Quote:
Originally Posted by Uhrenknecht
what's a good sample in 180s / turbos or not ?
I think it is a tricky question and a little vague.
It will depend on how you interrogate the sample and how accurate you want or need to be.
eg. if you pass 100 full game histories over to a good 180 coach they will be able to give you a decent estimate of your roi over the near future.
If you just look purely at the results of 100 games it is pretty useless for any accuracy at predicting your roi (it is better than no results but very, very wide)
So the more information you analyse along with the simple results the better.
If you just want to use tournament results then you do need a massive amount for 180s to get some accuracy. As a rule of thumb if you are happy at thinking 2,000 games is a decent sample to estimate your roi for 9 seat STTs then to get the equivalent confidence in your roi you need 10x the amout of games so 20,000 games.
In 180s after 8k games your spread of roi will be roughly roi +/-10% for a 90% confidence. (In STTs you would only need about 800 games for this accuracy).
Every time you use 4x the number of games you half the width of the spread so 8k x 4 = 24k game sample you can have a spread of +/- 5% with 90% confidence.
As you can see just waitng for enough results to provide feedback to your game is not likely fast enough and so you would have to aquire some confidence in your game from other means like study, coaching, forums etc.
As far as the difference between turbo and normal speed games goes there is no difference in the built in variance we see at both types just the difference that standard games are longer and easier for skill to win through and so higher possible roi's. The spread width due to the variance is the same, still roughly +/- 10% for 8k games.
It is best to judge by profit per hour and enjoyment factor rather than any roi variance. If you can sustain higher roi's you get less downswings.