Tuesday, 10 June 2014

Adjusting Scoring Rate for Age in CHL Prospects

Nathan MacKinnon, one of the youngest 1st round picks since 2003.
I wrote a post for Canucks Army this morning (read it here and click on all their advertising) in which I argued that Vancouver could expect to win a trade for the 1st overall draft choice this season - even if they gave up their 6th overall pick and Bo Horvat to do it - if they used the draft pick to select Kootenay's Sam Reinhart. The genesis of the pro-Reinhart argument is that Reinhart scored at a level comparable to other extremely highly-touted former prospects (and current NHL stars), therefore he was player worthy of being selected 1st overall, despite the skepticism among experts that this draft contains a "true #1 selection."

After publishing the article, it was pointed out to me that Reinhart is older than the average draft eligible prospect, and therefore his offensive numbers could not be fairly compared to a player like Sam Bennett, who is 8 months younger. The argument here is essentially that scoring rate increases with player age, and for the most part, this probably isn't an unreasonable generalization to make. 20 year olds score more than 19 year olds, who score more than 18 year olds, and so on and so forth. If this pattern exists on a macro scale, then it should logically exist on a micro scale, too. A kid who's 17.8 years old on in September should, in theory, score more than a kid who's 17.2 years old at the same point in time.

If this assumption is true, then we should be able to observe a consistent positive relationship between intra-year age and scoring rate. Fortunately, Josh Weissbock had previously scraped a whole bunch of data for a project we've currently postponed (we're instead working on Sham Sharron 2: Return of the Sham), so I had a repository of scoring and age data for every player to play in the OHL since 1990. Eliminating small sample size guys and correlating intra-year age with scoring rate, I was left with the following equation:

Pts/GP = (0.1672*Age) - 2.3714

This equation tells us that for every year a player ages, we can expect scoring rate to increase by 0.1672 points per game. This means that, for first time draft eligible players, we can adjust scoring rate for age through the following equation:

Age Adjusted Pts/GP = (1 - ([Age as of September of draft year - 17] * 0.1672)) * Pts/GP

This formula will estimate the predicted scoring rate of a player if they were exactly 17 years of age at the start of their draft year, based on their actual age and scoring rate. It is, for all intents and purposes, a measure of scoring relative to Nathan MacKinnon, who was almost exactly 17 years of age at the start of his draft year.

To see how it works, we'll use John Tavares as an example. Tavares was roughly 17.94 years old at the start of his draft year, and he scored 104 points in 56 games, split between Oshawa and London. His age-adjusted points per game is as follows:

Tavares Adj. Pts/GP = (1 - ([Age as of September of draft year - 17] * 0.1672)) * Pts/GP
= (1 - ([17.94 - 17] * 0.1672)) * (104/56)
= (1 - ([0.94] * 0.1672)) * 1.857
= (1 - 0.157) * 1.857
= 0.843 * 1.857
Tavares Adj. Pts/GP = 1.565 Pts/GP

Whereas if you attempt the same formula with Nathan MacKinnon:

MacKinnon Adj. Pts/GP = (1 - ([Age as of September of draft year - 17] * 0.1672)) * Pts/GP
= (1 - ([17.00 - 17] * 0.1672)) * (75/44)
= (1 - ([0.0] * 0.1672)) * 1.705
= (1 - 0.0) * 1.705
= 1 * 1.705
MacKinnon Adj. Pts/GP 1.705 Pts/GP

As you can probably see, the closer a player is to 17 years old at the beginning of their draft year, the smaller the adjustment to their scoring rate. As a side note, I don't know if this age adjustment works for any other age ranges than 17.0 years to 17.99 years old, as I basically just built it to look at first-time draft eligible players. I'll have to look into this later.

So there you have it, a formula to adjust scoring for intra-year age for first-time draft eligible CHL prospects. I don't know if anyone had done this publicly (if they have, I couldn't find it), but I hope this provides some useful insight going forward. 


As an aside, here are some top 2014-eligible prospects ranked by age adjusted Pts/GP:

Good news for Sam Bennett fans, however Reinhart still comes out on top when adjusting for league offense. Really though, both guys stack up well with some pretty elite company (and so does Nikolaj Ehlers, their closest comparables are here, here and here) and a team selecting 1st overall probably can't go wrong with either.

No comments:

Post a Comment