Friday, January 06, 2012

Fighting...Does It Matter?

Fighting is as much a part of hockey lore as the hat trick. You can debate whether it has a place in the game – that argument is as much a value judgment as anything else. But there is a parallel belief that fighting in the sport influences results. The latest exhibit comes from powerscouthockey.com, who engaged in some research “to seek out an answer to the only important issue about fighting in the NHL; does it matter?”

We have to give it to the folks who engaged themselves in the research, they certainly dove deep into the data, researching more than 1,500 fights going back to the start of the 2009 season. Their preliminary findings were two in number:

- A fight INCREASES momentum from one or both teams 76% of the time, but in only 34% of fights does one or both teams LOSE momentum.

- A fight INCREASES momentum from both teams about 1 out of every 4 fights (23% of fights), but rarely do both teams LOSE momentum (only 4% of fights) (emphasis the authors’).

To which we say…”so what?” Let’s go back to the original question the researchers asked, “does it matter?” This is essentially a question about outcomes. And in any competitive enterprise, the issue underlying the question “does it matter?” is not one of whether momentum is or is not influenced, but whether the outcome – which side wins – is influenced.

Wins, accumulated in sufficient number, will qualify a team for post-season play. So, do playoff-eligible teams rack up fights (if our hypothesis is that fights influence the outcome of wins and losses)? Well, we do not have to sift through reams of data to find an answer to that question. We need only look at how teams are ranked in fighting majors and identify those teams that were eligible for post-season play (the desired outcome).

We did just that for seasons starting with the 2007-2008 campaign (our decision criterion for which season to start with: laziness – we did not care to go back further). It is easy to display the results graphically. For the sake of general comparison, we broke the rankings into thirds – the top-ten fighting teams, the middle ten, and a lowest ten (playoff teams in yellow, eventual Stanley Cup winner in red).

In 2007-2008, we have the following:


Five teams in the uppermost third, six in the middle third, five in the lowest third, including the champion Detroit Red Wings finishing 30th in fights. In this instance, fighting did not appear to influence an “outcome,” if by the term you mean wins and losses. Detroit, Montreal, and Washington, among others, seemed to do fine without the fisticuffs.

In 2008-2009:


The numbers, by thirds, are: eight, three, and five (including, again, the eventual Cup winner). And, as was the case in the previous year, neither of the Cup finalists were ranked highly in fights over the course of the regular season.

In 2009-2010:


Seven of the 16 playoff-eligible teams were in the bottom third of fights, and 11 in the bottom half. True, Philadelphia (the Stanley Cup runner-up) finished second in the fight rankings, but fighting as a rule was not a path to wins in totals that would make teams playoff-eligible.

In 2010-2011:


Half of the playoff-eligible teams were ranked in the bottom third of the fight rankings, including one of the Stanley Cup finalists. Boston – the Stanley Cup winner – ranked second, and one might use this as an argument that fighting works. But fighting is not generally employed as a tactic (or even as a reaction) in the playoffs. Boston led the league’s playoff teams with five post-season fights, but that is perhaps as much a product of having played more games (25) than anyone else, except Vancouver (which finished second in fights with four and played 25 games). Two of Boston’s fights came in the first round, one each in the three rounds that followed. Vancouver had two of their fights in the first round, one in each of the last two rounds.

In the 2011-2012 season to date, here are the results:


The numbers by thirds break out into five, six, and five. While this does not provide any conclusive answer to the fundamental question, teams should by now have established a certain personality (with acknowledgment that personnel moves that might be made later might alter that personality somewhat). Fighting does not appear, at least so far this season, to have influenced outcomes in terms of whether it helps teams win and thus achieve playoff-eligible status.

This is the beef we have with the growing field of sports statistics. There is insufficient attention to the relationships between measures and outcomes. The measures are too often “output” related, and inadequately developed to demonstrate how those measures influences wins and losses. We believe this is a product of the relatively young state of the art, not a comment on the methodological aptitude of those trying to advance that art. And we do not want to lump those who have done much to improve the state of the art in with examples such as this.

But let us try to ask the right questions, too. If, in this instance, we are trying to ascertain if fighting influences a result, should the question be whether or not fighting influences (or is at least accompanied by) wins? Having read the preliminary research here and comparing it with win-loss outcomes, we are still left to ask, “does it matter?”


(thanks to a tweet from Neil Greenberg that pointed us to the research...although we do not get the feeling he endorses it...edit: OK, actually, the article by Jim Litke in the Washington Times that pointed us to the research)