+- International Simulation Football League (https://forums.sim-football.com)
+-- Forum: Community (https://forums.sim-football.com/forumdisplay.php?fid=5)
+--- Forum: Media (https://forums.sim-football.com/forumdisplay.php?fid=37)
+---- Forum: Graded Articles (https://forums.sim-football.com/forumdisplay.php?fid=38)
+---- Thread: *Rookies Read This First (/showthread.php?tid=21786)
*Rookies Read This First - 24redcrayons - 04-29-2020
Introduction
Sorry Rookies, but the title was clickbait. However, I hope I can convince you to stay and read this article. The idea of a home field advantage is something that has consistently proven to be a thing in the realm of professional sports. However, while I was looking at the results of the DSFL preseason for season 22, I had a feeling that for some reason, it seemed like home field advantages were a lot stronger in the league than usual. Following my hunch, I decided to do an analysis of the effects of home field advantage on the outcome of a game as well as the possible points scored. Furthermore, since I was combing through the data as well, I decided to include time of possession into what I was looking at and analyzing, so I also found the result of time of possession on the outcome of a game and the total points scored. All in all, I used the preseason data from DSFL season 22 as well as the preseason data and regular season data from season 21, excluding playoffs (week 15). This resulted in 88 games that were analyzed and part of my data set. For my regression, I wanted to create a 95% confidence interval, so the t-statistic that I used was 1.96, meaning that any t-value with an absolute value greater than 1.96 would be considered statistically significant. All of the data was created in Excel and imported into Stata, where 2 multiple regressions were done.
Variables
The key variables in this exercise were as follows: homeaway is whether or not a team was considered the home or away team. The home team was issued a dummy variable of 1 while away was 0. The second variable was totalseconds, which was the total amount of time that a team had possession of the ball, converted into seconds. This means that every single unit of time in this regression is in terms of seconds. The third variable is resultofgame, where a win was assigned a variable of 1 and a loss was assigned a 0.
Outcomes
![[Image: x47dqA3.png]](https://i.imgur.com/x47dqA3.png)
In the NFL, it is generally a given that the home team has some sort of advantage. Numerous statisticians have attempted to characterize this advantage. An article written in 2018 and published in the Chicago Booth Review stated that in the NFL, 57% of home teams win (1). As such, any number somewhere around there would be considered accurate. As we see in this small analysis above, that number was significantly higher last season in the DSFL. Out of 88 total games, 55 of those games were won by the home team, resulting in a 62.5% win rate for home teams. This was incredibly surprising to me, as it meant that somehow, there was a higher correlation between being a home team and winning in the simulation than in real life and there is a higher premium on the home team.
![[Image: rdJ1NKq.png]](https://i.imgur.com/rdJ1NKq.png)
Following this, I did two regressions. The first one was a regression of total points over total time of possession (in seconds) and home/away status. As seen above, home field advantage is also evident in terms of points. According to a CBS Sports article posted in August of 2019, home field advantage has long been equated to an additional three points (2). As such, I was looking for a result that was around there. My regression gave about what was expected, as the home field advantage pointwise in the DSFL is worth 3.39 points based on this data. This, once again, was almost 10% higher than what Vegas and the NFL predict, which means that the importance of home field advantage is once again overstated in the simulation, which is something that could very much impact the outcome of the game. As such, this is important to look at and keep in mind for the future. In that same regression, the number of points was also regressed over the total time of possession. The result of that regression, having controlled for the home field advantage was 0.009, meaning that an additional second of possession resulted in an additional 0.009 points for the team that had the ball. This was interesting to see in person, but not very surprising because it made tons of logical sense. Having the ball more means having more chances to score. Some simple arithmetic shows that this results in one point every 111.111 seconds of possession, rounding to a point every two minutes. Thus, this roughly results in 7 points every 14 minutes, or a touchdown every quarter (roughly). This was just interesting, but I really think that it’s not super significant datawise, because everyone already felt this way. Both of these regressions could be considered statistically significant, because as seen, the t-value for both were over 1.96.
![[Image: 6WURheD.png]](https://i.imgur.com/6WURheD.png)
My second regression was much more interesting. For my second regression, I regressed the outcome of a game over the same two variables, total time of possession (in seconds) and the home/away status. Interestingly, we see that the mere status of being the home team gives an advantage of 0.235 to win. I would also consider this statistically significant, as the t-value was significantly larger than 1.96. The final part of the second regression was the most important in my opinion, as the coefficient was 0.00068. This means that an increase of 1 second in time of possession meant that a team was 0.00068 closer to winning. Some simple arithmetic shows that in order for the away team to overcome the home field advantage, they would need to possess the ball for an extra 350 seconds or so, or around 6 minutes more than the home team in order to have the same chances at winning. Essentially, that would be close to around one possession. Thus, home field advantage is equivalent to a single possession. In order to further contextualize this, I did some math with these coefficients. By setting y equal to 1, I am able to calculate how many seconds of possession would be required in order to guarantee a win, with 95% certainty, for the home team and the away team. Specifically, an away team would need to possess the ball for 2705.55 seconds in order to guarantee a win, with 95% certainty, or 45.09 minutes. For the home team, after accounting for the home field advantage, they would need to only possess the ball for 2360.11 seconds, or 39.34 minutes in order to guarantee a win with 95% certainty.
Conclusion
All in all, we see that the idea of a home field advantage is very much alive and impactful in the simulations. Compared to the values given by the NFL and various sports networks, the home field advantage in the DSFL simulations is actually higher than that of real life, which is interesting. In the future, this absolutely should be looked into deeper, and the impact of the home field advantage should absolutely be tweaked in order to more closely resemble that of real life. Ideally, this analysis could have been done with much more data, likely all of the games from the beginning of the football league, in order to have the most legitimate and valid regression possible.
Code:
1202 words and a decent amount of research1. “Home Field Advantage: The Facts and the Fiction.” Chicago Booth Review. Accessed April 30, 2020. https://review.chicagobooth.edu/magazine/sp...and-the-fiction.
2. White, R.J. “NFL Betting Tips: How Much Home-Field Advantage Is Worth for Every NFL Team in 2019.” CBSSports.com, August 20, 2019. https://www.cbssports.com/nfl/news/nfl-bett...l-team-in-2019/.