Originally Posted by falstaff78
Hey Colin -
Thanks for the kind words! By the way I spend about half the week in Cambridge - are you a student?
Re: the discussion between you, Zagor and Tudwell.
The data would suggest that Federer is an outlier due to the LEVEL of his performance and not the SHAPE of the age-results profile. We can see that Fed's best years occurred precisely when the data suggested they would - i.e. August 03 - August 07.
The chart below disaggregates major victories by age for all 12 guys with >5 open era majors. Red highlights are late outliers, yellow highlights are early outliers. You can see the 5 important outliers to this trend in the entire open era.
Amazing work Falstaff! It's really interesting. I had the same problem than you. My wife thinks that I'm crazy to spend so much time in tennis number (instead of playing it!).
However I think their is a problem with your analysis of outlier. Are outlier really outlier when they are nearly half of the total cases? To solve this problem it would be necessary to include more cases. You can't make that with players who have won less than 5 slams, because they wouldn't be spread enough, so you have to include finals as well. The ideal thing would be to pick a bunch of top 10 players from each decades we know were not to much stoped by injuries and make a scale for their slam result. Thus we could see slam success by age for a variety of players, including the one who are not able to win or reach slam final.
Unfortunately I don't have a lot of free time now but I hope that I can someday participate to a community of tennis freak scientist.