A point about Tortorella and shot blocking

I’m not sure under what circumstances I wound up listening to the second intermission show on 1040, but there was no talk of any sort of substance. Rather than walking us through strategy or matchups, whoever the host was decided it was time to ask whether John Tortorella had the skills to properly motivate his club when down a goal. I’m not sure what there was of importance to talk about, particularly so early in the season, but the discussion was geared toward possibly the lowest common denominator of fan, or somebody who has no idea what a coach actually does.

If I were a space alien beamed to the Planet Vancouver during the summer and gained an immediate knowledge of the English language, based off what was being discussed in the press, I would know the following two things about Tortorella’s system:

1) He doesn’t like the media.

2) He likes shot blocks.

That is it. That is all I’d have learned about him. Every opinion about Tortorella seems to surround one or two of those things. My alter-ego the space alien would have been very confused. What are Tortorella’s breakout preferences? How does he like to deploy his players? What’s his opinion on sleep therapy? I’d talk to friends casually about hockey and even friends that have nothing to do with hockey media would preface their opinions with how Tortorella treats the media, as if it matters a lick to what happens on the ice. (Reactions were mixed. Some people liked the way Torts treated Larry Brooks)

This brings us to shot blocks.

When it was announced Alex Burrows would be out for a bit after suffering an injury blocking a shot, it compounded with Jordan Schroeder missing the start of the regular season in a walking boot with a broken foot, also while blocking a shot. Since John Tortorella likes blocking shots, his system must be the reason why Burrows and Schroeder are out early on in the season, right?

Well, not really. That was a point Blake Price tried to make:

I guess the theory is that Tortorella is the only coached that ever blocks shots, and Tortorella is the only coach that loses players to injury.

The point isn’t that what Blake said is demonstrably false (scroll below). It was that he had shown nothing to back up his opinion, other than the fact Burrows and Schroeder were hurt in the early going of the season. There is a tonne of information available online about injuries, and shot blocks. I didn’t have time to look at it that day, but I pretty much asked him straight up where he was getting his information from, or whether he had anything to back up his statement that blocked shots cause an increase in injuries.

Logically, it would make sense, yes? Burrows has been hurt to injury. The Rangers had three significant injuries during Tortorella’s time there thanks to blocked shots. Ryan Callahan missed the end of the 2010 season after stepping in front of a Zdeno Chara slapshot, Mats Zuccarello missed the 2012 playoffs with a broken wrist and Chris Drury appears to have hurt his knee blocking a shot back during the 2009 season.

Tortorella’s teams do often lead in shot blocks. They were 6th in the 2013 season, 4th in the 2012 season and 4th in 2011. If you looked at road numbers alone, which are probably much more valuable for properly using real-time statistics, the Rangers are 2nd, 3rd and 4th.

But Tortorella’s teams do not often lead the league in injuries, or come anywhere close to that. The blog Springing Malik tracks two injury-related statistics: man-games lost to injury, as well as CHIP, which I prefer. That’s the weighted salary cap hit of players lost to injury. A season without Alex Burrows would score higher (lower?) on CHIP than Dale Weise because Burrows is a more important player, but they’d be equal in the dubious man-games lost statistic.

Tortorella joined the Rangers during the 2009 season, and coached his first full year in 2009-2010. Between 2009 and 2013, the Rangers are 16th in the NHL in “CHIP”. Since then, the Rangers are 16th in the NHL in CHIP and 20th man-games lost, despite being 4th in blocked shots and 2nd in road blocked shots.

The highest-ranked team in CHIP, Pittsburgh, are 24th in road blocks and 18th in overall blocks.

Look, Blake may have a point on the surface: Putting yourself into a situation to block more shots seems like it would lead to more injuries by shot blocks. Shot-blocking injuries, though, seem to be a very rare occurrence. I posed the question on Twitter this morning and a Rangers fan answered with these players who had seen a significant injury blocking a shot:

Players can also get injured throwing checks. Dan Hamhuis missed the 2011 Stanley Cup Finals after coming up for the worst throwing a hit on Milan Lucic. Players get hurt standing in front of the net on a powerplay—Joffrey Lupul did so last season in Toronto. Goaltenders get hurt stretching for saves, players get hurt racing into the boards for loose pucks and Brent Sopel was once incapacitated infamously after picking up a cracker and straining his back.

With 120 team seasons available (I pro-rated the 2013 season by dividing CHIP, man-games and blocked shots by dividing by 48 and multiplying by 82) I ran some linear regressions to test the correlation between injuries and blocked shots.

The x-axis is CHIP, or man-games lost, and the y-axis is blocked shots. I ran four different regressions. As always, an r-squared of 1 implies perfect correlation, and an r-squared of 0 implies no correlation:


This data doesn’t conclude that there’s no risk in blocking a shot. But it can lead us to these other plausible conclusions:

  1. Getting injured while blocking a shot is a rare event, based on the thousands of blocked shots every season resulting in few injuries.
  2. If Tortorella’s players open themselves up to greater risk blocking shots, a different element of Tortorella’s strategy in New York protected his players.
  3. Perhaps the player’s training habits, not blocked shots or elements of a coach’s strategy, are the determining factors of injury.

There is a lot we don’t know about injuries and I’m curious on it. You hear about players being “injury prone” but I have yet to see studies done on whether a player that has suffered a certain injury in the past is more likely to suffer a similar injury.

I don’t think that Tortorella’s system opened up his Ranger players to any extra risk they wouldn’t otherwise face in New York when he was there. Hockey is an inherently dangerous game played at a high speed with 200 lb players carrying clubs and blades attached to their shoes. I’m generally anti-shot blocking because in most cases, you’re blocking a defenceman’s shot, and high shot block totals indicate you’re playing much of the game in your own end.

The most interesting thing I’ve read comes from Tyler, who showed that teams that block a higher percentage of forward shot attempts are less likely to see forwards take those attempts. I think as we learn about Tortorella’s system it’s not only going to be about “does his team block shots?” but about the situation, whose shots are being blocked, and whether that leads to a better breakout or not.

During his time in New York, the Rangers were 16th in Corsi and 11th in Fenwick. They went from a negative possession team to a positive one when you included blocked shots. I noticed this also during the 2012 Memorial Cup in Shawinigan with the London Knights. The Knights, despite losing the Corsi battle in each of their games, tallied more scoring chances. Simply blocking high volume shooters isn’t the answer to the shot blocking riddle. I think there’s more to it.

It’s still far too early to judge anything about Tortorella’s tenure in Vancouver, but it’s worth noting that Burrows isn’t the only player to be placed on the injured list after blocking a shot. I think we need to recognize that when things happen with very little frequency, such as injuries due to a shot block, you can probably attribute it more to randomness than anything else.

Remember, even the 2010 Canucks, who blocked the fewest shots (not counting the 2013 season) of the seasons Tortorella was in New York, blocked 57% of the shots the 2010 Panthers did, who blocked the most. A standard deviation is about 116 blocks in a season. Even the teams that block the lowest numbers of shots block a high amount. You’d have less injury risk if you never blocked a shot, but that’s not likely to happen anytime soon.

  • acg5151

    “If I were a space alien beamed to the Planet Vancouver during the summer and gained an immediate knowledge of the English language, based off what was being discussed in the press, I would know the following two things about Tortorella’s system:

    1) He doesn’t like the media.

    2) He likes shot blocks.”

    Don’t forget

    3) Stanley Cup winning coach and some nonsense about winning DNA…

    Very interesting post!

  • acg5151

    Yeah. How many players got injured under AV. Sami Salo, Ryan Kesler, Kevin Bieksa, Dan Hamhuis, Daniel Sedin, Chris Higgins, David Booth, Willie Mitchell, etc.

    I guess we can blame all injuries on coaches.

  • acg5151

    When empirical evidence disagrees with a theory, there needs to be an explanation. In this case, that explanation could either be that the theory is wrong or there is a problem with your estimation strategy.

    I think the theory that, all of things being equal, blocking more shots leads to more injuries is likely true. It’s hard to imagine that blocking shots would not have an impact on injuries.

    In this case, I would case that your estimation strategy is the problem. Your key evidence is that the R squared on the linear regression is low therefore there is no evidence of a relation between blocked shots and injuries. I can see a couple of problems with this strategy.

    First, your model is:

    injuries = a + b*shots block + error

    In order to get an accurate R squared, you need to accurately estimate a and b. Ordinary least squares will only give an unbiased estimate if and only if the error term is strictly exogenous from the shots blocked term. You’re probably going to have an endogeneity problem as players that are prone to injury would probably not be put in the position to be block shots. For example, you probably wouldn’t want Sami Salo or Danny Brier to block shots.

    Second, R squared on its own is a little trivial. Statistically, it doesn’t matter whether it explains all or none of the variation in the injuries. The key question is what is the magnitude of b and is it statistically significant. A low R squared is what I would expect from this sample but not because blocked shots don’t lead to injuries.

    Third and last, your samples might have some outliers in them due to either freak injuries or arena affects. For example, you’re gonna get a large CHIP for Sidney Crosby when he got his concussion and missed lots of games. I’m not sure empirically how large of an effect this would have.

    A better estimation strategy would be to look at blocked shots by individual players. You look at variation between players and within players. It would be easy to control for prior injuries and arena affects. Finally, you could also use a fixed effects model to control for individual player effects.

      • acg5151

        I would guess that the estimate of b is biased though. There’s too many factors for a simple linear regression to accurately estimate b. I think doing a more complex panel model with fixed effects would probably be insightful.

        • Well yeah.

          Blake Price seemed to have an awfully good idea of what “b” was, despite having no data to support it. b is probably nothing. It’s probably higher than 0, but it makes up a very small part of a+b+error, as shown.

  • I’m the only one who gave up on Team 1040 years ago? Ever since they allowed tha bush league David Pratt rattle on about bike lanes, I turned that crap off. Even though they got rid of him, their station still stinks of his leathery dirt-bagginess… Wait, that could because he’s back!? My word. Tha is one dumb station.

  • I have two courses of thought about blocking shots.

    1) I’m not sure how effective it really is statistically based on the fact that most shots being blocked are going to be from the perimeter. Which the goalie should technically have.

    2) With the amount of shots taken (on net, missed, blocked) would just pure luck and a random factor create a situation that no matter whether a team is playing a shot blocking system like the Canucks now do vs the system they were playing before?

  • Cale

    Cam you just gave me a nerd boner with this post. Seriously, my one beef with CA as of late is that the statistical analysis hasn’t been as strong or as prevalent as it had been. More posts like this, please.

  • Cale

    A commitment to shot blocking on purpose, done correctly, might well reduce the likelihood of injuries caused from shot-taking for all we know. Presumably, when you coach shot blocking, you teach how to do it properly. Facing up to the shot with pads and toes forward, stick blade squared to the ice is low risk. Lollygagging in the slot with your finger up your nose while waving your stick randomly in front of you pretending to do defence is how you are more likely to get injured from a shot.

    The evidence that blocked shots are taken from the perimeter and so are not likely to lead to goals is non-existent. Rather, all the evidence points to the conclusion that all shots are worth taking and increase the likelihood of scoring. Therefore, they must all be worth blocking.

    It may be that Crosby’s injury data arising from non-shot blocking injuries skewed Cam’s results a bit. However, does this not ultimately raise the equally pointless argument that players should not only let shots at their net pass them by, they should skate slowly and avoid the offensive zone altogether? Way more injuries are caused by human-human collisions than human-puck collisions. If you tell players not to get in the way of a shot, where are you sending them instead? To go hit somebody? I’d suggest that getting in the way of a Chara shot is less life threatening than getting in the way of Chara himself.

    The purpose of a body check, other than to entertain me, is to prevent a shot on net. The check may or may not accomplish that. A blocked shot always accomplishes that.

    Apparently, the object of the game is to collect a pay cheque, avoid injury and get out of the way of everything if not everybody.