Photo: Bill Smith
[Editor’s note: Eric Fingerhut is RMNB’s ombudsman. You can follow him on Twitter and read his other work on his blog here.]
I know everyone expects complaints and criticism from the ombudsman, but today, I want to start with some praise. The Winter Classic might have been the biggest day for hockey in Washington, D.C. history, and this website did a tremendous job chronicling it – before, during and after the game. Who did the best job of anyone covering the Caps in trying to figure out why Washington didn’t have an alumni game like other Winter Classic cities? Who did the best job covering the Caps alumni who came to town for the game? Who provided a constant stream of entertaining stuff to the lead up of the game – from players’ Instagram photos to reports from the family skate? Who provided a joyous recap of the game? Ian, Peter and Chris at RMNB did. The only complaint I could have about this site’s Winter Classic coverage is that there was so much of it, I found it hard to read it all. Really an amazing job.
Of course, that doesn’t mean I have nothing to complain about.
First, Patrick Holden has done a number of very interesting statistical analyses on the site, but I thought his recent post estimating Braden Holtby’s worth was problematic. I followed the math figuring out Holtby’s value per win and understood how he got the eye-opening $8 million cap hit for the Caps’ goalie over the next four years. And yet because that cap hit would make him the second-highest paid goalie in the NHL, there’s pretty much no way Holtby is getting anything close to that in his next contract, something even Patrick acknowledges in the comments to the post. (OK, a Caps Stanley Cup and a Conn Smythe for Holtby this year, and maybe he gets somewhere close to that— but the Caps should be lucky enough to have that problem.)
That’s why I thought the post’s failure to provide much context for how Holtby’s worth compared to the worth of some other NHL goalies was disappointing. Other than a $14 million/year calculation for Henrik Lundqvist, no one else was mentioned. How do the salaries of Corey Crawford or Jonathan Quick or Tukka Rask compare to a calculation of their worth? Perhaps by seeing how much they’re making and how that compares to their actual contracts, we really could project what would be fair market value for Holtby. Of course, then you get into further complications – all those goalies I just mentioned won either a Stanley Cup, a Vezina Trophy or both, which would factor into the amount of their current salaries but don’t factor into the worth-per-win calculation Patrick makes.
In other words, this post was an interesting intellectual exercise, but not all that useful in trying to figure out what Holtby might make in his next contract. I think readers would have benefited much more from the latter.
The other subject I wanted to gingerly bring up about the site is what has started to seem to me an overreliance on shot-differential/possession stats, particularly in small sample sizes. Let me explain.
In his very compelling half-year in review post, Peter noted that the Caps had not played well since the Winter Classic. Considering the team was 3-0-1 in the four games he was talking about, the statement struck me as a little odd. As evidence, Peter pointed to the Caps’ relatively high shooting percentage during that time (13.6) along with a 48.4 score-adjusted Corsi. Shooting percentage is always a very volatile stat in small sample sizes – it’s not particularly unusual for a team to score five goals on 25 shots one night, four goals on 30 shots another, and then even out a couple games later with a couple of games with one goal in 30 shots or two in 35. And sure enough, the next two games the Caps scored a total of three goals on 64 shots (4.6 percent shooting percentage). The Caps did survive that downturn, though, by giving up only 1 goal in those two games. But having a high shooting percentage doesn’t indicate by itself whether a team is playing well or poorly — it could mean a team has been facing poor goaltending, or converting on a lot of excellent scoring chances, or just getting lucky.
Peter’s contention is that the high shooting percentage was bailing out a team with a problematic 48.4 Corsi percentage. But is that Corsi really significant and indicative of a team not playing well, or are we exaggerating its importance here? Shot differential stats are predicated on the idea that because there is a decent amount of luck involved in hockey, the more times a team shoots the puck, the more chances that team has to score. If your team has a 48 percent Corsi percentage, that means that if every game had 100 attempted shots at 5 on 5, your team is going to have more than 300 shots less than its opposition over the course of a season. That’s a significant difference. But a 48 percent Corsi percentage in one game? In my hypothetical game of 100 attempted shots, that means that your opponent had about one shot more than you each period – you trailed 17-16 in shots in two periods and 18-16 in the third period. In one game, that doesn’t strike me as something too worrisome.
But even if that same thing happens three or four games in a row, that still doesn’t seem particularly problematic, especially when in two of the games, the Caps led by multiple goals for lengthy stretches. Yes, we’re using score-adjusted Corsi, but since it’s adjusted based on average score effects and we’ve seen that the Caps are generally below average in generating shots when leading, those leads certainly had an effect on their shot differential. The fact that the Caps do have a tendency to go into more of a defensive shell that most teams is problematic, but that would be a situational, and not overall, problem with their play. In other words, I find it hard to say that Corsi shows they’re not playing well when a chief reason their Corsi is under 50 percent is because they’re winning by multiple goals. Add that together with the fact that, as I said earlier, we’re talking about a very small sample size for a stat that is much more compelling when spread over dozens of games, and it seems like we’re placing a lot of importance on a handful of attempted shots per game. Sure, if a team has a 40 percent shot-differential over four games, that’s a problem – but at 48 percent, is that truly a sign of lackluster play or just some statistical noise in a lengthy season? Just my thoughts — I’d be glad to entertain evidence otherwise.
Finally, thanks Ian, for this Filip Forsberg post. It’s sad and makes me angry, but it’s important to never forget…
And that’s it for now. If you have specific issues you want me to look into, tweet me at @ericfingerhut.
