NHL Points-per-Game Peak Age Estimation
There are numerous ways to estimate the age at which NHL players peak. You could, for example, count all games played by players at each age or count the number of players at each age and assume that the largest number corresponds to the age at which player skills peak. This method is heavily impacted by the group of players who are chosen - if players with short careers are included, the peak age is skewed low because younger players play on potential and leave the league if they don't impress during a tryout; if only players with long careers are used, the peak age is skewed high by a bias towards players who avoid injury.
One way to avoid this is to include all players and account for their performance in leagues below the NHL. For my sample, I looked at all NHL players born 1962-79 who played exclusively in the NHL, AHL or IHL from Age 21 to Age 29. I assumed that AHL and IHL points were worth 45% of an NHL point. The average points-per-game at each age came out as follows:
The peak age is just slightly more than 25. The peak age actually falls at approximately 25 for a wide range of NHL equivalencies for the minors. This is also roughly the same result as you get if you restrict your dataset solely to players with careers longer than 200 games and you look at the number of NHL games played at each age. Other methods don't give substantially divergent results - even the most or least restrictive datasets result in peaks between age 24 and 26.
0 recs |
13 comments
|
Comments
I think you just depressed about 80 percent of the NHL with that one.
Bettman's Nightmare: A Blog Where Hockey Aficionados Dismantle That Mighty Empire, One Balsillie at a Time
http://bettmansnightmare.blogspot.com/
by Bettman's Nightmare on Jan 21, 2010 12:09 PM EST reply actions
I got an idea for a trainwreck…how about finding the curve for goalies?
Bettman's Nightmare: A Blog Where Hockey Aficionados Dismantle That Mighty Empire, One Balsillie at a Time
http://bettmansnightmare.blogspot.com/
by Bettman's Nightmare on Jan 21, 2010 5:02 PM EST reply actions
Surely offence peaks earlier than all-around effectiveness.
Can I suggest doing a similar study (perhaps NHL only) using plus/minus? Yes I know +/- is much-maligned and with some good reasons, but averaged out over large groups of players or careers there may still be useful data in there, especially age-specific stuff. I certainly found it useful in this study of high draft choices.
I would guess you would find a somewhat similar curve shifted a couple years to the right of the one above. Probably more jagged, but I don’t expect a random ~flat-line, let’s put it that way.
Writer for The Copper & Blue and primary shareholder of Zorg Industries
"Never be ashamed of who you are" -- Jean-Baptiste Emanuel Zorg
Yeah, I’ll echo Bruce’s point.
Since you seem to have the list of player ages to hand, Gabe … have you got the same for rel corsi, ice time per game, and qualcomp? And separating forwards from D would be really beneficial IMO.
There will be a survival bias for sure. Teams are willing to build the game plan around a young player (SEE Comrie and the kid line) but eventually they need them to take on more responsibility so the next promising youngster can be put into a position to succeed. And guys that never take that leap, or struggle with it … they usually end up with reduced roles, playing on bad teams, or in Europe.
The database with player ages in it is unfortunately separate from the Corsi database. A small pain to combine them…Anyways…
What kind of analysis are you thinking? I have two years (soon to be three) of Corsi and Qualcomp and 10 years of TOI. It doesn’t seem like enough data about individual players to see a pattern, though TOI is getting close.
A shame they’re in separate dbs, ah well.
A histogram of qualcomp, corsi, zonestart, etc by age would be interesting, though. cumulative for whatever data is handy. My collecting scripts don’t use player names, just numbers … it would be too much work to find the matching birth dates. It’s worth looking at, though, I’ll get around to it eventually.
That makes sense. In soccer and tennis the peak age is even younger (around 23 or 24) and hockey is fairly similar aerobically. Where did all this nonsense that someone’s peak age is 28 start? I have a feeling it’s sports writers and fans trying to make themselves feel better.
The West Coast is the Best Coast.
Superstars?
Does the curve change at all for superstars (eg, guys with > .75 PPG or > 1 PPG)?

by 






