Sunday, July 27, 2008


Fighting and Wins in 2007-2008

For some reason, late last night, I got to thinking how the Devils generally do when there's a fight during a game. Yeah, I don't know it popped in there, but it did. So this morning, I decided to look at all the games last season and see how many fights there were and whether or not there was any connection between it and winning.

To do this, I went to and check out the basic recaps for each game. Each recap has a list of penalties, so finding fighting majors and the score (of course) was fairly simple. Just as important as the number of fights there were, when they time they took place. I don't like fighting and I really wouldn't miss it if it were eliminated from professional hockey. That said, there is a train of thought that a fight would "wake a team up" so to speak. It does get the home fans all up in an uproar, I can't deny that. Whether or not it changes the momentum of a game is another question. What you, I, and your favorite auntie cheer or not cheer for is immaterial. The goal of a hockey game is to win, and that will be the focus of this then.

So in addition to counting the number of fights and seeing whether the Devils won, I subjectively decided whether that fight made a difference in the game. I would like to emphasize the word difference here. If the lead changes or a team ties it up after a fight in the same period, I would recognize the fight could have had an affect on the team's performance. Were the Devils losing, a fight takes place, and the Devils remain losing - I don't think the fight changed much of the game's course. I limited it to being within the same period only because an intermission has additional variables come into play when it comes to a team's performance - a leader making a speech, the coach giving the team what's for, and so forth. A fight in the first period, I think, isn't likely to be a catalyst or part of one when the Devils score twice in the third period to win the game. Also, it's not to say, "Well, David Clarkson threw down and the Devils scored 2 goals afterwards; the fight was totally responsible." I'm saying it could have. It's a subjective argument, but I felt it could provide an interesting insight. I understand it's flawed because a game could change in momentum but not yield any goals; however, there's no clearer way of showing that team "turned it around" or "turned it up" other than lighting the lamp.

I also split the difference between all fights and "in-game" fights. I'm defining a fight to be in-game is if it took place within the normal course of a game. Some fights/scrums/etc. take place at the very end of the game when it's all pretty much decided. Fists are thrown out of frustration more than anything else and it ultimately has no affect on the game. So I decided to only really count those in any analysis. Fights at the beginning of the game or in the middle of it would change any momentum on the one place that matters - the scoreboard. At the end of the game, not so much.

Fourthly, I'm only looking at this past season. Before you comment or e-mail me that "How dare I ignore Cam Janssen," I want to point out that the Devils committed 16 more major penalties this past season than in 2006-2007. While not all major penalties are fighting majors, the Devils fought enough times this past season such that it is mathematically impossible for it to be fewer than 2 seasons ago - regardless of "Cammer The Hammer." So much for cries that the team last season lost its "toughness."

Lastly, I'm only counting wins here. Shootout and overtime losses yield points, yes, but they aren't wins. And wins are what's important here. Anyway, there's no need for fancy charts or graphs because the numbers are straight forward:
  • Number of Devils fights: 45
  • Number of games with a Devil fighting "in-game": 32 (34 including the end-game fights)
  • Number of Devils fights "in-game": 42
  • Number of fights that changed the course of the score: 13
  • Devils record when they don't fight: 26-22 (Winning Percentage: 54.16667%)
  • Devils record when they fight "in-game": 19-13 (Winning Percentage: 59.375%)
  • Devils record when they fight but not "in-game": 1-1 (Winning Percentage: 50%)
  • Devils record when the fight changed the course of the score: 6-7 (Winning Percentage: 46.1538%)
Well, my original thought going into this was that fighting really wouldn't have a correlation. And I'm proven somewhat wrong. The Devils actually had better winning percentage when someone in red does throw down their gloves this past season. Not that they did bad without fighting - but the percentages are pretty clear about that. It's also not as if there's a small percentage of fighting; a good 93.3% of these fights are during the course of the game where things can change and approximately 39% of all Devils fights have this kind of fight.

That said, it's not that drastic of a change in winning percentage for fans to demand a fight every game. Of these 42 "in-game" fights, only 13 of them led to a significant change in the game. The Devils ended up slightly under the 50% winning percentage mark, as only 6 of those changes ultimately led to Devils wins. Therefore, most of the Devils fights do not led to a change in the game and when they do, it's almost a coin flip as to whether it works out for New Jersey.

Going into this analysis, I was under the impression that A) the Devils didn't fight all that much and B) the fights don't lead to too many wins - too many of them take place after the game's over, the Devils would likely continue sleepwalking through games regardless of the fight, etc. Thought A was quickly proven wrong, a point further justified when you consider New Jersey was in the middle of the league for major penalties. Thought B was refuted by the Devils happenstance to win 19 games out of the 32 where there were fights during the game. Only 2 fights were at the end of games; and there were even a couple that had multiple fights (most per game this season was 3, achieved three times).

However, it doesn't seem that the Devils fighting lead to the Devils playing better and more productive hockey from this past season. They are mostly just fights that happen in the game. This makes sense, otherwise, we'd see teams all over the league bend over backwards to keep enforcers, tough guys, and goons along for long contracts. In short, I don't think the Devils - or its fans - should be looking for fights to play better hockey - it likely will not happen. Considering most of the fights are by one David Clarkson, a young talented winger, I still think it's best if he's on the ice on shifts that would take place when he's not in the box for 5 minutes.

Labels: , , ,

Very interesting analysis. I've wanted to do something along these lines, but never took the time.

Not quite sure what your final verdict is, though. You say you were surprised as to how many games they win when they fight and how many change the course of the game, but then at the end you say it doesn't really matter. Which is it?
Well, I wasn't surprised as how many changed the game - just that the Devils went 19-13 in these "in-game" fights. 13 out of 42 isn't that impressive.

Ultimately, my point is that it likely doesn't seem to matter. There may be a correlation and the numbers do show a better winning percentage; but due to the fact that there were more games without fights than with them and that the fights didn't seem lead to a change the score of the game during that period. I still don't think the evidence isn't strong enough to say "fights are likely lead to wins."

If there was a better way to gauge momentum than scoring as well as a way to account for the differences in data (34 games w/ fights, 48 w/o), I think there'd be more evidence. I still stand by my ultimate conclusion is that the team shouldn't be looking for fights to play better hockey.
It would be interesting to take a closer look at the penalty that comes with the fight (the extra two that one team might get for instigator penalty), as any goal scored during that power play is a direct consequence of the fighting. Moreover, it would be interesting to use this instigator penalty as an indep variable to run a regression to make your point.
Post a Comment

<< Home
Blog Directory - Blogged