Five seasons should be plenty. You'd have hundreds of data points for each two-year set.It would be amazing to do similar age curves with statcast data. I'm not sure if 4-5 seasons is enough to do this kind of analysis, but in a few seasons they'll start to have entire cohorts that have entered and left their prime completely within the statcast era.
For fun, I looked at the top 20 of the StatCast sprint leaders from 2017, to see how those with elite speed fared the following two years (this is just a fun exercise; not intended to be representative, and understanding it is a tiny sample):
The overall average speed dropped in both subsequent years. Every player but one (Trea Turner) was unable to maintain his first year speed into his last, and only one player recorded a higher average speed in either year over his first (Harrison Bader).
In terms of speed lost by age, we're obviously working with a super small sample. But here they are, anyway. We have six two-year periods with at least 3 players in the group, and those results are:
Age 21-22 (3 players): -.43 ft/sec (I wouldn't put any stock in this sample, which is skewed by Barreto, who lost a full foot for whatever reason.)
Age 23-24 (4 players): -.175 ft/sec
Age 24-25 (5 players): -.32 ft/sec
Age 25-26 (5 players): -.2 ft/sec
Age 27-28 (3 players): -.233 ft/sec
Age 28-29 (3 players): -.434 ft/sec
While this certainly doesn't prove anything, the mid-20s numbers are right in line with the general observation that on average, a player will lose about .2 ft/second per year at that age, and suggests that the heuristic applies even to elite speedsters.