Some further thoughts on goal quality

By Ian Cooper

Since we officially launched last week, we've had a lot of feedback on our first article in The Toronto Star.  I wanted to take a moment to thank everybody who took the time to tell us what they think and to respond to some of those comments.

Because we only have 700 words or so in each article it's never possible to address every counter argument, so I wanted to briefly cover a few here.

I recognize that goals vs. playoff and non-playoff teams is a very rough way of measuring goal quality and that we should have looked at goals per game since some players (e.g. Ovechkin) played in a division where they were usually the only playoff team and therefore had far more games against non-playoff teams.

That's true, and in fact I did look at that.

Now it's possible that all of those teams didn't make the playoffs because they had to play against the Great Number 8 so many times and he really is just that good.

Personally, I don't find that argument compelling given Washington's low overall standing in recent years and the fact that Ovechkin has never advanced beyond the second round of the playoffs.

But if you're convinced, that's OK.  As any good lawyer will tell you, arguments about how we judge agreed upon facts just aren't that interesting.

If we look at goals per game against playoff and non-playoff teams, the story doesn't get better for Ovechkin.  In 21 games against playoff teams last season, his 9 goals amounted to 0.43 goals per game.  In 27 games against non-playoff teams, his 23 goals amounted to 0.85 goals per game.  Now the non-playoff teams will give up more goals for sure, but not twice as many.

But wait!  This is a small sample size.  Well, that's true as well. Regrettably, because hockey players don't play thousands of games a year, get injured, get traded, etc., we are always limited to small sample sizes and external factors.  But unless we're all willing to throw up our hands and say "who knows - it's all a matter of luck - one year Ovechkin's the best player in the league, the next it's David Clarkson", we do need to look at the data that's out there and draw conclusions.

Does this mean Ovechkin's a stat padder?  Maybe.  But then again, maybe not.  We'd need to look at a lot more data to make that claim more forcefully.

From where I'm sitting, as much as Don Cherry's rants may be a rhetorical tool that lets me offer up a fun read, that doesn't necessarily matter.

Regardless of whether or not there's intent here, we've observed that Ovechkin scored twice as often last year against teams that let in more goals for sure, but not twice as many.  So unless I'm prepared to make the (rather odd) argument that if only he had the opportunity to play a tougher schedule, he would have had even more goals last year (e.g. because he would rise to the occasion, learn how to beat those better teams with more practice, etc.), I'm forced to admit those 32 goals in 48 games mean less than some might initially think.

Some have set up a false dichotomy and claimed that I'm arguing Ovechkin sucks.  Of course I don't think that.  He's an elite player and a very effective goal scorer by any measure (0.43 goals per game against anyone, let alone the better teams in the league is impressive).  That said, his goals total is overstated because of where he's getting them.  If he played better teams last year, I'm pretty sure he would have had fewer (I'll take a wild leap here and say it would have been somewhere closer to that 0.43 per game number).

There are lots of different ways of thinking about the quality of a goal (state of the game when it was scored, power play vs. even strength vs. short handed, quality of goaltending (if you score against a good team's backup, did you really score against a good team?), value of an empty net goal, etc.), and we are looking at as many of these as we can think of.

When we write about this next, we will offer a measure that accounts for some of what's going on here.  When we do, we'll make sure it's something comprehensible so that fans and teams can start using it.  That may ruffle some feathers among the advanced stats community, but I would point out that Corsi and Fenwick, which are only barely gaining mainstream acceptance, do not involve complicated math.

So stick with us, tell us what you think, call us idiots if you'd like.  And let's keep the conversation going!


No Comments Yet

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>