The Big Crunch

Hello Everybody,

I just finished coding all point-by-point summaries of all table tennis matches that I observed from the 2010 World Table Tennis Championship. Since I had to code the server, direction, court, and game progression (who won the point) for all 1,826 serves. This process took, as expected, a really long time. For the last month, I have been coding all these point-by-point summaries, which is the reason why I haven't posted in a while.

There is good news and bad news. The good news is that after confirming all the data with these point-by-point codes, there are only very slight changes in the overall results of the data. The winning probabilities for serving to either the left or right side are still the same, which is great news. When controlling for Ad and Duece court, there are slight changes to that hypothesis, but I would stress that the difference is VERY slight.

Now, for this to be a comprehensive study, I have to test for the equality of winning probabilities for each player in each match with distribution statistics such as the chi-squared statistic (P-Value), the run test, and the joint statistic test (a combination of the previous two). The bad news is that in an individual table tennis match, there aren't very many serves for any one player (on average, only about 35 per match). It is likely that these statistics will reject the null hypothesis in more matches than expected. However, I can do an Empirical CDF test, which tests for uniformity of P-Values. If I have a uniform distribution of P-Values, I believe that it would okay to assume an equality of winning probabilities. Of course, there aren't enough serves in any individual match to say whether or not there is an asymptotic distribution, which is what the confirmation statistics assume. But studies like these, especially high school studies, are never perfect.

So now it is a big crunch. A big crunch of numbers and a big crunch of work to have a good paper by next Friday. I'll be sure to work hard, because I don't want my presentation to be like a black hole singularity.

Jayanth