I’m answering this off the top of my head, so nobody gripe at me for not sourcing properly:
The generally accepted peak year is the Age-27 season. James, IIRC, did some research and found that it might be a year or two behind that. You’ll still find mainstream (for stats people) acceptance behind 27, but somewhere between 27 and 29 is going to hit 95% of observers. If you’re looking for research, one of James’ Abstracts (hand to God, I don’t remember which) is the way to go.