Arbitrage and market efficiency in sports betting markets

(1)

Arbitrage and Market Efficiency in Sports Betting Markets

Bachelor Thesis

Rob Clowting

10071881

Economie & Bedrijfskunde

Finance & Organization

Supervisor: P. Versijp

07-‐02-‐2014

(2)

1

Abstract

Section Page

Introduction 1 Literature and Background 2 Methodology 7 Results 9 Practical application of arbitrage betting 12 Conclusion and Discussion 15 Appendix 16 Reference List 16

(3)

Introduction

Arbitrage is defined as the simultaneous purchase and sale of the same security in two different markets for different prices with a risk free return. In modern financial

markets this mispricing is increasingly difficult to exploit for individual investors, but in the market for online sports betting recent literature suggests arbitrage may arise frequently and in an easily expolitable way.

The online sports betting market has grown enormously over the past decade due to the rise of the internet and mobile internet devices allowing bettors to place bets on any sports event, at any time, anywhere in the world. According to Bwin.Party, one of the leading global online betting agencies, the global online sports betting market, excluding the US, was estimated to be worth €10.5 billion in 2012 and expected to grow at 7.3% per year for the period 2012-‐2015 (Bwin.Party, 2013).

The market for online sports betting is divided into two seperate markets. The first and best known is the bookmakers market, where individual bettors bet on sports events where the odds have been determined by the bookmakers preferences and information. The other market, which is relatively younger and less well known, is the exchange market, where odds/prices are determined by supply and demand of

individual bettors. In this market, bettors can not only bet on a certain outcome "Team X

wins/draws/loses", but also against these outcomes, giving "Team X does not

win/draw/lose". This exchange market can be approached in the same way as a normal

stock market, with the opportunity to go long (bet on a certain outcome) and to go short (bet against a certain outcome).

In the difference in odds between these two markets that arise due to bookmaker preferences and other factors, the literature suggests that arbitrage

opportunities arise frequently. Franck, Verbeek and Nüensch (2012) researched the five big European football leauges for arbitrage opportunities, and suggested that for smaller leaugues different results may arise due to different levels of market efficiency.

To research this question, this paper will investigate the Dutch Eredivisie in the season 2012-‐2013 and research the arbitrage frequency and returns. These results will be compared to the results found by Franck, Verbeek and Nüensch to see if there are any significant differences.

(4)

Literature and Background

One of the most important concepts in modern day finance is the Law of One Price (Lamont and Thaler, 2003). This law states that an identical good must sell for identical prices in all markets. Theoretically, if this law is violated, meaning an identical asset is sold for different prices on different markets, an arbitrage opportunity arises. The concept of arbitrage is generally defined as the simultaneous purchase and sale of the same asset to profit from a price difference. Arbitrage is a crucial concept in modern day financial theory. Because arbitrageurs instantly exploit the arisen arbitrage

opportunities, prices theoretically never fluctuate far from equilibirum for extensive periods of time. The arbitrageurs make a small riskless profit and prices are quickly back at effiecient levels. This is the basis of modern day financial theory such as Fama's classical efficient market theory, the CAPM model and Ross's Arbitrage Pricing Theory (Schleifer and Vishny, 1997).

In their paper on anomalies in the Law of One Price, Lamont and Thaler (2003) suggest that this Law might be violated on a larger scale than economists expect. A famous example is the mispricing of Royal Dutch/Shell, where shares for the firm were

supposed to be traded at a 1.5 ratio in two different markets (London and Amsterdam), but this ratio varied from being 30 percent too low to 15 percent too high. This example, amongst others, is a blatant violation of the Law of One Price and together with bubbles on financial markets there seems to be evidence that other factors, still unknown to academics, may be influencing prices in financial markets (Lamont and Thaler, 2003).

Arbitrageurs exploit price inefficiencies in a market to make a riskless profit. By doing so they also rebalance the price to its equilibrium, because the exploitation of arbitrage opportunities leads to prices changing back to their efficient level. In modern financial markets, arbitrage opportunities often arise and disappear in a matter of nanoseconds and complicated software and trading systems are required to exploit these. In online betting markets, recent studies performed by amongst others Vlastakis et al. (2009) and Franck et al. (2012) show that arbitrage opportunities arise frequently and in an easily exploitable way.

(5)

Betting markets are interesting for academic research because they function similar to normal financial markets (Vlastakis et al., 2009). Academic literature has focused on market efficiency in betting markets because it allows for easy testing compared to financial markets. The advantage of conducting research on betting markets is that bets have a period of life that is certain beforehand and the resulting value of the bet is fully known afterwards. This paper will research the extent to which betting market

efficiency varies over differently sized leagues in European football betting.

Inefficiency and mispricing in betting markets has been researched quite extensively. Inefficiency in betting markets implies that the odds given by the bookmaker or market do not reflect the true probabilities of an outcome. Papers on this subject often research biases in bookmaker pricesetting, such as the favourite-‐longshot bias, as described by Vlastakis et al. (2009). The favourite-‐longshot bias states that betting on favourites (low returns with high probability) yields higher average returns than betting on longshots (high returns with low probability). This bias has been consistently proven in many different sports, ranging from american football in the United States and regular football (soccer) in Europe to horse racing.

Another bias found by several papers on the subject and described by Vlastakis et al. (2009) is the overestimation of the home team advantage. Because bettors tend to have a statistically unjustified bias towards the home team, bookmakers set their odds inefficiently to exploit this biased betting behaviour. The reasoning behind this will be explained in the next section. Vlastakis et al. (2009) find that the most profitable strategy is betting on 'away-‐favourites', where it should be remarked that this strategy still has a negative average return.

To answer the question why bookmakers set prices inefficiently the literature agrees on several causes. One reason, as given by Vlastakis et al. (2009), Kuypers (2000) and Franck et al. (2012) can be found in human psychology. Bookmakers set inefficient odds but maximize profits by exploiting bettor biases over a certain sports event. For

example, a bookmaker with a strong presence or customer base in England may expolit sentimental betting over an England national football team game against Germany. By setting inefficient odds the bookmaker may still maximize its profit by exploiting the

(6)

potentially irrational betting behaviour of its English customers. The loss they take on their inefficient odds is compensated by higher trading volume.

A second reason for inefficient pricesetting, also described by Vlastakis et al. (2009), is that bookmakers want to exploit the 'favourite-‐longshot' and 'home

advantage' biases. For example, if the team that is in first place in a league plays against the team that is in last place, a bookmaker may set odds for the team in last place to win higher than they should be, attracting more bettors that take a 'long shot' on this

outcome. Again, the higher betting volume compensates the mispricing, maximizing profit for the bookmaker.

Another simple but important reason, described by Franck et al. (2012), is that bookmakers purposefully set their odds inefficiently for advertisement or promotional reasons. By offering odds that give short term negative returns for the bookmaker during the promotion or advertisent period, the bookmaker hopes to attract new customers that will stick to their company and give postive net returns in the long run.

To understand how bookmakers can compensate mispricing with betting volume we look at the bookmaker's revenue model. The bookmaker earns its money by charging a commission that is integrated in its odds. This implies that a fraction of every bet a bettor makes goes directly to the bookmaker as a commission. Franck et al. (2012) find that bookmakers charge 11.3% commission on average. This gives rise to the

opportunity for bookmakers to set odds inefficiently to boost betting volume. Because they charge a commission on all bets, high betting volumes have the potential to compensate losses taken on mispricing.

This mispricing should potentially give rise to arbitrage opportunities. Vlastakis et al. (2009) researched arbitrage in the bookmaker market by itself, also called intra-‐market arbitrage. Arbitrage opportunities are found by spreading bets across multiple

bookmakers that have different odds for the same match. They found that arbitrage opportunities arise in only 0,5% of matches. To examine potential arbitrage between the bookmakers market and the exchange market first the betting exchange market will be examined.

Betting exchange markets are relatively new in the online sports betting world. The best known and largest betting exchange platform is Betfair, with almost 1 million active

(7)

users worldwide and an annual revenue of £387 million in 2013 (Betfair, 2014). In a betting exchange market, the bookmakers are completely replaced and prices are set by a constant auction process of supply and demand of odds, determined by individual bettors themselves. On the bookmaker market, the commissions bookmakers charge are already included in the odds, but betting exchange platforms charge commissions on the net profits of the bettor. Betfair charges a 2-‐5% commission based on betting activity and does so only on the bettor's won bets.

Smith et al. (2006, 2009) extensively researched the performance of these betting exchange markets compared to the bookmakers market. In their research, they found that betting exchange markets perform significantly better at predicting match outcomes than the bookmakers market, demonstrating that the exchange market is more efficient than the bookmakers market. In this situation where there is the bookmakers market that sets prices inefficiently and the exchange market that is superior at predicting outcomes, arbitrage opportunities should arise frequently (Franck et al., 2012).

The betting exchange platform offers bettors the possibility to take both sides of a betting contract, similar to taking long and short positions on a regular financial market. Not only can a bettor bet on certain outcomes of an event 'Team X wins/draws/loses', he can also bet against a certain outcome 'Team X does not win/draw/lose'. This offers the opportunity to 'buy' a bet on the bookmakers market and then 'sell' this bet on the exchange market. This strategy is called inter-‐market betting and because of the mispricing and setting of inefficient odds on the bookmakers market, as discussed above, the literature suggests that arbitrage opportunities may arise frequently. Franck et al. (2012) find that in the big five European football leagues (England, Spain,

Germany, Italy, France), arbitrage opportunities arise in 19,2% of all matches resulting in an average positive return of 1,4% on these arbitrage opportunities. They also suggest that in smaller leagues the inefficiency may be even bigger due to less available information and bookmakers difference in opinions on pricesetting. This paper will research the extent of this implication by taking the Dutch Eredivisie in the season 2012-‐2013 and compare the arbitrage frequency and return to the results found by Franck et al. (2012). To test this the hypothesis for this thesis will be:

(8)

H0: Arbitrage frequency and returns in Dutch Eredivisie = Arbitrage frequency and

returns in European markets

H1: Arbitrage frequency and returns in Dutch Eredivisie ≠Arbitrage frequency and

returns in European markets

An internet search reveals there is a big online market for these arbitrage bets, also called 'sure bets' or 'free bets' on many websites. Sites like 'Oddsportal.com' and

'Oddschecker.com' all offer advice on how to profit from arbitrage trading. There are also websites that offer complete software packages to engage in arbitrage trading in sports betting, such as 'Rebelbetting.com' . These websites offer information about arbitrage bets for a monthly subscription fee and claim to offer a 10-‐20% monthly profit from their information. Some media exposure in respected media such as the British

newspaper The Guardian confirms that these arbitrage subscriptions can be relatively simple and quite profitable for an individual bettor (The Guardian, 2012). But as Franck et al. (2012) also mention in their paper, bookmakers do not appreciate arbitrage trading on their websites and reserve the right to restrict betting or cancel any account without further explanation when they suspect arbitrage trading activities. This rather crude form of market regulation potentially limits the profitability of arbitrage trading for a bettor in the long run, as he is more likely to be caught by the bookmakers as his profits and trading volumes increase (Franck et al. 2012).

To answer the question why these arbitrage opportunities still arise given that there even is software available, Franck et al. (2012) argue that bookmakers look at the long run and not at their individual bets. They set their odds in such a way to attract new customers that will bring in money in the long run because they face transaction costs to switch to a different bookmaker. Bookmakers have information about all their

customers and by examining their trading history they can simply cancel the accounts of suspected arbitrageurs and keep the profitable regular bettors.

Another argument is given by Montone (2012), who states that arbitrage trading is Pareto-‐efficient. As the start of a match comes closer, odd volatility increases and arbitrageurs help the bookmakers identify the arbitrage opportunities so that they can rebalance their books to remove these arbitrage opportunities from their odds.

(9)

Methodology

Data collection for the bookmakers market is performed thorugh the website football-‐

data.co.uk. This website offers historical data for match odds from 10 major bookmakers

(B365, Bluesquare, Bwin, Gamebookers, Interwetten, Ladbrokes, Sportingbet, Stan James, VC Bet, William Hill). Odds from football-‐data.co.uk are collected every friday afternoon for weekend matches and tuesday afternoon for weekday matches (although weekday matches are not releveant for the Eredivisie). For the exchange market, we will use the data from the exchange platform Betfair.com. As described above, this is the biggest and best known exchange platform and historical data is available thorugh their website data.betfair.com.

The Betfair data has to be cleaned first to get relevant odds for every match. For every match, first the odds have to be filtered so that only the odds are taken that were traded at the time odds were recorded at football-‐data.co.uk. Because the Betfair data allows for filtering on the criteria "First-‐traded" and "Last traded", we can filter the odds so that they match the odds in time for the football-‐data.co.uk data. All Betfair odds odds that were traded for the first time after football-‐data.co.uk odds were recorded (friday

afternoon) and for the last time before football-‐data.co.uk odds were recorded have been dropped. Following the research performed by Franck et al. (2012), the odds selected are the odds with the highest trading volume 'VOLUME_MATCHED' in the data. After the relevant Betfair odds have been selected, they are matched with the data from football-‐

data.co.uk, so that a list is created with all Eredivisie matches in the season 2012-‐2013

and corresponding bookmakers' odds and Betfair odds. This gives a database of 306 matches, where one match had to be removed because there had been an error in the Betfair database that made the difference between in-‐game traded odds and before-‐ game odds impossible to seperate. In total, the database consists of 305 Eredivisie matches with corresponding odds.

For arbitrage, three strategies can be used. The first one that will be examined is the 'long-‐position intra-‐market arbitrage' strategy, which simply means the bettor 'buys' his bets on the bookmakers market and then 'buys' the opposing bets from other

(10)

bookmakers so that in the difference in odds an arbitrage opportunity arises. Mathematicaly this can be expressed as follows.

This formula stands for the sum of the highest odds (!_!) given for each match outcome on the bookmaker market. If this sum is smaller than 1, an arbitrage opportunity exists. The return from this strategy can be calculated using the following formula.

The second strategy is the 'long-‐position inter-‐market arbitrage' strategy. This is the same strategy as the intra-‐market strategy, but includes the exchange market for placing bets. The formulas used for this strategy are the same as for the intra-‐market, the only difference being that the highest odds !_! also include the odds from the exchange market. These exchange market odds have to be corrected for commission charged so that max(!!) for the exchange market is (!!(1 − !) − !) with ! being the commission

charged.

As Franck et al. (2012) and Vlastakis (2009) show, these former two strategies do offer arbitrage opportunities but do not fully exploit the other more relevant betting strategy that inter-‐market betting allows. This third strategy is the 'short-‐position inter-‐market arbitrage' strategy. This 'short-‐position inter-‐market arbitrage' strategy implies that a bet is 'bought' on the bookmakers market and 'sold' on the exchange market. This

strategy fully exploits the benefits of using the exchange market as it allows the bettor to hedge his bookmakers bets on the exchange market.

As previously discussed, for arbitrage opportunities to arise using the short selling strategy, the odds on the bookmakers market have to exceed the odds on the exchange market including commissions. This gives the following formula for calculating arbitrage possibilities. !̅_! > !!",! − ! 1 − ! ∑ 1 !̅_! < 1 ∏ = 1 ∑!̅_! − 1

(11)

Where !_! stands for the highest odds given by a bookmaker for a certain match, !_!",! stands for the odds given on the exchange market and ! stands for the commission charged by the betting exchange website. The commission chosen in this paper is 5%, as this is the maximum charged by Betfair. It should be noted that by increasing betting activity, Betfair lowers its commission for bettors to a minimum of 2%, potentially allowing more arbitrage opportunities with higher returns.

When an arbitrage opportunity is found, the expected profit can be calculated by the following formula.

Similar to other papers on this subject, the concept of arbitrage will be considered in a theoretical way and no cutoff value will be used for arbitrage returns. It can be discussed whether an arbitrage opportunity yielding a very small profit has any practical value for a bettor, but to compare the results from the data used in this paper to other academic literature these small return arbitrage opportunities will also be included in the results.

The results following from these formulas will then be compared to the results found by Franck et al. (2012) by using a t-‐test to test if they differ significantly.

Results

The main results about the Eredivisie data for the arbitrage strategies used are summarized in Table 1. The results of the tests for equality of the average returns are summarized in Table 2. The most remarkable difference in arbitrage frequency found between the Eredivisie and the big five European leagues are that intra-‐market opportunities arise more frequently: 5,25% in this paper compared to 0,5% found by Vlastakis et al. (2009) and 0,8% found by Franck et al. (2012). As discussed before, intra-‐market arbitrage is caused by difference in opinions amongst bookmakers about match outcomes. Because information is more uncertain and harder to come by for smaller leagues this should lead to higher arbitrage frequencies as is found in this paper.

Π = !̅!(!!",!− !)

(12)

The average return on the intra-‐market arbitrage opportunities is not significantly different from the returns found by Franck et al. (2012), with the remark that the number of observations is small: 16.

To see if the bookmakers' difference in opinions about match outcomes is bigger in the Eredivisie compared to a big league a test was performed to test if the odd spread is different between the Eredivisie and the Premier League in the 2012-‐2013 season. The odd spread is the highest odd minus the lowest odd given by the football-‐dat.co.uk set of bookmakers for a match outcome. The odd spread in the Eredivisie is found to be

significantly bigger at the 1% level compared to the Premier League. The exact results can be found in appendix Table A1. This confirms the idea that information uncertainty is higher in smaller leagues and thus leads to more frequent arbitrage opportunities.

Arbitrage frequency from the inter-‐market strategy is a bit higher than what was found by Franck et al. (2012), 6,56% compared to 5,0%. This strategy does not result in much higher arbitrage frequency in comparison to Franck et al. (2012) like the intra-‐market strategy does. The inter-‐market long postion only leads to four more cases of arbitrage compared to the intra-‐market strategy. An explanation for this is that the bookmakers odds have a larger spread and therefore are more often higher than the bet exchange

TABLE 1

ARBITRAGE OPPORTUNITIES IN DUTCH EREDIVISIE 2012-‐2013

All matches Arbitrage opportunities

Long position intra-‐market

Return on hedged bets (std. dev.) -‐0,0197 (0,0107) 0,0074 (0,0073)

Observations 305 16

Percentage 5,25%

Long position inter-‐market

Return on hedged bets (std. dev.) -‐0,0159 (0,0109) 0,0103(0,0092)

Observations 305 20

Percentage 6,56%

Short position inter-‐market

Return on hedged bets (std. dev.) -‐0,0149 (0,0135) 0,0085(0,0091)

Observations 305 73

Percentage 23,93%

(13)

odds (corrected for commissions). Adding exchange market odds to the bookmakers odds does not result in a rise in arbitrage frequency like it does in a market where the bookmakers spread is smaller as is the case in the paper by Franck et al. (2012).

The average returns are significantly smaller than the returns found by Franck et al. (2012). A possible explanation for this is that the number of observations is too small (20) to give an accurate representation of the population. Another explanation

might be that there is also more uncertainty on the bet exchange market about the true outcomes of the match. Although bookmakers odds are more spread out and lead to higher arbitrage frequency, the exchange market uncertainty might lead to a smaller average return on these positions.

For the short position inter-‐market strategy, the arbitrage frequency found is 23,93%, almost five percent higher then the frequency found for this strategy by Franck et al. (2012). Again this can be explained by bookmaker uncertainty leading to bigger differences in odds and allowing for more arbitrage opportunities.

TABLE 2

T-‐TESTS FOR AVERAGE RETURNS FOR ARBITRAGE OPPORTUNITIES

Long position intra-‐market

Hypothesis ! = 0,009

Sample mean !̅ = 0,0074

Sample std dev ! = 0,0073

Sample size ! = 16

T statistic ! = −0,876

Conclusion (at 5% significance) Do not reject !_!

Long position inter-‐market

Sample size ! = 20

Conclusion (at 5% significance) Reject !!

Short position inter-‐market

Sample size ! = 73

Conclusion (at 5% significance) Reject !_!

(14)

The average return found for the short position inter-‐market strategy is also significantly lower than the returns found in the big five European Leagues. Here, the number of observations is large enough to assume a normal distribution. An explanation for the lower returns could again be that the betting exchange market is not as good at predicting outcomes as in the big five leagues because trading volume is lower.

Arbitrage opportunities arise more frequently because of bookmaker differences but cannot be exploited fully because the exchange market is also less efficient.

The results found are ambiguous. Although arbitrage frequency is higher in the Dutch Eredivisie as expected from the literature, the average return found on the betting strategies is lower. These lower returns can be partially explained by theory but a case could also be made that because of higher frequency, average return could also be higher. The research performed in this paper suggests that there might be a negative correlation between increasing arbitrage frequency on the one hand and decreasing average returns on the other. No definitive conclusion can be drawn yet on this matter as only two 'categories' of leagues have been compared. The big five European leagues on one side and the Eredivisie on the other. To test whether there is a relation between increasing frequency and decreasing returns a full set of European leagues should be made categorzing them from big to small to see whether this relationship persists amongst these leagues.

Practical application of arbitrage betting

In this section the data is analyzed to see what the Eredivisie arbitrage could have resulted in for an individual bettor. Several scenarios are created in which is discussed what the profits from being an arbitrage bettor could have been. As discussed earlier in this paper, bookmakers may limit or cancel accounts which they expect to be involved in arbitrage betting. The bet exchange market does not have such limitations. For the sake of simplicity, it is assumed that the bettor places his bets in a such a way that he avoids being detected by the bookmakers. This can be done by placing rounded bets at the bookmakers (e.g. €100 instead of €98,76), avoiding frequent cash withdrawals and not only bet on arbitrage bets but also bet on normal matches, which have an expected negative return. By hedging these losing bets at the exchange market a zero profit

(15)

portfolio can be constructed where the losing part of the portfolio is taken at the bookmakers market to avoid detection.

In general, bookmakers sometimes charge commission costs on the use of credit cards for cash deposits. By using an E-‐wallet these costs can be avoided so transaction costs for subscribing at a bookmakers would be zero in terms of money and very small in terms the of effort of subscribing, often only a matter making an account and depositing money.

To identify arbitrage opportunities the bettor can either do research himself or use arbitrage software such as Rebelbetting. A Rebelbetting subscription costs €799 per year (Rebelbetting, 2014). The Eredivisie football season only lasts ten months but a 10 month subscription at Rebelbetting would be substantially more expensive due to discount pricing on longer subscriptions (10 months cost €1015). The Rebelbetting software offers an overview of arbitrage opportunities for many sports for more than 50 bookmakers and bet exchanges. It also includes a wide array of different betting options such as half time scores, total goals scored (e.g. over/under 2 goals) and individual goalscorers. All these arbitrage opportunities would be too time consuming for an individual bettor to investigate himself, but since this paper and its data only look at the match result at one point in time, an argument can be made that this software is

unneccesary for the individual bettor we describe in this section.

Because our data is matched in time for the friday afternoon, as discusssed in the Methodology section, we can also make a case for the argument that our bettor can identify the arbitrage opportunities himself without the help of paid software like

Rebelbetting but by using one of many websites for checking odds like oddsportal.com or

oddschecker.com and Excel or similar computational software.

The first scenario regards a bettor who chooses the most conservative approach for arbitrage betting. Each weekend he chooses one match and strategy out of the 9 matches played and chooses the match with the highest arbitrage return. He then reinvests his profit the next weekend using the same principle. Using this strategy, the bettor can earn a 55,8% return over 32 rounds in the Eredivisie (2 out of 34 round did not have an arbitrage opportunity).

(16)

The second scenario discussed is where the bettor looks for every match if there is an arbitrage opportunity using one of the three discussed strategies and then chooses the strategy with the highest return when an opportunity is available. He does not reinvest his money but bets seperately and independently on every arbitrage opportunity.

Using this strategy, a total of 77 matches yield an arbitrage opportunity, with an average return of 0,877%. Assuming he bets on every opportunity with an equal amout of money and without reinvesting his profits he earns a 67,5% profit on his bets.

In the third scenario we combine both scenario one and two to look at the bettor who invests in every arbitrage opportunity and then reinvests his profits from each weekend in the arbitrage opportunities for the next weekend. This would require frequent

movement of money from bookmaker to bookmaker but we assume our bettor is a skilled arbitrage bettor who avoids detection by the bookmakers. This would be the strategy that would be most profitable but also most difficult to implement because of potententital bookmaker detection. Using this strategy, the bets on every arbitrage opportunity like in scenario two and after every weekend he reinvests his profits in the next weekend as in scenario one. He bets on a total of 77 matches reinvesting 32 times yielding a 95,5% profit over the 2012-‐2013 season.

These scenarios are provided to give an insight in the potentital profitability of arbitrage betting in the Dutch Eredivisie. Scenario one is most likely to be succesful, since it only requires 32 bets and a relatively small amount of transferring of money between bookmakers. Scenario three is most likely to fail, since it requires a fair amout of

transferring money between bookmakers and more frequent bets on arbitrage bets that are more likely to evoke bookmaker suspicion.

It seems to be the case that for a small investor/bettor, arbitrage betting can be quite profitable. If the arbitrage bettor places his bets well and follows the instructions to avoid detection by the bookmakers, profits could potentially be anywhere between 50% and 100% per year, but online accounts note that this might be limited to several

thousand euros per year (The Guardian, 2012 and SBR Forums, 2014). The online community seems to be lively with several forums and websites dedicated to arbitrage betting, also called surebetting or smartbetting (Arbusers.com). In these forums, tips

(17)

and strategies are exchanged to avoid accounts being limited or closed down by bookmakers.

Conclusion and Discussion

This paper researches the frequency and returns of arbitrage in the online sports betting market for the Dutch Eredivisie in the 2012-‐2013 season and compares these to the big five European football leagues.

As the academic literature suggested, due to information uncertainty arbitrage frequency is higher in the Eredivisie. Average returns found are significantly lower than the returns found in the paper by Franck et al. (2012). An explanation for these lower returns is that information uncertainty is also bigger in the exchange market leading to a decrease in average returns. A possible negative correlation exists between increasing arbitrage frequency and decreasing arbitrage returns, but this paper does not offer enough evidence to draw this conclusion. Further research should be done to compare more leagues of different sizes to see what relation exists between arbitrage frequency and returns.

Another suggestion for further research could be the examination of different betting strategies. This paper and other academic literature only research the final match outcome, but betting markets offer a complete set of betting options like half-‐time scores, first goal scorer, first corner kick taken etcetera. Possibly these more exotic betting options have even more arbitrage opportunities lying within them.

Regarding the practical application of using arbitrage betting to make a profit as a bettor several scenarios are available that each yield different profits. Higher profits can be achieved by using more complicated arbitrage betting strategies but these also involve more risk of being detected. Arbitrage trading is profitable as long as the arbitrage bettor remains undetected by the bookmakers. Total profits can be as high as 10% per month, but is probably limited to a few thousand euros per year because of bookmaker detection.

(18)

Appendix Reference List

Arbusers (2014). Online forum for arbitrage betting. http://arbusers.com/ (Accessed 06-‐ 02-‐2014).

Betfair (2014).Online betting exchange platform. Key facts.

http://corporate.betfair.com/about-‐us/key-‐facts.aspx (Accessed 8 January 2014)

Betfair Data (2013). Historical data from the exchange platform Betfair.

www.data.betfair.com. (Accessed 16 December 2013)

Bwin.Party (2013). Leading online bookmaker. Market statement.

http://www.bwinparty.com/AboutUs/OurMarkets/OnlineSportsBetting.aspx (Accessed 8

January 2014)

TABLE A1

Table A1 shows results from the testing for difference in odds spread (highest odd minus lowest odd from the 10 different bookmakers for a certain match) in the bookmakers market.

Variable "x" is the English Premier League, variably "y" is the Dutch Eredivisie. #obs includes all matches and match outcomes. The average Eredivisie odd spread (0,7739) is significantly bigger at the 1% level than the Premier League spread (0,6112). The numbers given are the actual differences in odds (i.e. a 0,5 spread would be the diference between for example an odd of 1 and 1,5).

(19)

CROXSON, K. and JAMES READE, J. (2013). Information and Efficiency: Goal Arrival in Soccer Betting. The Economic Journal, doi: 10.1111/ecoj.12033

Footballdata.co.uk (2014). Historical data from 10 major bookmakers for the Dutch Eredivisie. http://www.football-‐data.co.uk/netherlandsm.php (Accessed 16 December 2013)

FRANCK, E. VERBEEK, E. and NÜENSCH, S (2012). Inter-‐market Arbitrage in Betting.

Economica, 80, 300-‐325

The Guardian, (2012). Free bets mean you can clean up while bookies meet their match.

Jason Shearer, The Guardian, 24-‐07-‐2012

KUYPERS, T. (2000). Information and Efficiency: an Empirical Study of a Fixed-‐Odds Betting Market. Applied Economics, 32, 1353-‐1363

LAMONT, O.A. and THALER, R.H. (2003). Anomalies: the Law of One Price in financial markets. Journal of Economic Perspectives, 17, 191-‐202

LEVITT, S. D. (2004). Why are gambling markets organised so differently from financial markets? The Economic Journal, 114, 223-‐246

MAKROPOULOU, V. and MARKELLOS, R. N. (2011). Optimal Price Setting in Fixed-‐Odds Betting Markets under Information Uncertainty. Scottish Journal of Political Economy,

58

MONTONE, M. (2012). Optimal Mark-‐Up and Arbitrages in the Betting Market.

Unpublished, University of Cassino & University of Naples

Rebelbetting.com (2014). Subscription costs. http://rebelbetting.com/pricing. (Accessed 05-‐02-‐2014)

(20)

SBR Forum (2014). Forum for sports betting. Section: Sportsbooks and the Industry.

http://forum.sbrforum.com/sportsbooks-‐industry/ (Accessed 06-‐02-‐2014)

SCHLEIFER, A. VISHNY, R.W. (1997). The Limits of Arbitrage. The Journal of Finance, 52,

35-‐55

SMITH, M.A. PATON, D. and VAUGHAN WILLIAMS, L . (2006). Market Efficiency in Person-‐to-‐Person Betting. Economica, 73, 673-‐689

SMITH, M.A. PATON, D and VAUGHAN WILLIAMS, L. (2009). Do bookmakers possess superior skills to bettor in predicting outcomes? Journal of Economic Behaviour and Organization, 71, 539-‐549

VLASTAKIS, N. DOTSIS, G. and MARKELLOS, R. N. (2009). How Efficient is the European Football Betting Market? Evidence from Arbitrage and Trading Strategies. Journal of

Arbitrage and market efficiency in sports betting markets