Not sure where to post this but here seems reasonable.
So I compiled all the stats from PCM DBs from 2011 to 2018 into one database and cleaned it all up to allow analysis of stats over time (link below). Analysis based on this relies on the accuracy of PCM's evaluation of riders but as I note in the DB post I think that is decent assumption.
So I played with the data to develop age curves by primary stat type. I filtered to those riders who reached a value of 76 or higher on the stat sometime between the ages of 24 and 30.
For this pool of riders I calculated two values shown in the graphs below:
The first graph shows how far the average rider at that age is from the average rider at the peak age. Keep in mind this is limited to rider who reached at least 76, so average means average within that pool.
The second graph shows the % of riders at that age who will get more than 1 point better within 3 years. I ignore 1 point improvements on the basis they could be noise.
At 21 sprinters are within 4 points of their peak and they peak at 27 (and are very close by 24). Whereas at 21 cobblers are 12 points from their peak and they don't peak until 30.
At 21 most riders will still improve within 3 years, although 20% of TT riders won't. But by 25, almost 80% of CO will still improve within 3 years while it has dropped to 20% for SP riders. TT improvement rates start low but flatten out slowly, I this reflects it being a more trainable skill.
Think these outcomes reflect what I would expect but thought it was cool to see it in numbers. Want to play with stat interactions next particularly riders with multiple skills.
