Very few people in this country don’t know about the hockey birthday thing — either you’re in hockey in some way, or you’ve read Malcolm Gladwell, and the intersection of those groups seems to cover about 29 million Canadians — so I won’t rehash it. Gabe Desjardins has a pretty good takedown of Gladwell if you are new to the idea.
The bias, of course, is most present in junior hockey. The argument, as Desjardins describes it: “The junior coach is only going to have his players for 2-3 years, so he needs to pick kids who will help him win now. Kids born in January will be bigger and stronger than kids born in December of the same year, so it makes sense to pick them with such a short horizon.”
But does it make sense, though? Yes, they are bigger and faster, but are they necessarily better players because of that?
Unfortunately we don’t have the same advanced statistics in CHL to analyze that question that we would at the NHL level. (Building a ranking system around points per game is imperfect, but all we can do. Lemons, lemonade, etc.) We can try, though.
First-round selections: fall birthdays need not apply
What I did first was take every first-round choice in the OHL Priority Selection from 2000 to 2006, inclusive. Eric Staal, John Tavares … and others who weren’t so lucky.
All together there were 139 skaters chosen. As you’d expect, those born in January, February, or March (Q1) constitute more than half (71). Those Q1 players ended up scoring 0.70 PPG in their CHL careers. The October/November/December (Q4) players, all 18 of them, averaged 0.66 PPG, virtually the same scoring rate.
Of course if a Q4 player is drafted in the first round, the team who drafted him probably already knows he has the skill to match up with the Q1 players also available. What this (probably) tells us is just that for every Q4 player who the OHL teams want, there are four Q1 players who are just as good at that age. To find evidence of an age bias (that is, to find evidence teams are overlooking younger players for no reason other than their month of birth) we would need something more than this.
So what if we just look at those who played junior hockey this past season? That’s when it gets kind of interesting.
Why the birthday bias exists
I have 720 skaters (and their ’10-11 stats and birthdate), which obviously isn’t everyone on 60 teams. But it’s basically the top half or more of the roster, so we are focusing on the players who “matter” (for lack of a better term). We’ll break them down by birth quarter.
First, the simple counts:
Again for most people this chart is nothing new. I include it here only to establish the trends of those with early birthdays (in purple) and late birthdays (in blue). Take particular note of the counts for Q4 in one year and Q1 in the following year.
When we switch out “number of players” for “average points per game” then those peaks and valleys go away:
This is why the birthday bias exists in junior hockey. Older players are generally better — that’s a pretty decent correlation, actually, and shows there is something to the disproportionate selection we see in the first chart. Even if we focus only on 1993-and-later birthdates, under the assumption that any 1990- and 1991-born players who are still in the league have already shown themselves to be deserving of a CHL roster spot, there is a dropoff in PPG and thus we can infer a dropoff in quality.
But not a large one. While PPG and age are certainly correlated, it’s only to a small degree — less than 0.02 PPG per quarter. Which means over the course of 65 games, say, a Q1 player can be expected to put up about three and a half points more than a Q4 player born the same year. Assume about 1.5 assists per goal and that means about one and a half goals. Which is about one-fifth of a win, or one-tenth of a point in the standings. At most.
Of course if you take a roster full of Q4s and replace them with a roster of Q1s, then we’re talking about one-fifth times 18, or 3.6 wins (assuming the rates from above hold), and that difference is certainly enough to matter.
What to do about it
The issue isn’t “are older players better in their CHL years than younger players?” The answer to that is almost certainly yes.
Rather, it’s “are teams responding to this situation in an optimal way?” And I’d have to say: probably not.
For one, Q4 players from one year (say 1992) are better, all else being equal, than Q1 players from the next year (1993) but there are more than twice as many Q1-next-year players (100 vs. 231).
And even in the same year, remember the Q1-to-Q4 difference is tiny if you’re choosing between two players (less than 0.06 PPG). So there should be a much more even distribution of birthdays in the draft once you are into the rest of the player pool and your really high scorers are (presumably) no longer available.
For example, CHL teams could, in later rounds, go “off the board” and select late-birthday players who are otherwise undervalued. Why not, right? They’re not that much worse than Q1 players. Maybe you can pick a local kid with a December birthday rather than trying to convince a January kid who lives a 12-hour drive away to join your team.
I say “could” above rather than “do” because we’re obviously not seeing that behaviour:
Birthday distribution of 2000-2010 selections (non-goalies only)
(The worst offenders, by the way, are the Kingston Frontenacs. Of their selections in the sixth round or later, more than half were Q1s and only 3% were Q4s, for a ratio of nearly 18:1.)
Incredibly, the last decade of OHL drafts have seen even fewer Q4 players drafted after the first round. There is simply no way we should still be seeing a 4:1 ratio in the 11th round. If you’re down to “projectability” at that point, rather than stats and performance, then fine, pick the tallest kid if you have no other information about them.
But there does seem to be an over-reliance on age, even beyond what we’d expect there should be.