I've got over 10000 hands played, and only 17 people with over 250 hands (this is at PS). At a smaller site, it might be different.

I think 100 is probably a good bet -- you'll get enough to get a feel for the player but not so many that you don't get any to export.

[BRMan]

Taking a look at it, I've got 16500 hands total (all on PP)and I've got 33 people that I have at least 200 hands with. I think I'll drop my filter down to 150 to get a bigger list of players. Thanks again for your help.