It’s a scene anyone who watches NHL hockey has seen: two players squaring off, dropping the gloves, and going at it. One player might dominate the other, the two players might draw more or less even, but after the fight the teams will generally get back to playing 5-on-5 hockey (barring an instigator penalty to one player.)
Usually, one team will come out playing better after the fight. And usually, the commentator that evening will make a point of mentioning that Player X’s big fight win and/or willingness to go toe-to-toe with a beast like Player Y has given his team momentum.
Is there any truth to that story?
According to Terry Appleby of Power Scout Hockey, fights do swing momentum. Appleby measures ‘momentum’ by recording how often NHL teams take shots – if a lot of time passes between shots, they have less momentum; if they get a lot of shots off in a short time period, they have more momentum.
Appleby’s full piece on fighting is here. It has two key problems, both of which he recognizes – and since he recognizes them, I’ll use his own quotes to highlight them.
If a team’s fights occur without any regard to team or opponent momentum, then the team will score at an average gain of about a 0.1 of a goal. If so, this would require about 60 fights to score a win. But good coaches don’t do things randomly – preferring to pick the opportune time to change team momentum. In such cases scoring will double to about 0.2 of a goal and the team will require only about 30 fights to get a win.
Appleby notes that a team will pick up an extra win, on average, for every 60 fights on their team. This number drops to every 30 fights if a coach is judicious about when he uses his fighters. This, however, is contingent on ignoring the impact on the opposing team – because Appleby finds the momentum swings both ways; in other words, both teams shoot more after a fight. The extra 0.1 goals the team might get after a fight, in other words, are generally cancelled out by the 0.1 goals the other team will get after a fight.
Here’s the second problem identified by Appleby:
For interest sake, further study into the impacts of fighting on team momentum could add more context to these results. Other factors that could influence the fight/momentum analysis include 3 minutes after the fight versus other time ranges, who won the fight, the score, period, shots by line, powerplay opportunity, time in period, time of year, and quality of opposition. These will be explored with future research depending on availability of the data required.
There’s lots of other things that happen at the same time as fights. Most fighters are fourth line guys – and most fourth lines don’t generate a lot of shots. How much of the shot increase is generated by the fact that the fourth-liners are off the ice and a team’s skilled players are on it? Some fights come with other penalties – Appleby’s piece seems to indicate that they haven’t allowed for the impact of penalties on shot volume.
Let’s look at the example Appleby used – a December 7, 2011 fight between Buffalo’s Matt Ellis and Philadelphia’s Marc-Andre Bourdon. According to Appleby, “a fight between Ellis and Bourdon at the beginning of the second period sparked an immediate increase in momentum for Philadelphia and a subsequent goal by Matt Read.” Looking back at the game log for that night, we see that Ellis was assessed an instigator penalty for picking the fight with Bourdon after Bourdon took a boarding penalty. Yet, Philadelphia walked away with the ‘momentum’ while Buffalo remained flat-lined – despite the fact that Ellis did what is typically considered to be the right thing in standing up for a teammate after an illegal hit.
(Also interesting in that same game is the fact that another fight – between Corey Tropp and Zac Rinaldo – at the end of the first period seems to have had a negative impact. Both teams were pressing heavily in the first, but after the fight both teams came out flat in the second period.)
Fortunately, there are other studies into the impact of fighting. Statistical wizard Gabriel Desjardins tried a different approach, looking at only fights with a clear winner and then checking to see how scoring changed for the team that won the fight. The full piece is here, the conclusion is quoted below:
Now before you get excited about your favorite team signing a new goon, remember that an NHL team needs to improve its goal differential by approximately six goals to win one additional game. So winning a fight is worth a little more than 1/80th of a win in the standings; given that the best fighters might win at most ten fights in a season, the direct benefit is probably on the order of having the equipment guys make sure nobody’s playing with an illegal stick.
Then there’s Phil Birnbaum’s post at Sabremetric Research. Inspired by Appleby’s piece, Birnbaum runs through a long list of simulations and reviews truckloads of game data (the full post is well worth reading if you have any tolerance for math), and then comes to the following one-line conclusion:
At best, there might be a small effect in certain specific circumstances … but much, much less than sportscasters make it out to be.
My personal take on all of this is primarily based on the evidence presented in the three posts above, watching a bunch of hockey, and some basic knowledge of how confirmation bias works. It can be boiled down to a few key points:
1. In some cases, it’s all but impossible for a coach to know if a fight will benefit his team or the other team more on the scoreboard – it’s always a double-edged sword.
2. The impact of a fight on the scoreboard is generally pretty small – we all remember fights that were quickly followed by a goal, but tend to forget the ones that weren’t.
3. Given the relatively low impact and the chance of a negative impact, a coach with a fighter at his disposal would be well-advised to use him if a) his team is already trailing in the game and b) he’s reasonably confident that his fighter will win the bout. A coach with the lead or a fighter outmatched by the other guy is probably better off leaving him on the bench.