I would take their results with a huge grain of salt, basically because their sample is a. small, and b. restricted to a period 10 seasons ago. Their curve was introduced here, and here is how it was constructed: The other fail here is that they do not bother to show variation of range or even standard deviation from that data. Part of the issue here is that a lot of the pitchers who are not that great as starters just go away after age 26 or so (or they move to the pen; check Joe Nathan's or Glen Perkins's curves as starters - and both are included there.) Plus a lot of players just blow their arms. So, to talk about quality pitchers and what an aging curve looks like for a good pitcher, it would have been more meaningful to get pitchers who have satisfied the following criteria: a. They pitched for more than 10 seasons, b. their ERA+/FIP- is on the top 25% of the pitchers, and then look what happens. For every Greinke and Velander there are about 50 Duffys, Romeros, Guardados, Hawkinses, or Mayses, and including those pitchers in the graph, is just noise, not meaningful data...