Sample size?

[golfboy7]

Been playing lots of the 10+1 NL SnG's on PP and was wondering about....

How large of a sample size should i have before I can draw any conclusions from the results?

100, 500, 1000, 5000, 10000?

[AaronBrown]

That depends on your definition of "conclusion." Even 100 hands tells you something, 10,000 hands isn't enough to draw a firm conclusion. But I would say something like 1,000 hands is enough to analyze seriously, looking for changes in your performance. But you will have runs of 1,000 hands with good and bad stats that will be nothing more than random fluctuations, you don't need to respond to every one of them.

Generally, my feeling is that it's better to overanalyze than under. If you were playing well but had some bad luck, and made an unnecessary change as a result, it's not a big loss. It's good to mix things up anyway. But if you're playing badly and refuse to change because you're sure it's just bad luck, you will lose a lot.

[golfboy7]

keep in mind this is a complete tournament that Im analizing.

Basically I set up an excel spreadsheet that gives me the % of time I will finish in any given spot in a 10 man sit and go at a given level of buy in.

it also will give me the % of time that I have the best/worst hand when I went all in and lost...

more stats than just that, such as my ROI and so forth, but I'm just wondering how many tournaments should i play in before I decide if playing them has + EV or not. right now i'm at 80 tournaments and have been placing in the money (top 3) about 40% of the time.

I'm thinking i'll play 1000 + of them before I make any real conclusions, just wondering what everyone else thinks.

[spaminator101]

id say 1000 is about right however i would start anilyzing at 500 seriously

[pzhon]

The standard deviation of the result of the tournament is about 1.7 buyins. After n tournaments, the standard deviation of your average result is about 1.7/squareroot(n) buyins.

[Maulik]

Anything less than 500 tournamets is meaningless. 1000 is a good starting place.

[pzhon]

[ QUOTE ]
Anything less than 500 tournamets is meaningless. 1000 is a good starting place.

[/ QUOTE ]
That's not right.

[ QUOTE ]
[ QUOTE ]
The standard deviation of the result of the tournament is about 1.7 buyins. After n tournaments, the standard deviation of your average result is about 1.7/squareroot(n) buyins.

[/ QUOTE ]

[/ QUOTE ]
After 100 tournaments, the standard deviation of your ROI is about 17%.

After 300 tournaments, it is 10%.

After 1200 tournaments, it is 5%.

There is no sudden change from insignificant to significant. You gradually increase your confidence in the accuracy of your results. At lower levels, it is often possible to conclude after a hundred tournaments that you are likely to be a winning player, or that you have a lot to learn from established winners.