Preseason bias in the NFL and NBA betting markets

ABSTRACT

This study extends research in the sports gaming literature by examining the efficiency of betting markets related to preseason professional sporting events. Using NFL (1995–2014) and NBA (2005–2014) data from preseason games, we examine the pricing efficiency of point spreads in these markets and consider evidence of systematic mispricing. Findings suggest point spreads are too large in these situations, providing a profitable betting opportunity for those willing to systematically wager on underdogs. Similar findings are not seen within the context of NFL or NBA regular seasons. These findings are more pronounced as preseason point spreads become larger. Further stratification by week of the NFL preseason demonstrates that underdogs discontinue their superior performance for the one week (Week 3) in which clubs tend to expel a higher level of effort.

KEYWORDS:

• Sports betting
• market efficiency
• uncertainty

JEL CLASSIFICATION:

• G14
• G10
• D82
• D80

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Given the lower scoring nature of baseball and hockey, we concentrate on only those US-based professional sports leagues offering regular point spreads.

² https://www.washingtonpost.com/news/sports/wp/2014/08/03/nfls-preseason-experiment-with-longer-extra-point-begins-sunday-night/.

³ http://www.repole.com/sun4cast/data.html.

⁴ http://www.covers.com/sports/NBA/matchups?selectedDate=2012-10-6.

⁵ This results in somewhat conservative results as betting lines, in our full samples, more frequently move towards underdogs (so that they ‘get’ fewer points) at the closing lines we study, as compared to opening lines.

⁶ See, e.g. the calculation of Levitt (2004), footnote 14.

⁷ ‘Pick ‘em’ games, where teams are equally likely to win (i.e. the point spread is 0) are omitted from analyses considering the performance of betting on underdogs.

⁸ These breakpoints are chosen so that, collectively for both sports, roughly a half and a quarter of observations from the NFL and NBA are on either side of the respective breakpoints. Other breakpoint selections result in very similar results.

⁹ We classify early, fifth preseason games, such as the annual NFL ‘Hall of Fame Game’ as ‘Preseason Week 1’ contests, and the subsequent games played by these 2–4 participants, per year, in the sample are still classified as ‘Preseason Week 1’ games as well because they are held at the same times that all remaining NFL teams are playing their typical, initial preseason games.

¹⁰ E.g. ‘With two weeks of preseason action in the books, we’ve reached the most important game of the exhibition season: Week 3. It’s the dress rehearsal for most teams, where the starters will play anywhere up to three quarters before basically being shelved until the regular season.’ (http://www.nfl.com/fantasyfootball/story/0ap3000000517189/article/fantasy-storylines-for-week-3-of-the-nfl-preseason).

¹¹ We forego providing significance levels of the individual results given the small sample sizes which result from our stratification into subsamples.

¹² E.g. ‘The third week of the NFL preseason is traditionally the most important. Starters normally play until halftime and sometimes into the third quarter as a dress rehearsal for the regular season before receiving little to no playing time in the exhibition finale.’ (http://bleacherreport.com/articles/2554780-nfl-preseason-schedule-2015-week-3-tv-coverage-ticket-info-and-key-storylines).