In a past article, I explained the method I use to create a model to produce NFL Power Rankings or figures for each team in the league. I discussed how I prefer starting with numbers to identify possible plays and then further examine the matchups to finalize my weekly choices. Well, when I began developing my quantitative NFL model a few years back I soon realized that I also wanted some type of logical, numbers based method to determine a team’s home field advantage as well. Hence, another labor of love began in which I attempted to uncover each NFL franchise’s home field advantage. The method I chose was to examine two aspects of how a team performs at home compared to on the road and combine those findings into a number which could then be added to the home team. I will explain below.
First I examined each team’s winning percentage at home compared to the road over a ten year period (a 160 game sample) excluding playoff games and neutral site games. I choose to look at this data simply because teams who have a significantly greater win percentage at home versus the road accrued this because they are scoring more points than their opponent within their home stadium and allowing less. Unsurprisingly, 31 of the 32 NFL teams had a better winning percentage at home then the road over this ten year period. In fact New England, Buffalo, Minnesota, Green Bay, San Francisco, and Seattle all have won 20% or more games over the past 10 years within their home stadium than on the road. Dallas is the sole team to win less games at home than on the road over that sample size and land at the bottom of this list along with Tampa Bay, LA Rams (yes some of that was in St. Louis), and Oakland. Surprising Kansas City, who is commonly thought to have a great home field advantage only has won 10% more games at home than on the road during that time span which is below league average.
The second factor which I examined is the average differential between points scored and points allowed by a team when playing at home versus the road. Again this was examined over a ten year period from 2009 to 2019. Understandably, equal weight cannot be given evenly to all those years as players, coaches, and even stadiums change over that period of time. To account for this I used a weighted scale which gives greater credit to more recent seasons in determination of a home field advantage grade. Therefore the further back in time we examine, the less weight that point differential has. Additionally, to further emphasize this recency effect, I calculated point differential in 10, 6, and 3 year weighted chucks. I then took these weighted averages and averaged these together to create a final number. To summarize, this home/road point differential number reflects the average of weighted 10, 6, and 3 year groups. When examined in this manner, the Green Bay Packers came out on top followed by Miami, Houston, Buffalo, and Arizona. The bottom group was very NFC East as it contained Dallas, the New York Giants, Washington, and the LA Rams.
Upon the suggestion of a few fellow handicappers, I further teased out the home vs road point differential by looking at divisional vs non-divisional opponents. What was found was that for a majority of NFL teams, 17 of 32, there was a significant decrease in the number of points scored at home vs road when playing teams from within the division vs non divisional opponents. Eight teams were neutral to divisional vs non divisional competition while 7 teams performed better in this metric when playing divisional opponents. For this reason, each team has 2 home field advantage numbers depending on whether they are playing a team from within the division or outside of the division.
To combine the two factors considered in this exercise (home vs. road winning percentage and home vs. road average point differential) I took the metrics and averaged them separately. What I found is that on average NFL teams win about 14% more of their games at home then on the road while the average home vs road point differential is about 4.6 points. Then, I used the assumption that the average HFA used by linemakers is approximately 2.5 and worked backwards examining each team individually by comparing their specific home win % and point differential to the league averages in these metrics. The closer these data points fell to the league average, the closer their 2019 HFA was placed to 2.5. So for example, if a team had a league average home win percentage and a league average home point differential then their HFA would be 2.5. If they were slightly above this they would likely land around the 3 area for HFA while a few teams (such as Green Bay and Minnesota) were significantly above average in both categories and fell around a 4 point HFA. These metrics were then modified slightly based on the numbers I had used in previous years and a home field advantage number for divisional and non divisional opponents for 2019 was devised. It should be noted that this number is not necessarily “static” throughout the season. Situations such as back-to-back road games for an opponent or extra rest off a bye could push these numbers up or down depending on the specific scheduling situation.
To summarize, there is a very real home field advantage within the NFL however, creating a reliable number that reflects a team’s true advantage when playing in its home stadium is quite tricky. The numbers you’ll see in my articles represent one method of examining home field advantage. Other statistical examinations of the topic are certainly possible and would be interesting to study. Further analysis could focus on a home team’s relative advantage during different months of the year (for example: I would add +.5 points to Denver HFA for Weeks 1 and 2) or traditional warm weather teams playing cold weather teams. Data could also be adjusted to reflect new stadiums, injuries to key players, or blowout games which are statistical anomalies. In the end however, there is no perfect way to examine the phenomena but, having a logical, evidence based system to account for this elusive number is better than none at all.
Note: I am pretty open minded and always willing to consider other ideas or suggestions. If any reader with a strong math/statistical background would like to review my data and approach feel free to reach out to me on Twitter @sciflyguy.