Mathieu Perreault

Early this morning, Mike Vogel broke the news that the Caps reassigned forwards Jay Beagle, Andrew Gordon and Mathieu Perreault to AHL Hershey. Then during Media Day, Bruce Boudreau stated that he thought Marcus Johansson had more defensive upside than Perreault to try and explain why the French-Canadian was sent down over the Swede. I am far from a Perreault supporter, so you know this is coming from the numbers heart.

Looking at the pre-season stats of MJ90 and MP85, do they really show MJ90 is better defensively?

If we go by conventional stats then yes: MJ90 was a +5 in 5 games played with 0 Goals and 4 assists while MP85 was an even player with 1 goal and 1 assist in 5 games. However this is RMNB and we are anything but conventional, so let’s take a look at scoring chances.

A scoring chance is a shot directed at net – an actual goal scored, a shot on goal or a missed shot – and if we look at scoring chance differential we get a completely different story.

Both pivot hopefuls played roughly the same amount of time in the preseason: MJ90 63.5 minutes, MP85 60.4 minutes. MP85 was on the ice for 55 scoring chances for and 53 against, while MJ90 was outchanced 53 to 37 when on the ice. In addition, the Caps were outchanced in 4 out of his 5 games – including a -14 clunker – when he was on the ice. Last time I checked, defense meant keeping the puck away from your own net.

MP85 was outchanced in only 2 of his 5 games but never worse than a -5. Some of that can be attributed to MP85 spending 32.55% of his 5v5 time with Knuble and Ovechkin – Backstrom’s first line scoring threats. Some of it could be chemistry. Some of it could be his AHL/NHL experience. All of it is better than MJ90′s on-ice numbers.

This brings up a question: If MP85 was a legitimate 3rd line center candidate, why not see how he would do with players he would see regular time with? MJ90 spent most of his 5v5 time with Andrew Gordon and Matt Hendricks – more realistic linemates before RMNB’s favorite blogger was sent down through waivers.

So is MJ90 better defensively? Too soon to tell, but to this Corsi-aware blogger it looks like MP85 was sent down because he could be and not because he is an inferior defensive player.

  • strungout

    So how many of those total minutes for each player were shorthanded?

    That right away should tell you who is better defensively.

    Stupid corsi nonsense.

  • CDizz

    There is no doubt in my mind Perreault should be a full-timer. What he lacks in size, he makes up for a hundred fold in skating ability, playmaking, and an overall incredible hockey sense. It’s a shame that we have a full roster, but if Johansson is really as good defensively as he’s hyped to be then maybe it’s for the best. God knows we could use some strong defensive forwards.

  • KHtaD

    Okay, so some things I’d like to know before interpreting this article:

    Is all the data 5v5? Johansson played significant PK minutes, Perreault did not.

    What were the score effects for each player? Did one get significantly more ice at a +1 score differential, harming their Corsi?

    What was the quality of competition like?

    What were the quality of teammates like? I seem to remember Perreault centering AO (a noted Corsi monster) for two games, while Johansson got a game with Semin and third or 4th line time with heavy PK duty otherwise.

    A final note: Corsi scoring chances. The ratio converges in the long-term, but definitely not over a single game. You can define scoring chances in that way, but the rest of the statistical community has been using a different definition.

  • Neil – RMNB

    I should have clarified that all the numbers (TOI and Shot differential) are all 5v5. That’s what happens when Ian rushes me after Sons Of Anarchy.

  • Sean

    I really don’t think a 5-game sample size in the pre-season is adequate statistical data. Especially given the nature of the pre-season to shuffle lines (as frequently pointed out, MP playing with Ovie, for instance) and a higher variance of competition (everyone from current superstars to guys who end up in the juniors are playing).

  • McKinley

    I like both players but I wonder if Matty’s “energy” doesn’t make him more of a fan favorite than the more conservative playing Swede. Not sure this team is well served if all the centers are Backstrom clones but Perreault isn’t the gritty or big center alternative either. Whatever reason MJ got the nod, we should have learned from previous years that the opening day roster isn’t nearly as meaningful as the roster during the season. Besides, offense is not this team’s concern. No need to second guess just because we can.

  • NBTW

    Sample size, sample size, sample size.

    You really think 60-odd minutes of ice time is an adequate sample size to start making statistical conclusions?

    Watching the players, it’s pretty clear to this fan who the better defensive player is now, and it’s not MP.

    And another note: down with Corsi. I’m starting the movement. It’s a garbage statistic that tells you nothing about a player’s defensive value. Case-in-point: Corsi ratings will tell you that Daniel Alfredsson (and his -8.4 relative Corsi) is one of the worst defensive players in the game.

  • http://www.russianmachineneverbreaks.com Ian Oland

    What I take from Neil’s story is that Bruce’s reasoning to the press for why he kept MoJo is flawed. It’s clearly deeper than just “defensive upside” issues and I think Neil did a great job of illustrating as such. If this was an open competition, it should be assumed that Bruce’s decision was based primarily off the sample size that you guys are complaining about.

    For me, it’s more evidence that MoJo’s roster spot was decided long before training camp. And there’s nothing wrong with that. He’s going to be a very, very good player. But this mantra that there are roster spots up for grabs for anyone in training camp seems to be a farce. Especially for the Hershey guys not named Carlson and Alzner.

  • NBTW

    There’s a difference between making scouting conclusions and statistical conclusions based on a small sample size. Bruce did the former, which I have no problem with. Neil did the latter, which isn’t useful in the slightest.

    Again, what I saw with my own eyes tells me that Mojo is the better defensive player right now. The statistical analysis doesn’t contradict that, or Bruce’s comments, because the sample size is too small to make the statistics meaningful.

  • iwearstripes

    “But this mantra that there are roster spots up for grabs for anyone in training camp seems to be a farce.”

    Considering half the roster spent time in Hershey on their way to the big club this hardly looks like an organization that thumbs its nose at players working their way up from the AHL.

  • Sean

    @Ian
    BB also had the ability to work with both players in practice and scrimmages, something that isn’t reflected in these stats. That’s already a larger sample size.

    Statistics should represent a players’ abilities over the long term, but in a short window such as the pre-season, they’re not always going to. You can, and players frequently do, have bad stats despite playing well or good stats despite playing poorly.

  • http://redlinestation.blogspot.com RedArmyLine

    @NBTW: Context is everything. Alfredsson faces the toughest minutes on Ottawa with, if I recall correctly, tough zonestart as well. Of course his Corsi won’t be great.
    Read Behind the Net, the Copper and Blue, and Irreverent Oilers Fans.

    Also,
    “So is MJ90 better defensively? Too soon to tell, but to this Corsi-aware blogger it looks like MP85 was sent down because he could be and not because he is an inferior defensive player.”

    “Too soon to tell” sounds to me like just noting what the trend of the results thus far have been, not a conclusion.

    Neil,

    I don’t think scoring chances and Corsi are interchangeable in such small sample, and the fact MP played with AO is critically important since AO is one of the best Corsi forwards in the NHL (as far as I know only Datsyuk and Zetterberg are better). The Capitals actually kind of lack such Corsi-driving forwards; AO, Backstrom, Semin, and Fehr all get the job done, but aside from them guys like Laich, Flash, Chimera, and the fourth line don’t really drive the play as much and as such Johansson’s Corsi would suffer mightily playing with them.

  • http://www.russianmachineneverbreaks.com Russian Machine Never Breaks

    @NBTW “down with Corsi. I’m starting the movement. It’s a garbage statistic that tells you nothing about a player’s defensive value. Case-in-point: Corsi ratings will tell you that Daniel Alfredsson (and his -8.4 relative Corsi) is one of the worst defensive players in the game.”

    Like any statistic, it provides context – and you discard it at your own peril. As for Alfreddson: he is 37 and clearly on the downside of his career. His Corsi value from last year is not good but that doesn’t mean we discount his other 13 seasons of play.

    “the sample size is too small to make the statistics meaningful.”

    I would argue that I used ALL the preseason data available to make a conclusion. Granted, in a perfect world I would love to have all the data for SEL, AHL and NHL available, but that wasn’t possible. Would I use 60 minutes of 5v5 TOI to project the careers of centermen? Nope, but does it provide that much more clarity of how they used their opportunity.

    @RedArmyLine “the fact MP played with AO is critically important since AO is one of the best Corsi forwards in the NHL”

    The fact MP85 played on the 8-22 line AND had positive Corsi means something. Does it mean he is a top line player? No, but it *does* mean he didn’t drag the line down. MJ90 dragged a line of Fehr/Chimera and Fehr/Laich down to negative numbers.

    Again, it is important to remember that this is pre-season and stats like these only provide context. Am I saying Perreault is going to be the next coming of Datsyuk? Of course not. Am I saying he will be a better long term player than MJ90? Not at all. All I am saying is that based on the preseason numbers MP85 should have gotten a shot (at the very least over Hendricks – how many 28yo rookies amount to anything? Mark Streit and ????) but didn’t under the guise of MJ90 being better defensively long term.

    -Neil

  • NBTW

    Don’t like the Alfredsson example? How about Sami Pahlsson? In 2008-09, he had the 4th worst relative Corsi amongst regular forwards. Does that mean he’s a bad defensive player? Of course not. Those are two of the best defensive forwards in the game, despite their age.

    Hell, Marc-Andre Bergeron was in the top 10 amongst all players with 10+ minutes of 5-on-5 and 60+ games in 2008-09. Peter Regin and Carlo Colaiacovo were second and third this past season.

    Many discount +/- as a relevant hockey statistic, why not Corsi? Because it has a fancy name?

    And no, the Corsi ratings for MJ and MP this preseason don’t give any context whatsoever, because the sample sizes are statistically meaningless. I know I sound like a broken record, but it’s the major flaw with the analysis. It’s like trying to determine who’s the better hitter by looking at the statistics for 15 at-bats in spring training.

  • SantaC

    Jesus what a joke of a blog post. MJ is 2 years younger than MP and has much better upside. Of course you should go with the more talented player.

  • Neil – RMNB

    I will leave it at this: Corsi is helpful. It doesn’t tell us who wins the game (Goals For/Against does), but it is a VERY solid indicatior of where the game is being played.

    And you cannot possibly disagree that a team’s chances of winning improve if they’re playing more of the game in the offensive zone.

  • strungout

    Corsi is not, in any way, helpful.

    For you doing whatever it is you do? Maybe. But for real NHL teams making actual decisions on playing time and player development…utterly useless.

  • http://redlinestation.blogspot.com RedArmyLine

    @NBTW

    Notice that Sami Pahlsson consistently plays some of the toughest minutes in the league. He’s exclusively used in a checking role with other checking line players.

    Marc Andre Bergeron plays some of the easiest minutes in the league, with easy zonestart as well. You’re plucking Corsi out of thin air, without looking at quality of competition, quality of teammates, and zonestarts.

    People discount plus minus because it’s without context. Corsi has context, but you’re being blind to it right now, and hence it seems bad to you. Find me a real example.

    Also, you’re not understanding that Neil just said thus far Perreault seems to be better. He’s not drawing any conclusions–just pointing out the short-term trend. It’s exactly the same as calling Jordan Eberle “clutch.”

    @SantaC–that reasoning would put 18-year old goalie Jack Campbell as the starting goalie for the Dallas Stars right now.

    @strungout–the Buffalo Sabres were the ones who started keeping track of Corsi. I’d say they think it’s helpful. The NHL teams probably also have loads of video data and advanced stats we don’t have.

    Corsi is not a causation of anything. It’s an indicator that correlates pretty strongly to winning. Once people like you get over this phobia of all things non-traditional then you’ll see how useful it is. It showed me that Marc Staal is a top-5 shutdown defenseman, Duncan Keith is flat-out the best defenseman today, that Tomas Fleischmann is overrated, that Eric Fehr is a good hockey player.

    @Neil–it’s pretty much impossible to drag Alex Ovechkin down. Brooks Laich, on the other hand, doesn’t drive Corsi; he’s more of a passenger (and to be fair, so are guys like Holmstrom and Knuble).

    On the other hand, you’re probably right that MP is better than MarJo defensively right now. I don’t know if you’ve read Copper and Blue’s “Unrealistic Expectations” piece, but it’s worth a read if you haven’t.

  • Sean

    @RedArmyLine
    Corsi has no more context than +/-. You can give Corsi some context by looking at other stats, the quality of competition, etc., but you can also do the same with +/-.

    I love (and I mean love) statistics. I really buy into most baseball sabermetrics. Corsi has its uses, but any hockey statistic (traditional or otherwise) must be taken with a grain of salt. Events in hockey aren’t isolated, but have tons of variables interacting at once, much like most defensive metrics in baseball (thus why articles on fangraphs often also reference the fans’ scouting report to corroborate any defensive metrics they use). Over the long-term, things should normalize (though players’ abilities, linemates, and match-ups are not going to remain static), but short-term stats like this is pretty much worthless. Even to try and note “short-term trends.”

    The “trend” is that Roy Halladay is not only the greatest playoff pitcher of all time, but a better hitter than everyone on the Cincinatti Reds.

    Heck, the numbers don’t even support the argument that MP85 is better than MaJo defensively. Both gave up 53 scoring chances (and MaJo even played 3 more minutes of time), so the defensive aspects of Corsi were the same. It was only in the chances generated (the offensive aspects), that MaJo lagged behind.

    I realize this is all the data that is available, but at this point the data set is just too small to have any value. A 3rd line center is probably going to play around 850-1000 minutes over the course of a season. Using a 60-minute sample size (even in the regular season, much less the pre-season) would be the equivalent of using roughly 2 starts to analyze a starting pitcher in baseball. It’s just not going to provide any meaningful information.

  • Pingback: Caps beat Devils 7-2 as the Circus Comes to Town