NFL Home Field Advantage
Many people with an interest in forecasting future outcomes of NFL games have taken a pass at creating a home field advantage number that goes beyond a standard, default amount of points regardless of the teams or time of year. Is there anything to be gained from more sophisticated HFA calculations?
NFL Home Field Advantage Studies, part I
Statistics can prove that there is some amount of advantage for the home side in a given matchup. Whether this is because they don't travel, they know the stadium particulars, the fans make a difference, or any number of other theories is unclear. What we do know is that when you take a reasonably large sample of games, the home team will wind up scoring on average between two to three points more per game than the away team. Since all teams play the same number of home and away games during the regular season, this is significant evidence of a real home edge.
The vast majority of people can agree with the concept of the home team having some benefit to playing at home. Where to go next from this simple assessment becomes more controversial. Some of those in an interest with predicting upcoming NFL games like to stick with a constant number to represent the home field advantage, regardless of the stage of the season, the opponent, and any recent events for the home side. At the other extreme there are folks advocating that you craft team specific HFA's and adjust them on the fly. We'll take a more disciplined approach to exploring the subject and first examine whether there is any change to the general league-wide home field advantage based on what week of the season a game is being played.
Of course, what data sample to use is a first choice, and for starters we will go with a simple 'last four season' model based on the 1999 to 2002 regular season games. We'll present the true average by week, and a 'smoothed' average which uses the week itself plus the two weeks on either side (or one week on either side for weeks 2 and 16). In addition we tallied up the home versus the point spread nets as well for another take on the matter.
Home Field Advantage by week number, 1999 to 2002
||# of games
||Average HFA per game
||Adjusted HFA per game
||Home vs Spread Net
||Average 'HSA' per game
||Adjusted 'HSA' per game
Analysis: now, four years is not a huge number of games in NFL-land, yet there are some pretty evident patterns emerging over this span. Essentially the home field advantage seems to be over-stated when people claim it's worth a generic 2.5 or 3 points, since from week one to week twelve the average is 1.3 points, and then from week 13 on it averaged a whopping 5.3 points!
The 'smoothed' averages provide a cleaner look, but even there it is clear that some jump happened at week 13. Likewise using the 'home versus spread' sums instead of the 'home minus away' score average, shows that the smoothed effect has the away sides on top from week one to twelve, and then the home sides coming home strong down the stretch.
Another interesting area is the first two weeks of the season, when maybe the 'jitters' and high expectations of a sellout crowd can make it a tougher than normal game for a home team.
Is this simply a mirage from not using enough years? That's a fair question to ask, and we can certainly revisit this subject after we run the numbers for another decade or so beyond our original sample, but we have an inkling that these stats may hold up under further testing.
Even if the data wasn't in agreement from an older sample, a case can be made that the NFL changes from year to year, sometimes dramatically, which means treading into the waters of ancient history may not be too pertinent to what is happening now. However, we'll crank out the data, and let the numbers stand or fall on their own merits.
At a first glance though, the wise men may elect to use a variable Home Field Advantage adjustment to their predicted scores, based on the time of year that a game takes place.
Home Field Advantage by week, 1983-2002 (Subscriber Feature)