Tournement Sample Size

[toxzen]

What is a good sample size to have to show you're true ROI in tournements with an average field of 500? I've tried searching for an answer with no luck.

[DWarrior]

1000 or something, MTT stickies?

[toxzen]

Checked the stickies, didn't see it. Is that all tournies or for an individual tournie? Like does that mean I have to play the 100+9R 1000 times to see if I'm a winner or not in that particular tournement?

[oddsock]

[ QUOTE ]
Like does that mean I have to play the 100+9R 1000 times to see if I'm a winner or not in that particular tournement?

[/ QUOTE ]

No, you are a winner if you are ahead using any sample size and if you can consistently do well in these games then you may consider that you are simply a winner regardless of the number of games played.

However the smaller the sample size, the more likely your results will be skewed due to short term variance - especially in MTTs. You can play 50 500 player tournaments and fail to money in any of them and win the 51st one thus moving your ROI from 0% to around 270% in one go (assuming top prize is 30% of prize pool and including entry fee).

Any sample size which allows this sort of wild fluctuation is obviously statistically too small which is why 1000 has been quoted above. I'm pretty sure from a mathematical point of view that an acceptable sample size would be much larger than that. Realistically, don't even think about it ROI until you've played 500. Seriously.

I read in the MTT boards that an acceptable ROI is to stay even until hitting a big payoff!

[toxzen]

Thanks for the reply!