Exclusive Content   Join Now

Turning over new leaf? Not after 20 years!

Sep 2, 2003 9:28 AM

Two of the most successful "contrary" measures for predicting the outcome against the spread we have devised are Turnover Difference, and Cumulative Spread Margin.

These tools have both been solid for over 20 years. Lately, the results have been better than the lifetime average, especially the past two years.

Here’s how it works. If team A has been averaging a —2 turnovers net per game (i.e. committing two more TOs than its opponent), while team B is +1, team A would be the pick.

For the CSM approach, look for a team’s average net per game ATS of six points or higher.

If a team has been beating the line by an average of 4 per game, while the opponent has been losing to the line by 3, the latter team would be the one favored by the ”˜jumpers.

In both cases the counter play works best in Weeks 5-12, the heart of the NFL season. For reasons not entirely clear, these handicapping tools are less successful after Week 13.

The turnover difference is 57 percent over the last 20 and the CSM is 56. The past two years have hit 60. That’s some strong picking considering you are dealing with quite a few picks in a given week.

The temptation is always strong for a higher percentage, so we have run some tests to see what happens when considering both turnover and spread margin together.

One of our graphs lists a 10-year span of results (1991 to 2000) when both the turnover difference and the CSM have an active play on the same team in a game:

Comments: The overall performance improves to 59 percent when both factors work at the same time. The record was 41-24 (63%) over the past two seasons. From Week 13 on the numbers fall to 52% and should be considered non-predictable. Note that games with a spread of 5 to 9½ were impressive posting 63 percent in the initial 10-year sample and 60 in the two-year retest. Likewise "double contrarian" favorites during those same 10 and 2 year test periods were at 63 and 60 percent respectively.

Here are some other results: INT picks + CSM agree 56%; FUM picks + CSM agree 61%; All three T/O models agree 56%; All three T/O + CSM agree 57%.

Historically, the really high CSM game differences have not been as good. Limiting the range to a 6 to 13 difference between the teams, will invariably improve the results of the overall samples.

Our conclusion: Using Turnover Difference and the CSM together can add some extra punch to an already solid approach. features innovative statistical coverage of the NFL to help you win. Visit us on the web at or contact us at [email protected]