NFL games have the tightest lines of any pro sport in North America, and as a result it is in theory the toughest sport to beat. Still, we tested a method last season that combined offensive yards per point and defensive yards allowed to forecast spreads, and when the difference from the real spreads was three points or more, the results were better than 55 percent!
The 2011 NFL Season is scheduled to begin in two months with training camps scheduled to open this month, although the latter looks to be in jeopardy due to the current lockout. This article assumes that there will be an NFL season this year, and even if the season is shortened, the principles we are about to discuss are still applicable.
Now it is no secret that NFL lines are the tightest of any professional sport, as bookmakers have plenty of time to get the lines right with teams playing only one game per week, and these lines have to be sharp with the NFL having easily the largest betting volume of any North American sport. It also has the largest betting limits, which is a strong indication that the books are confident in their numbers.
For this reason, bettors that think they can beat the NFL by analyzing basic widely available statistics are doomed to failure, as the sportsbooks have already analyzed all of those basic stats when making the line, and they are way ahead of the gamblers in that regard. Bettors have to look deeper at more advanced stats just to turn a profit, and maintaining a profit over many years is tougher in the NFL than in any other pro sport.
To that end, we feel we have come across one statistic that seems more predictive than the other generic stats, and understanding how to use this stat properly is the key to bettors having a chance to eek out a profit. All records against the spread discussed in this article are based on past closing betting odds from Pinnacle Sports.
Yards per point (YPP)
Yards per point, usually referred to as YPP, just may be the single most predictive statistic in football, but it is often misused in its relationship to the actual point spread. YPP is a simple efficiency calculation where you just divide total yards by points scored, and the lower the result, the better the number.
The New England Patriots were the most efficient NFL team during the 2010 regular season, as they totaled 5820 yards and 518 points for an excellent YPP of 11.24. The Carolina Panthers were the least efficient, with 4135 yards and 196 points for a horrendous YPP of 21.10. The average YPP for all NFL teams in 2010 was 15.25.
Now, simply taking the team with the better YPP has historically worked well in the NFL Playoffs in the past. The problem with that system is that it does not take the spread into account, so the books can counteract that trend by simply shading the lines on the better YPP playoff teams.
Our approach uses offensive YPP and defensive yards allowed to create a theoretical point spread, which we would then compare to the spreads on upcoming games. Now back-testing our approach takes time, because we cannot simply look at year-end statistics from past seasons and apply them to that year. Rather, we need to know the season-to-date numbers heading into each week, which is a much more complicated test.
With that said, we did test the method for the 2010 season, and the results are encouraging to say the least. But before revealing those ATS results, let us explain our method, which is actually quite simple but potentially very effective based on last year.
What you need to gather each week is the offensive YPP and the defensive yards allowed for each team. To get a team’s projected score, you divide its opponent’s defensive yards allowed per game by the teams own YPP. For example, if Team A has a YPP of 15.00 and Team B is allowing 300.0 yards per game, Team A’s predicted score would be 20 (300.0/15.00). After doing the same thing for Team B, you compare the predicted scores to get a point spread.
With our method, since you are using overall numbers, your prediction is now done. We do not make any adjustments for home field advantage. If you so desire, you can experiment using home numbers only for the home team and road numbers only for the road team, but that would account for fewer games and would in our minds be less reliable.
Next, you compare the theoretical line you just created with the actual betting line, and if the variance is at least three points, you have a play.
For a real life example, let us look at last year’s wild card playoff game between the New York Jets and Indianapolis Colts. We chose a game from that week for convenience, as it would use numbers from the entire 2010 regular season, which are easier to find than partial season-to-date numbers. It is also easier to use that week than future playoff games, where you would need to combine the regular season with the previous playoff weeks.
So the Jets entered that game with a YPP for the season of 15.3 and allowing 291.5 yards per game. The Colts entered the game with a better YPP of 14.00 (remember, the lower the better) but allowing a significantly worse 341.6 yards per game. To calculate the Jets score, you divide the Colts yards allowed of 341.6 by the Jets YPP of 15.30 to get 22.3. To calculate the Colts score, divide the Jets yards allowed of 191.5 by the Colts YPP of 14.00 to get 20.8.
So that gives a final prediction of Jets 22.3, Colts 20.8 for a theoretical spread of Jets -1½ .Note that when calculating the projected spread, any fraction would be expressed as ½, so the theoretical line would be 1½ is the difference is anywhere between 1.1 and 1.9 Then, since we need a difference of at least three points in order to make a play, we would need a spread of at last Jets +1.5 (or conversely, Colts +4.5).
Well, the closing line at Pinnacle was Colts -2, so the Jets were indeed the play. They proceeded to win the game 17-16.
Using this methodology for the whole year starting with Week 2 and using only the year-to-date numbers entering each week, our final results were an outstanding 109-86-4, 55.9 percent against the spread! This includes a 104-84-4, 55.3 percent regular season and a nice 5-2 ATS run in the playoffs. In case you were wondering, the method had no play on the Super Bowl, as it spit out a forecast of Green Bay -1, which was only two points less than the actual line.
Now one season with about 200 plays may not be statistically significant, but it was certainly a nice beginning and worth keeping an eye on in the future.