Photo Credit: Tom Szczerbowski / USA TODAY Sports

Is Matt Martin Actually Good Defensively?

I think it’s fair to say that Matt Martin has been a polarising figure in Toronto this season.  Unsurprisingly, the divide mostly seems to be between traditionalists who see Martin as a player who performs a valuable role protecting the Leafs younger stars versus more statistically-inclined fans who see Martin is a low-scoring player whose tangible contributions are not very significant.  But one point of agreement has typically been that Matt Martin is good defensively.

Back when Martin was signed, I said this on Twitter:

When I took a statistical look at Martin a couple weeks later, I said that “the New York Islanders have typically done a good job of limiting shots attempts against their own net while Martin is on the ice.”

But lately I’ve been wondering if it’s actually true that Martin is good defensively.  It’s certainly true that the Islanders allowed fewer shot attempts with Martin on the ice relative to his teammates, and that trend continued in Toronto, where Martin’s Corsi Against relative to his teammates per 60 minutes has been -4.63 (for CA/60 Rel lower is better; it means a player allows fewer shots than his teammates).

Thinking about the issue further, I came up with a theory for why those statistics might be misleading.  There were two parts to the theory:

  1. Martin plays most of his minutes against bottom 6 players, especially other 4th liners.
  2. These are players who typically play lower event hockey and aren’t necessarily comparable to top 6 forwards.

If that were true, it could be the case that Martin allows less shot attempts not because he is better defensively than his teammates, but because his coaches put him on the ice in low-pressure situations against opponents who are not much of a threat to score.  This would explain why Martin’s overall shot attempt ratio is quite poor despite seemingly having good shot suppression skill.  After all, if you’re playing low event hockey but still getting consistently outshot, that’s not actually good defence, it’s just getting bad results at a slower speed.

In order to see whether that theory holds up, I decided to see what Martin’s results looked like against players who have a strong track record of generating good Corsi results.  Back in March I put together a list of the best centres on each team by Corsi Rel for another article, so I decided to use that list again.  Using Hockey Analysis I took a look at Martin’s results against those forwards over the past three seasons.  Hockey Analysis has a 10 minute ice time minimum in order to display “against you” stats for any given player, and Matt Martin played less than that cut-off against most of the players on my list.  There were 12 players who met the cut-off, and here are their results against Matt Martin over the past three seasons:

Player Team TOI CF/60 Dif CA/60 Dif Pace Dif
J.Staal CAR 18:55 18.04 -9.53 8.51
Toews CHI 15:18 -5.91 -14.82 -20.73
Trocheck FLA 23:59 -4.44 -22.58 -27.02
E. Staal MIN 17:19 10.86 -13.32 -2.46
Galchenyuk MTL 20:00 26.26 -6.04 20.22
Johansen NSH 15:17 12.83 -7.37 5.46
Josefson NJD 27:05 -11.62 -5.35 -16.97
Stepan NYR 19:53 -21.35 11.82 -9.53
Smith OTT 22:06 22.28 -9.19 13.09
Crosby PIT 36:25 -18.48 5.79 -12.69
Perreault WPG 15:14 30.51 -12.27 18.24
Eller WSH 26:46 29.03 -18.58 10.45

CF/60 Dif is whether a player generated more or fewer shot attempts when playing against Martin as compared to everyone else in the league (lower is better for Martin).  CA/60 is the same stat for shot attempts against (higher is better for Martin).  The last one is the overall pace of shot attempts (Corsi for + Corsi against), which I included to see whether Martin seemed to have any skill at slowing the game down, which might be a useful skill even if the other results weren’t ideal.

In this chart, CF/60 Dif is the measure of Martin’s shot suppression (a positive result means a player was on the ice for more shot attempts when playing against Martin vs playing against the rest of the league).  7 out of 12 players shot more frequently when playing against Martin, while the other 5 shot fewer.  There’s a pretty mixed bag in terms of results here in terms of how big the differences are, but on the whole the results do lean towards the league’s top Corsi players generating more offence when playing against Martin, which does back up my theory that his results are influenced by the easy competition he typically plays.

The results in terms of CA/60 Dif (the amount of offence generated by Martin’s team with him on the ice) are far more clear.  Other players do significantly better defensively playing against Martin than against the rest of the league.  Because of this, even if Martin had positive shot suppression abilities (and it’s looking more doubtful that he does), his team still comes out behind because he hampers the offence so much.

However, this might be something that you could live with if Martin was good at limiting the overall pace of play.  There are times, especially late in hockey games, when lowering the pace of play may be advantageous even if your offence is limited because the expected value of preventing a goal against is higher than the expected value of scoring another goal.  However, there’s no real evidence here that Martin is reducing the pace of play.  There’s a 50/50 split between players who had a higher pace of play against Martin than against the rest of the league, and the magnitude of the effect is similar in both directions.

However, 12 players isn’t a great sample size.  I decided to go back to the Hockey Analysis data to increase the amount of information we’re looking at.  I collected 12 more forwards, using a cut-off of at least 20 minutes ice time, picking out the ones Martin played against most frequently who struck me as good examples of solid top 6 forwards.  I limited my selections to one player per team to broaden the results as much as possible.  This is admittedly a much more subjective approach than the way I put together my first chart, but given Martin’s low ice time there’s only so much that can be done to collect a detailed sample against any individual players.  Here’s the second list (like the first list, it’s almost all Eastern Conference players, which is unavoidable given Martin’s low ice time):

Player Team TOI CF/60 Dif CA/60 Dif Pace Dif
Krejci BOS 21:31 -2.36 -12.34 -14.7
Skinner CAR 20:27 29.81 -14.22 15.59
Dubinsky CBJ 25:57 13.24 -13.7 -0.46
Zetterberg DET 27:58 -0.79 -1.51 -2.3
Jagr FLA 22:00 5.74 3.11 8.85
Pacioretty MTL 21:59 -12.5 -0.48 -12.98
Zuccarello NYR 28:02 17.08 5.46 22.54
Brassard OTT 30:20 19.49 -1.16 18.33
Giroux PHI 24:12 13 -13.34 -0.34
Kessel PIT 22:35 30.08 -17.69 12.39
Kucherov TBL 27:56 -18.68 -19.73 -38.41
Ovechkin WSH 23:16 -12.76 10.72 -2.04

The results here are pretty consistent with the earlier chart.  A small majority of these players generate more shots when playing against Martin, while almost all of the players do much better defensively against Martin.  And once again, there’s no clear evidence of Martin reducing the pace of play.  What seems to happen is that most players play at a similar-ish pace against Martin, but they get a bigger share of the shots because Martin and his linemates generate so little offence.

The results here back up my theory about why Martin’s results look good.  He does not show any real shot suppression ability against top 6 players (in fact, he seems to be a bit worse than average), nor is there any evidence that he reduces the overall pace of play.  Indeed, if you look at his results on a player-by-player basis it quickly becomes apparent that most of his “shot suppression” is against lower quality competition, players who get minimal ice time and play at a much slower pace overall.  The evidence just doesn’t support the idea that Martin is good defensively.  He plays low event hockey against bottom line players for most of his ice time, but when faced with the kinds of competition that his more talented teammates typically face, Martin does not appear to have any notable shot suppression ability.

(As an aside, this also supports a theory I outlined in a post about the effects of Quality of Competition on Morgan Rielly, which is that the true gap in Corsi talent league-wide is artificially suppressed because depth players spend so much of their ice time playing against other depth players, raising their results and making them look better than they would if they played better competition more regularly.)

  • Stan Smith

    I have always said that Corsi sucks as a way to rate defence. It rates everything as the same, without taking into consideration quality of shots. Take shot suppression for example. A player that is more aggressive at the blueline or along the boards takes the risk of overcommitting himself. If he is successful, great, but if he misses his chance, the opposing player, or players, has a clear path to the net. A more conservative player might give up the line easier, but keeps position between the puck and their own net. The first player may give up fewer shot attempts but better scoring chances on the attempts he does give up. The second player may give up more shots, but fewer good shots and scoring chances.

    Looking at the Leafs goals against per 60 minutes 5 on 5, Martin has the lowest of any forward on the team at 1.72. Yes his quality of competition is less, but you cannot expect your 4th line players to be better than an opposing teams best players, just better than the players he plays against. If every line you have is better than the opposing teams same line (4th line vs 4th line, etc), then you win. I would say if he hads the best GA60 on the team he is doing the job he is expected to do quite well.

    • Gary Empey

      I thought when Boyle finally came on board and Kapanen joined the line they all started to be a force to be reckoned with on both ends of the ice. Babcock seem to feel the same way and increased their ice time. Whether one is talking about Stamkos, Ovechkin, or Martin you must have suitable line-mates for them to be their most effective.

    • Kanuunankuula

      Goals against is also extremely fickle, so it’s quite probable he won’t repeat that. It also is not against the point of the article. He does good since he plays so much of his TOI against weak competition. I’d check how he does on xGA/60, but Corsica is down. I’d also show Fenwick against and such in this article.

      He was also marketed as a shot suppressor 4th liner, which he apparently is not.

    • Bob Canuck

      I looked at Naturalstattrick.com 5v5 data for the 2016-2017 regular season for forwards with a minimum of 400 TOI; that works out to be 368 forwards. Given that there were 30 teams and 12 forwards per team, the demarcation points for lines 1,2,3, and 4 is 90, 180, 270, and 360, respectively. I looked at scoring chances and high danger scoring chances per 60.

      Matt Martin’s ranking in each of the categories is as follows (the corresponding line number is in brackets):

      SCF60 – 319 (4); SCA60 – 165 (2); SCF%60 – 291 (4); HDCF60 – 325 (4); HDCA60 – 107 (2); HDCF%60 – 285 (4).

      The data above suggests that Matt Martin is a fourth-line forward who, in terms of net scoring chances and high-danger scoring chances per 60, is closer to a third-line player than a bottom-end fourth liner.

      The Matt Martin discussion reminds me of R.A. Dickey’s tenure with the Blue Jays. Leaving aside the merits of the Dickey-Syndergaard trade and Dickey’s salary, Dickey had a league average ERA as arguably the Jays fifth starter. If your fifth starter has a league average ERA, that is a competitive advantage. Yes, he’s a fifth starter but he compares favourably to other AL fifth starters.

      When it comes to Martin, yes he is a fourth liner but he compares favourably to other fourth liners in terms of net scoring and high danger scoring chances per 60. The contract is a separate matter as are the intangibles but this data supports the contention that Matt Martin is not detrimental to the Leafs when compared to other fourth liners.