Winning a Fight Has Impact on Future Outcomes

Travis Yost
August 04 2013 09:45AM

 

http://a.espncdn.com/photo/2013/0304/mag_a_nhl_sharks_canucks_576.jpg

 

INTRODUCTION

Back in July of 2009, Gabriel Desjardins at PuckProspectus tried to identify whether or not winning a hockey fight could be the impetus behind a boost in team scoring. He found virtually nothing.

More than four years later, I've found a much more significant set of data. The numbers may surprise you.

METHODOLOGY

I used a slightly different approach from Desjardins, mostly because I really wanted to target the league's fighting-specialists who are mythologized as on-ice intimidators and certifiable momentum changers by guys like Don Cherry and the Hockey Night in Canada crew.

Sixty-five fighters logged four or more fights last year. I took each player and looked at his 2012-2013 fighting record as decided by the public at HockeyFights.com. The site does generate hundreds of votes per fight, and theoretically can serve as a decent representation of the much-larger hockey-fan base.

I pooled all of the fights won by each of the fighters, then noted whether his respective team scored or conceded the following goal. There were a handful of instances where a fight occurred and neither team scored for the rest of the game -- as such, these instances aren't included in the data.

How did they fare in the one-hundred and seventy-seven fights collected?

 

 

So, just to be clear here: if you're an enforcer / fighting-specialist and beat the living hell out of some poor guy in a hockey game last year, there was a better than 54% chance your team was going to concede the next goal.

Colton Orr, Mike Brown, Jared Boll, Brandon Prust -- a bunch of guys who fought last year, and a bunch of guys who saw the opposition go on to score the next goal more often than not.

Everyone's favorite top-six power-forwards Milan Lucic and Chris Stewart won a combined seven fights in more or less bloodbath-fashion last year. Never once did the Bruins or the Blues, respectively, score the next goal.

CONCLUSION

Anyone with even a rudimentary understanding of how little something like fighting should have on future outcomes can see that this 54%-to-46% data set is nothing more than getting lucky (or, unlucky) on a series of coin-flips. Realistically, the number should be just about 50%-50%. The methodology in this study is questionable, too. My guess is that an even-more rigourous aproach, with score-effect adjustments and the lot, will turn up what everyone expects: there's truly nothing to see, because fighting is irrelevant to future performance.

But if you're Nick Kypreos and still under the impression that fighting can impact games (and, more specifically, goal-scoring), there's your answer. It absolutely can. For the other team. And the goal for all fighting-types should be the same from here forward. Start a fight, but take a dive. Lose at all costs. If you're a gritty, heart-and-soul guy looking to sacrifice your well-being for the success of your hockey team, wear a flurry of rights on your face. Do it, and you'll have a 4% better chance at scoring the next goal. Also, enjoy the concussion that comes as a proximate result. Those can end well.

For the sickly-types interested in the inputs for the above, I've opened a Google spreadsheet with all of your favorite fighters and included it below.

 

 

73100a182a18dbb0cba9bf8f547ed466
Hockey and hoops. @TravisHeHateMe
Avatar
#1 leafnerd
August 04 2013, 12:49PM
Trash it!
1
trashes
Props
9
props

Good stuff. For all the bias for and against fighting, it is hard to show that fighting is at best immaterial.

Did you exclude PP/PK goals after a fight (instigator rule, or minor penalty etc) and look at 5v5 goals only?

Avatar
#2 leafnerd
August 04 2013, 12:50PM
Trash it!
1
trashes
Props
5
props

@leafnerd

Also, instead of goals can you or someone look at the change in corsi differential before and after a fight. [It is hard to do and I never got around to it]. That would be better evidence as shots are both a more reliable (predictive) and more valid data set (better regressor) and not prone to luck in small sample effects.

That is, after a fight does a team's Corsi differential improve, worsen or is there no material impact. That said, home/road and score effects are always a problem but this would be a more statistically meaningful dataset.

Avatar
#3 gongshow
August 04 2013, 12:54PM
Trash it!
1
trashes
Props
3
props

Thanks Travis.

What are your thoughts on 2 different ways of looking at the data subset that you've already compiled:

1) What happens if you throw out games that are already out of reach - say more than 2 goals apart? As many fights occur in runaway games (which may be more about settling an old score or sending a message for the future), does a won fight in a close game have any impact (of statistical significance) on the outcome?

2) Does a won fight have any statistical impact on the results of a followup game? In terms of sending a message/intimidation, is there any effect?

Avatar
#5 Patrick D. (SnarkSD)
August 04 2013, 02:44PM
Trash it!
0
trashes
Props
2
props

Not surprisingly, the SE of a 50/50 event after 177 observations is 3.75%. Just shy of 4%.

Avatar
#6 beloch
August 05 2013, 05:49AM
Trash it!
1
trashes
Props
2
props

Goons tend to fight goons and an injured (or pained) goon isn't much of a loss. I'd be willing to bet that momentum changes are more strongly correlated with who wins a fight between first-liners. I'd also be interested in seeing data for specific elite players known for fighting. e.g. Iginla has a reputation for getting fired up and playing better after a fight. Do the stats back that up?

If some fights do have a clear impact on games, the use of enforcers might just be a misguided attempt to replicate them. Given that enforcer fights are probably much more numerous than fights between first-liners, looking at the impact of all fights together might also hide the effect of first-liner scuffles.

Avatar
#7 Chuck Miller
August 05 2013, 10:02AM
Trash it!
3
trashes
Props
1
props

If fighting is so important how do the Blackhawks get 45 of 48 possible points to start the season, then win the Cup with the fewest majors in the league?

The league will be forced to outlaw fights because the potential for head injuries, and mega law suits, will be too great.

In the end, Derek Boogaard's estate will have more to say about fights than Don Cherry.

Avatar
#8 Dave R
August 05 2013, 10:44AM
Trash it!
1
trashes
Props
0
props

Hey there - nice analysis! Looking through the raw data I noticed something though - for the Flyers, Rinaldo, Sestito (note: not sure if his fights were with Phi or Van), and Simmonds combined to go 6 - 1 in the "Next Goal" category which made me think, maybe this "Next Goal" thing is not a league wide trait, but instead reflective of the subtleties of the team and coaching dynamics. What I'm saying is even if overall, league wide "Next Goal" balances out to 50/50 maybe some teams are actually good at it and some are bad. Also, maybe winning the fight is less important, and some teams feed off the energy of a teammate putting it out there and some don't... Just a thought and pretty easy to look into I'd imagine. Thanks!

Avatar
#9 gongshow
August 05 2013, 07:59PM
Trash it!
1
trashes
Props
1
props

@Travis Yost

So, how about this idea - do teams with more fighting majors or teams with a nuclear threat in the lineup have a lower incidence of certain types of injury? How about the stars staying healthier?

If not , present day tough guys belong in WWE.

I don't know if the Boys On The Bus Oilers gave much credit for wins to Semenko, McSorley or the other tough guys in the lineup but Wayne and friends seemed to be really happy to have them around anyways. (Granted, there was no instigator rule at that point).

Avatar
#10 gongshow
August 05 2013, 08:00PM
Trash it!
0
trashes
Props
0
props

@Chuck Miller

Well put.

Comments are closed for this article.