• Tampa Bay’s awful turnover margin led to an inordinate amount of close losses
  • The New York Giants perceived poor close game record in 2018 was deceiving
  • Dallas Cowboys’ close game record may not be due for regression as once thought

As you recall in my first article back in early July, I wrote about some key team stats that tend to regress from year to year. Stats like close game records, turnover margin, etc are being talked about everywhere but I think it’s important to look under the hood a little bit more. 

Sharp Football Betting Package
Take 15% off NFL Betting + Props

Turnover Margin in Close Games

You often hear NFL analysts on tv declare keys to a game that aren’t always predictive, but one stat they correctly identify is turnover margin. In the last 10 years, NFL teams have a 134-959-1 (.123) record when they lose the turnover battle by two or more. When the margin is three or more that record plummets to 22-459 (.045) with an average margin of defeat of 16 points. If you are capable of somehow predicting who will turn the ball over much more often than their opponent, you will absolutely be able to predict who will win that game. The problem is that it’s virtually impossible to do this, mainly because turnovers are often random. But the good news is because of this randomness, the turnover margin figure should regress the following season. 

Because losing the turnover battle by such a wide margin is so important in determining the outcome of games, let’s look at the teams that lost the turnover battle by an abnormal margin but only lost coin flip type games. By using the pro football reference season finder, we can filter games for teams that had a turnover margin of minus-3 or worse and still only lost by seven points or less. In the last 20 years, there have been 152 of these instances but only five teams suffered this distinction at least three times in one season. The 2002 Minnesota Vikings, 2005 Green Bay Packers, 2010 Dallas Cowboys, and 2012 Detroit Lions all lost at least three games by seven points or less in a single season with a minus-3 turnover margin or worse. 

Add to that list last year’s Tampa Bay Buccaneers. Of the four teams listed above, not including last year’s Bucs, they all increased their win totals the following season by at least two games. Every team finished 6-10 or worse in the season with this distinction and each finished the following season at 7-9 or better. 

Will the 2019 Buccaneers see the same fate? It’s hard to know for sure but in looking at last year’s results, they lost three games by exactly three points in each of the games when they turned the ball over four times and never took the ball away from their opponent more than once. As noted in the table below, not only did Tampa lose the turnover battle by a wide margin but they also outgained each of three opponents by a significant margin.

In a couple of the cases, against Cincinnati and New York, Tampa needed a second-half flurry to make those games close. In both games, however, Tampa turned the ball over multiple times in the first half or first 2.5 quarters. For example, Tampa only trailed by seven at halftime against the Giants, but then proceeded to throw interceptions on three straight drives. Tampa proved all season long their offense could score with any opponent, but the turnovers completely derailed many of their games. If Tampa can clean up the turnovers, they can certainly win more close games.

Close Game Analysis

One of the most often cited stats to determine potential surprise or bust teams is record in one-score games, seven points or less. The theory is that in the NFL, 1-2 random plays could make the difference in games that are determined by one score. Essentially these are coin flip games and during a season, teams should win near 50% of these games. However, there are times that teams score last-second points that could make a double-digit outcome turn into a perceived close game and vice versa. Because these occur multiple times per year, I decided to look to see if any teams were affected more often than others. The biggest observation is that the Giants’ poor close game record of 4-8 was a bit bogus. In reality, we should look at the Giants one score game record more like 4-5 than 4-8 that you will hear in the mainstream media. 

  • In the Giants second game last year at Dallas, the Giants were down 20-3 with 5:45 left in the game. Following an Evan Engram touchdown with 1:27, the Giants recovered an onside kick, drove down the field and kicked a field goal with 0:11 left in the game to make the final deficit a one-score game at 20-13. New York was outgained 5.9 to 4.1 yards per play. Needless to say, this was far from a coin flip outcome.
  • In Week 8 at home against the Redskins, the Giants trailed the entire game and Evan Engram scored a touchdown with 0:17 left to make the final deficit 20-13. The Skins outgained the Giants 5.5 to 4.9 on a yards per play basis and were never truly threatened.
  • In Week 7 in Atlanta, the Giants trailed the entire game and were down by 14 with 7:47 and by 11 with 1:55. Odell Beckham scored a touchdown and the Giants followed with a 2-point conversion with 0:05 left to make the game “close.” The final score of 23-20 was not indicative of the flow of the game and should not be included as a coin flip type loss.

Another team that was involved with multiple games that were not truly one score outcomes was the Cowboys. As noted above, Dallas played one game against the Giants that shouldn’t be considered close and another one at home against the Saints in Week 13. Dallas led 13-0 at halftime and was up 13-3 before New Orleans scored a touchdown with 1:08 left in the game to turn a one-sided game into a one-score outcome. Dallas outgained New Orleans 5.2 to 3.7 on a yards per play basis.

When you exclude these two games, Dallas’s one score record goes from 8-2 to 6-2. The Cowboys go from winning the most one-score games in the NFL to being tied with six other teams. Maybe their regression in close games in 2019 won’t be as dramatic as some might have concluded.