wanted to know how many sng's you have to play before you can get an accurate win rate? because in about 140 sng's i have a about a 20 dollar profit, just want to know how many more sng's i have to play to get a good accurate rate. thanks people

You didn’t say what kind of tournament you play. Below is a table I developed for a book I’m working on that shows the sample size you need for a 9 man Sit-n-Go to be 90% confident you are a winning player.

		9-Man SNG.	Sample Size for 90%	Confidence	WR>0 vs. ROI
			ROI
ITM Placing Distrib.	10%	15%	20%	30%	40%
Uniform 1-1-1	350	162	92	42	24
Symmetrical 1-2-1	331	149	85	39	22
First Likely 3-2-1	417	189	109	50	29
Third Likely 1-2-3	275	124	70	31	17

So, if you were playing these kinds of tournaments, and finished first, second and third about equally, with a 15% ROI you would need about 162 tournments to be 90% confident of being a winner.

Collin Moshman said what you need ~ 5000 sng's before you can class your ROI. 140 sng's is nothing, just up-streak

This can not be mathematically accurate. It obviously has to do with your margin of error though. Maybe Colin meant that in order to be 99% confident that your ROI is within +-0.1% of recorded ROI you have to play that many. If you lower confidence level to 90% or even 95% and the ROI deviation to 1% instead of 0.1% the number is significantly lower.

Besides that maybe he was referring to sngos with 180 people which would be an entirely different senario from stts

i play single table sng's and its the only game i have a profit in. i have no profit in cash or tourneys but sng's i have but like that guy said i think i need 5k tourneys to determine but i only play maybe 10 -20 games per day so getting to 5k games would be impossible because by the time i come close my game could be different. i guess i can't say for sure if im profitable at sng's. Unless statmanhal is right. thanks everyone and cheers and gl at the tables

Hey, what's with the UNLESS?

Note the table I presented only considered 90% confidence and fairly good
ROI's.

If your ROI to date was, say, 5%, and you wanted to be 95% confident of being a winning player, the required sample size is 2,310 for equally distributed placings, much larger than what I showed in the table.

unless means except and if your joking then ignore and thanks for your help.

Could you elaborate a little on the equally distributed placings?
Does that mean if I achieve my ROI with a higher% 1st places -> the sample goes up,
While same ROI with higher 3rd places -> sample sizes goes down?

Yes, the required sample size varies with how a winner usully finishes.

Sample size needed depends on the variance of winnings. I looked at four placing distributions when In The Money for 3 placings: Uniform 1-1-1 (all places equally likely) Symmetrical ( 1-2-1), First likely (3-2-1) and Third likely (1-2-3). Of course there is a fourth possibility and that’s out of the money.

The highest variance is First Likely where I assumed winning first is 3 times as likely as finishing third, given you are ITM. The reason it has the highest variance is because when it wins it most often wins the highest amount and that means it has the largest dispersion from not winning at all.

Finishing third most often has the smallest deviation from not winning.

Hope this helps clarify.

That's actually a really cool method to estimate samples.

So for a 6-Max Hyper KO Game with equal number of 1st and 2nd and ~8% ROI, what should be the needed sample size? Or have you done this for 9 max only?

I can easily do it with the program I wrote. What is the prize pool buy-in and fee and what is the prize structure?
e.g. 9 + 1 and 65% 35%

I assumed the buy in was 10 + 1 and the prize structure was 65%, 35%. The results were that a sample size of 530 is required to be 90% confident you are a winner. Click the following link to show a summary of the program's results