NFL Betting Against the Public 2003-05

  • Article: NFL Outlook and Review of 2005’s “Year of the Favorite”

What can we expect from the 2006 NFL Season?  Was 2005’s “Year of the Favorite” a fluke?  How does “Betting Against the Public” fare during different types of “market environments?”  To get a better perspective of things, we thought we would take a step back and study some longer-term data.  The information on this site is for entertainment and educational purposes only.  Use of this information in violation of any federal, state, or local laws is prohibited.

2005: Year of the Favorite

As you may know, the 2005 NFL season was very different from other years.  We have seen academic studies that verify the bias towards underdogs in the NFL (as well as in other major sports).  Indeed, before 2005, various studies showed that NFL dogs beat the point spread about 52%-53% of the time.

Here is a table of NFL underdog performance over the past few years.

Table 1: NFL Underdogs vs. Point Spread

Year NFL Underdogs
2001 53.0%
2002 55.8%
2003 53.2%
2004 50.0%
2005 42.8%

As you can see, 2005 was a very unusual year.  Although NFL Dogs might normally beat the spread 52%-53% of the time, the average over the past 3 years is now 48.3%, and over the past 5 years, 50.8%.

Betting Against the Public in the NFL

The aberration of 2005’s dogs and favorites dragged down the results of many professionals, cut into sportsbook’s profits – and even put a dent in Betting Against the Public.  Even though 2005 was a difficult year for dogs, Betting Against the Public in the NFL still shows a 53.6% winning percentage (compared to 48.3% for all dogs), since inception three years ago.

The NFL does not generate the large number of games that other sports such as the NBA and MLB generate.  The limited number of games makes it more difficult to obtain “statistically significant” results.  As a result, we thought we would slice and dice the numbers in various ways.

If Betting Against the Public in the NFL can show consistent results – no matter how we study the data, it will improve our comfort level and the “robustness” and “effectiveness” of the results.  We analyzed the performance of blindly betting on all underdogs versus betting on the dogs that the public didn’t like – on a year-by-year basis.

Table 2: Benefit of Betting Against Public (vs. Spread)

  All Dogs Dogs, Bet Against Public
2003 53.2% 61.8%
2004 50.0% 50.0%
2005 42.8% 48.1%
2003-2005 48.3% 53.6%


Over the past 3 seasons, betting on dogs would have resulted in a 48.3% winning percentage.  If you Bet Against the Public (in the NFL, most of these plays would be on the dog, at the 75% level), you would have hit 53.6% over the past three years.  Results are fairly consistent each year, with Betting Against the Public adding about 5% value per year.

This shows that Betting Against the Public can be beneficial to your handicapping.  Indeed, when NFL dogs “revert to the mean” and hit at 52%+, we might expect Betting Against the Public to also bounce back – at least to the 55%+ range.

Conclusion & Rationale

There’s a saying, “A statistician is someone who can have his head in an oven and his feet on ice – but feels great, on average.”  It’s good to have the numbers and stats behind your handicapping.  It’s even better to have a solid rationale for why the system works.

Why does “Betting Against the Public” work?  Fading the Public puts you on the side of the sportsbooks – and sportsbooks are in business to earn a profit.  As a result, Betting Against the Public – and on NFL dogs – will have their day again in the near future.


We do not guarantee that the trends and biases we’ve found will continue to exist.  It is impossible to predict the future.  Any serious academic research in the field of “market efficiencies” recognizes that inefficiencies may disappear or fade over time.  Once inefficiencies are discovered, it is only a matter of time before the market corrects itself.  We do not guarantee our data is error-free.  However, we’ve tried our best to make sure every score and percentage is correct.