Re: SNG statistics
[ QUOTE ]
[ QUOTE ]
Well I have no idea what that crazy math means, but I can say that I'm a winning player and I can easily go over 150 sngs with a negative roi or a ridiculously inflated one.
[/ QUOTE ]
Yeah, but if you split up all your SnGs in 150-SnGs-chunks, 95% of the chuncks' ROI should be between 13% and 25% if your true ROI is 19%. If you've played thousands of SnGs, of course many of the chuncks will have a ROI that is below or above that confidence interval.
[/ QUOTE ]
Nonsense. This is true if you are measuring a physical process such as counting coin flips or weighing widgets. These things have independent observations. The same is not true when attempting to quantify human behaviour.
The result of a SnG is not independent of those immediately preceeding it. I hereby posit the STTF First Law of SnG Dynamics: the law of inertia. A player on a heater tends to remain on a heater, and a player on a downswing tends to go on tilt. In other words, long runs of winning or losing tend to be self-reenforcing, and will be greater in magnitude and duration than runs of heads versus tails in a coin flipping games.
A sample size of 150 will certainly give you a statistically significant result to decide if a coin is unbiased. Even if the OP had a sample size of 150 (and he doesn't), empirical evidence here indicates that would not be large enough to average out the psychological effect and the resultant long swings up and down.
Experienced students of the game here have concluded that a sample size of 1000 will average out the mind swings in the game and give a good estimate of true skills. This is a subjective analysis that you will not find in any stat text, but one that stands the test of STTF time.
|