On the efficiency of racetrack betting market: a new test for the favourite-longshot bias

ABSTRACT

A number of empirical studies on the efficiency of racetrack betting market have shown the ‘favourite-longshot bias,’ which means longshots are overbet while favourites are underbet. Asian markets such as Hong Kong and Japan, however, have produced some contradictory empirical evidence to the bias. One critical element in the efficiency test procedure is how to assess the unobservable objective winning probability of a horse in a race. This paper proposes a new test framework with a more general evaluation of the objective probability of winning than the traditional method. Unlike the traditional method, our model allows the heterogeneity of the horses and the races. We apply the new empirical method to test whether the favourite-longshot bias is present in racetrack betting market of Korea. We found that the favourite-longshot bias exists in the racetrack market of Korea and the result distinguishes Korean racetrack market from other Asian markets.

KEYWORDS:

• Market efficiency test
• favourite-longshot bias
• racetrack betting
• probit regression

JEL CLASSIFICATION:

• D81
• G14
• C25

Acknowledgments

We are grateful for the helpful comments from the anonymous referee and the editor. This study was partially funded by the National Research Foundation of Korea (Grant Number NRF-2016S1A3A2923769). The authors declare that they have no conflict of interest.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The studies of market efficiency in sports betting markets include basketball (Camerer 1989; Brown and Sauer 1993; Paul and Weinbach 2005), baseball (Woodland and Woodland 1994), football (Pope and Peel 1989; Golec and Tamarkin 1991; Forrest and Simmons 2000; Cain, Law, and Peel 2000, 2003; Forrest, Goddard, and Simmons 2005; Deschamps and Gergaud 2012; Graham and Stott 2008; Vlastakis, Dotsis, and Markellos 2009; Borghesi 2012; Koning 2012; Direr 2013; Nyberg 2014), ice hockey (Gandar et al. 1988; Dare and MacDonald 1996; Gray and Gray 1997; Woodland and Woodland 2001; Gandar, Zuber, and Johnson 2004; Paul and Weinbach 2012) and tennis (Cain, Law, and Peel 2003; Lahvička 2014; Abinzano, Muga, and Santamaria 2017) among others.

² FLB has been found in the US (Ali 1977; Gramm and Owens 2005), Australia (Tuckwell 1983), Canada (Hausch, Ziemba, and Rubinstein 1981), Germany (Winter and Kukuk 2006), New Zealand (Gandar, Zuber, and Johnson 2001), and the UK (Williams and Paton 1997; Bruce and Johnson 2000).

³ There are studies on reverse FLB in other sports betting markets outside of horse tracks, including Swidler and Shaw (1995) and Schnytzer and Weinberg (2008).

⁴ Those two manuscripts are Kim (2008) and Ro (2014).

⁵ As a referee points out, it would be ideal if we could compare Korean market results to Hong Kong and Japan using the same methodology of ours. Unfortunately, however, we do not have access to race-level data from those countries.

⁶ Even though the market efficiency hypothesis allows market prices to be imperfect in the short term, there exists the literature showing the true values will win out in the long term. (Malkiel 2003; Gray, Gray, and Roche 2005) .

⁷ Hurley and McDonough (1995), Ottaviani and Sørensen (2010) and Sung, Johnson, and Peirson (2012) explains two different types of bettors, one is well informed and the other is uninformed, can exaggerate FLB.

⁸ On a Jeju island, racing is reserved for Korean native-pony descended from the Mongolian horse.

⁹ Sung, Johnson, and Peirson (2012) have tried to explain FLB using different bettor types between weekends and weekdays. In this study, this explanation has not been considered because racetrack betting is made on weekends only.

¹⁰ Deduction rates are between 20% and up to 27% depending on the bet type. If it is 27%, it consists of 16% deduction is for three taxes (leisure tax, 10%; local education tax, 4%; and rural special tax, 2%) and 11% is for the KRA (the operation cost: 7% and the profit: 4%).

¹¹ Refer to 2016 Annual Report IFHA. 1 Euro (€) = US$1.052 as of 12/31/2016.

¹² Trifecta was introduced in April 2009.

¹³ We are grateful to KRA (Korea Racing Association) for providing the data.

¹⁴ One of the advantages of our data is the fact that id and name of each horse exist in our data set. Given the same ids and names, we could estimate the objective probability of the same horse participating multiple times. (Objective probability means the winning probability of the sample horse running n times).:

¹⁵ Of course, the nine horses in the first group are also included in the second group.

¹⁶ There exist 16 ‘favourites’ in our data, as the maximum number of horses in a race was 16.

¹⁷ To conserve space, we suppress the results by each horse. The detailed table is available from the authors upon request.

Additional information

Funding

This work was supported by the National Research Foundation of Korea [NRF-2016S1A3A2923769].