One of the interesting things about Football is just how big of an effect Turnovers have on the game, especially at the professional level. We've all seen the stats: "If a team has a +3 turnover margin in a game they win the game 90+% of the time", etc. But just how big of an effect do Turnovers have on a season, how many games is a large deviation from the mean (+ or -) in the turnover department worth?

I say turnovers are a variable because they really can't be controlled. It is true that good defenses force them and good offenses don't give them up but that margin doesn't really count for a whole lot when you consider the majority of turnovers happen to be "right place right time". Think about it, how many interceptions are the result of tipped balls just happening to land in the right hands? And how many times does a rolling fumble happen to roll a defensive players way when it could have easily bounced the other way?

The mean # for turnover margin for a team in a season is 0. Anything other than that is a deviation and IMO anything + or -10 is a significant deviation that I would classify as an anomaly. If a team has for instance a -15 turnover ratio in one season that probability is extremely high that in the following season their turnover margin will be significantly greater as it "regresses to the mean" of 0. Regression to the mean is a key element of statistical analysis that is often overlooked in sports betting but many astute bettors know this well, and it applies in football with Turnovers.

Keep in mind this is a relatively small sample size (6 seasons) but take a look at some turnover margin anomalies from the last 6 years and how each team fared win/loss wise the following season:

I classified an anamoly as a turnover margin in a full season of +10 or greater and -10 or greater.

Teams with a +10 or greater turnover margin could be classified as somewhat "lucky" as they have gotten significantly more turnovers in their favor than what is expected. So a regression the following season should be expected right?

For the last 6 seasons teams that have a +10 or greater turnover margin have seen their win totals in the following season after the +10 or greater year drop by an average of -2.94 wins, that's 3 wins! Mind you the sample size is not the greatest at 36 teams over the 6 years but IMO is still statistically significant. Also significant is that 81% of those 36 teams saw a drop in their win totals, 4 stayed the same and just 3 (8%) increased all contributing to the average 3 win drop off. 3 wins less is a lot in a league with just 16 games.


Although the numbers aren't as large for teams that "should" be improved based on regression to the mean but they too are significant. Of the 35 teams the last 6 years that have had a turnover ratio of -10 or more for a full season they combined to average +2.2 more wins in the following season. With about 70% of the teams winning more games.


What does it all mean? Probably not a hell of a lot betting wise but it does show how some years teams can be "lucky' or "unlucky" based on how the ball bounces and are more likely than not to regress (either + or -) the following year.

In the 2008 season the following is a list of teams that had turnover ratios of +10 or greater and -10 or greater (in other words lucky and unlucky teams):

Miami +17
Tennessee +14
Baltimore +13


Houston -10
Dallas -11
Denver -17
San Francisco -17


Not surprisingly Miami, Baltimore and Tennessee all had great years last season with their high turnover ratios. However what is surprising, to me at least, is that despite their bad turnover ratios, Houston, Dallas, Denver and San Francisco really weren't all that brutal in the win/loss column with records of 8-8, 9-7, 8-8, and 7-9 respectively. If the +2.2 ratio from above is factored in these could have been playoff teams last season had they had a few bounces go the other way and near normal turnover margins or "better luck".


Unfortunately the Win Total futures for this coming season for Baltimore, Miami, and Tennessee seem to already have their predicted "regression to the mean" in turnovers already factored into the lines as Miami and Tennessee's totals are a full 4 wins less than what they finished last year with and Baltimore's is 2.5 less. I've already seen people asking why the totals are so low on these teams, and to that I saw, BE VERY CAREFUL betting Miami, Baltimore or Tennessee OVER season wins for 2009 as recent stats suggest .

However, they don't seem to be factored in with the teams that should be better next year. Houston won 8 games last year, their win total is set at 8.5 with + juice. Dallas won 9, win total at 9. Denver won 8, win total at 7, and San Fran won 7, win total at 7. There could be some hidden value on the OVER Season Wins with these 4 teams.