The role of small bettors in price formation in betting exchanges

ABSTRACT

The presence of small bettors in betting exchanges generates mispricing, which can lead to exploitation by informed traders or result in permanent price deviations. This paper shows that mispricing from this source is also dependent upon variables of established relevance such as tournament round, the level of attention to the event, the volume of the betting, and bet type, further confirming these findings by means of cumulative accuracy profile (CAP) curves. It also offers evidence of the relevant role played by the type of device used to place the bet, whereby higher mispricing is observed in live bets placed via a mobile device, which appears to be associated with impulsive betting. This last finding could have practical implications for the regulation of the use of mobile devices to access gambling platforms.

KEYWORDS:

• Betting Exchanges
• market Efficiency
• noise Bettors
• Behavioural Finance

JEL CLASSIFICATION:

• L83
• G00
• G12
• G14

Acknowledgments

We thank carcaj.com, Raúl Ibañez and Ruben Mangas for their help and support with the data. Financial support from the Spanish Ministry of Economy and Competitiveness, Project ECO2016-77631-R and Fundación Caja Navarra FUNCAN-07445 are gratefully acknowledged.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The data for this study were supplied to Carcaj.es by their technology provider, Xanadu Consultancy (Matchbook.com) and made available to the authors for research purposes. Unfortunately, www.carcaj.es ceased its activity on 1 June 2017.

² To facilitate comparison, the stakes on all lay bets were converted to the equivalent for back.

³ Various studies have found considerably higher average bet size in Grand Slam tennis matches. For example, from a betting database covering Wimbledon Tennis Tournaments from 2008 to 2013, Brown and Yang (2018) report an average bet size of £860 for back orders and £929 for lay orders.

⁴ When analogous CAP and AR analyses were performed for the rest of the variables shown in Table 3, higher AR values were found in matches not involving any of the TOP5, in matches not involving any Spanish players, in bets on favourites, and market order bets (results available upon request).

⁵ The joint consideration of the three variables gives rise to multicollinearity problems, as indicated by the condition index values. The presence of multicollinearity issues might suggest that they are capturing the same information, although the QF variable would be more complex in content since it would also be capturing effects due to the increasing closeness in skill levels of the players as the tournament progresses.

⁶ It is worth recalling that the stakes were converted from lay stakes to back stakes.

Additional information

Funding

This work was supported by the Fundación Caja Navarra [FUNCAN-07445]; Ministerio de Economía y Competitividad [ECO2016-77631-R].