What is true is that world records only happen a few times a year during championship meets yet swimmers do pretend races every day during training. Moreover, records have a different statistical distribution compared with the underlying time series, so the list above says nothing about how the daily improvements would look for a typical top swimmer.

At this level of competition, all bodily functions are controlled and training happens every day for most of the day so there's plenty of scope for steady progress with tiny improvements of the order of 1/100th second every couple of days, which add up to 1/10th progress by the time that the next big race occurs.

However, I couldn't find a time series supporting this with casual googling, and in truth I expect it's too valuable to be found on the web. At this level of competition, the swimmers are measured daily, and all their bodily functions are controlled. A complete time series of daily improvements is likely worth a small fortune. Swimming coaches at this level of competitionare unlikely to give out daily timing progress

--
$E(X_t|F_s) = X_s,\quad t > s$

Indeed, if it were repeatable to that precision, there would be little point in racing.

In Oz, where they take their Olympic swimming coverage quite seriously, you not only hear the winning time and whether it broke a World Record or Olympic Record, but also the PB's (personal best time) of the individual ozzie swimmers and whether they did a new PB at the Olympics.


I've been accused of being a Marxist, yet while Harpo's my favourite, it's Groucho I'm always quoting. Odd, that.