North London Derby projections and stats, plus a discussion of home field advantage

I wrote my first post here at Cartilage Free Captain in advance of last spring's North London Derby. Since then Spurs have gone from favorites for the Champions League to the Europa League group phase, and now the club has turned over whole squad. I've gone from underemployed doctoral student to a still-underemployed doctorate-holder with an increasingly involved hobby of writing at a popular soccer website. So some real stuff has gone down, is what I'm saying.

Now we've got another NLD on Sunday. I'm projecting Spurs will lose. You're welcome.

Basically, this is because of home field advantage. My numbers had Spurs as a better club than Arsenal in the pre-season, and the performance of both clubs so far has not changed my ratings hardly at all.

Home Field Advantage: An Unnecessarily Academic Discussion

The problem is, playing away and playing at home are totally different propositions. There were 573 goals scored by home teams in the EPL last year and only 443 by away teams. Home teams put 1920 shots on target, away teams 1564. 1316 SiBoT for home teams, 1040 for away teams. Over the last several seasons, home field advantage in the English Premier League runs in the range of 25% for both goals and expected goals.

I think Spurs are better than Arsenal, but are we 25% better? That's a lot. As a note, the numbers for the top clubs are no different, that ~25% HFA is pretty constant up and down the table.

There are a lot of theories as to what causes home field advantage, which is seen to greater or lesser extents in basically every professional sport. There is actually an extensive academic literature on this subject, because academics like sports too. (Someday I'm writing a book on fandom and the ancient arena.)

There's a good publicly available meta-analysis by Jeremy Jameson in the Journal of Applied Social Psychology. There have also been some Freakonomics-style analysis of the home field advantage which have put forward the Levittastic extreme and headline-grabbing thesis that home field advantage is *only* a function of refereeing. (See Moskowitz and Wertheim, Scorecasting) The much more measured discussion by Jameson demonstrates the existence of a variety of factors, including both teams' familiarity with the stadium, which are clearly not just a function of refereeing. (For a good public response to the Scorecasting chapter, see Phil Birnbaum's review.

While I have my issues with the Freakonomics style of ignoring complexity in the academic literature and pretending contrary data does not exist, the evidence that referees play an outsized role in HFA is nonetheless quite impressive. For instance, on average home teams score about 25% more goals and put about 25% more shots on target. Home teams are awarded about 50% more penalties. You would expect the more attacking club to get more penalties, but an effect twice the size of the effect on shooting and scoring suggests strongly that the referees are swayed by the crowd. A study of added injury time in major European Leagues found that referees added on average two more minutes of extra time when the home team was trailing than they added when the home team was leading. (Scorecasting, p. 140)

These facts don't demonstrate that home field is *only* about referee effects, but it's clear they play a significant part. We should not expect the consistent reality of home field advantage to just disappear when Spurs travel to the Emirates on Sunday.

Projections for Sunday

I have run my projection program for Sunday's game, and I have us losing about 40% of the time. But I also have us winning about 30% of the time. That's pretty close. I have also run the numbers for how different outcomes of this match affect both Tottenham's and Arsenal's projected points and chance of finishing in the top four. The basic story is that a win would certainly make a difference for both clubs, but no one is going to be particularly crippled by a loss.

Tottenham would fall back to roughly even with Arsenal in the projected table with a loss. (Since an Arsenal win would probably also require Arsenal to outplay Spurs, likely their projection would be a little better next week than I have it here.) A Tottenham win would put our chances at finishing top four into a pretty nice place. But as I have said before—indeed, as I said after our last NLD win—a 25% chance is the chance of a coin coming up tails twice. You've seen that happen all the time. It's hardly unlikely. So a win would be nice, and loss would be annoying, but nothing is going to be determined in the third match of the season.