By Andrew Puopolo
This past weekend marked a major derby match on the Italian soccer calendar, with Roma “hosting” Lazio at the 60,00 seat Stadio Olimpico. For those unfamiliar with Italian soccer, Lazio and Roma share the Stadio Olimpico, with both teams playing all their home matches at the stadium. It is well established that playing in your own stadium provides a significant advantage, but what if your opponent is also playing “at home?” When the two teams meet, they are both playing in their home stadium, but for this match Roma were the designated home team. This means they got to change in their own changing room, had the adverts around the stadium and got a larger percentage of the tickets available to fans. In this particular match, Roma defended their “home advantage” and defeated Lazio 2-1, but how does the effect of home field advantage in this situation hold up in general?
When people generally think about home field advantage, there are multiple benefits that the home team enjoys. The first is that teams get to play in front of their home fans and benefit from the crowd support that the fans provide. The presence of partisan fans can also influence referees to unintentionally give more beneficial calls to the home team. The second main advantage is the familiarity of surroundings that the home team enjoys. Players get to sleep in their own beds, do not have to travel, and are playing on the same field that they play half their matches on.
This blog attempts to test which of the two factors contributes more to home field advantage. In ground-share derby matches, the first advantage sticks but the second advantage does not. Specific details on ticket allocation for these matches is incredibly hard to come by, but for the Milan derby between Internazionale and Milan, the “visiting” team is allocated the typical away allocation provided to all teams, and the tickets for the rest of the stadium are sold by the “home” team. This means that the “home” team is primarily playing in front of their own fans. However, both teams are playing in familiar surroundings (it is both of their home stadiums, after all) so that advantage is eliminated.
In Italian soccer, there are multiple instances of teams sharing the same stadium, and we can use the results of these derbies to isolate the impact of a partisan crowd. We used the following five derby matches in Serie A, dating back to the 2001-2002 season for this analysis. The 2005/6 season was omitted from this analysis as a result of the Italian match fixing scandal.
AC Milan vs Internazionale (15 seasons)
AS Roma vs SS Lazio (15 seasons)
Genoa vs U.C. Sampdoria (9 seasons)
Juventus vs Torino (4 seasons, only until 2010/11 when the teams stopped ground-sharing).
Chievo Verona vs Hellas Verona (4 seasons)
For those unfamiliar with the structure of European Soccer, each league has 20 teams (Serie A had 18 teams until the 2003/4 season) and every team plays its opponent twice (once at home and once away). We wanted to look at how these teams performed against each other in their derby matches, and if being the designated “home” team had any positive effect.
For each team in each season (for a total of 94 rows), we computed a “derby home field advantage” proxy statistic: the number of points in home match minus the number of points in the away match. For example, in the 2015/16 season, Inter defeated AC 1-0 at “home” while in the reverse match, Milan won 3-0. This resulted in the “home field advantage” statistic being equal to 3 for both teams. In 2016/17, the two teams drew both matches, so the statistic was equal to 0 for both teams.
We wanted to test to see whether, on average, teams glean less advantage from playing at home against sides that share their stadium than they do against other teams. To test this, we bootstrapped a null distribution. We did this as follows:
Step 1: For each of the 94 teams in the dataset, pick a random opponent in the league (not the ground-share opponent).
Step 2: For the randomly chosen team, compute the “home field advantage” statistic for that season described above.
Step 3: Compute the average “home field advantage” statistic for each of the 94 randomly chosen opponents in that particular season.
Step 4: Repeat Steps 1-3 10,000 times and store each of the results.
For the dataset we are using to test, we find that the average home field advantage statistic is .12. This means that for the 94 teams in our dataset, on average they collected an extra .12 points in their “home match” against their ground-sharing neighbors than they did in their “away match.” This is despite both matches being playing in the same stadium.
Our null home field advantage statistic is distributed as shown in Figure 1 with a vertical red line at .12 to indicate the value of our test statistic. This null distribution had a mean of .56 and a standard deviation of .17. This means that on average, Serie A teams win an extra .56 points in their home match against an opponent than they do in their away match.
Only 40 of the 10,000 generated null distributions have a home field advantage statistic of less than .12. Thus, with a p value of .004, we can conclude that there is less home field advantage in these ground-share derby matches compared to regular home matches.
But does the home field advantage disappear completely? Surely there are some benefits to being the designated home team. We decided to run a t test just on the variable “DerbyAdv” and obtained the following results:
Here, we find that there is no statistical significance to suggest that there is any home field advantage when playing your ground-share partner “at home”. This is not entirely surprising either, but is less obvious than our first result.
In conclusion, we find that the home field advantage in these ground-share derby matches is substantially reduced, and possibly eliminated altogether. Therefore, our results suggest that the advantage gained by playing in familiar surroundings and not having to travel appears to outweigh the advantage of playing in front of your own fans (at least, for these 10 Serie A teams).
One final thing to note is the quality of the teams. In the case of the Rome and Milan derbies, all four teams are regularly amongst the best teams in Serie A. This means that matches between those sides are always extremely competitive, whereas matches against other teams in the league might not be. This difference might affect our results a little bit.
What do you think about this? Is there a similar effect when the Lakers play the Clippers at the Staples Center or when the Jets play the Giants at MetLife Stadium? How about for other European soccer leagues? Let us know in the comments below.
Editors Note: If you have any questions or comments about this post for Andrew, feel free to reach out to him at andrewpuopolo@college.harvard.edu.