The favorite longshot bias in tennis tournaments
Early evidence for the favorite longshot bias
Download 419.07 Kb.
|
Early evidence for the favorite longshot biasIn this section of the literature review the early literature is discussed. Early literature mainly focuses on horse racing. This kind of market is called a pari-mutual market. A pari-mutual market is a market where all the money betted in a race is pooled. From this pool of money the costs for organizing the race are covered (called track take). After the race, the remaining money is paid to the winners according to the height of their bet (bettors that played the winning horse). In the early literature the studied countries were the United Kingdom and the United States. The first paper which mentioned the favorite longshot bias was [RMG49]. As a psychologist he was interested in how people behave in a race track setting. In such a pari-mutual betting market setting the betting behavior creates the odds the people play for which is an interesting phenomena for a psychologist. Griffith was interested if the bettors at the horse race tracks in the United Kingdom could predict the exact probabilities of a horse winning the race. He researched this by comparing the odds derived by the betting procedure of the bettors and transforming them in probabilities. Then he compared this probability with the exact probability that is derived by calculating how often a horse from each odds category actually wins. His first finding was that the socially determined odds (probabilities) were very similar to the true probability of a horse from an odds category winning. So the best horse was given the highest probability and the worst horse was given the lowest probability from the bettors. There was only one deviation between this socially determined probability and true probability which occurred at the extremes. Namely there was a systematic overvaluation of the bettors of the chances of a longshot (very low probability) horse to win the race. While the horses with the highest true chance to win were systematically undervalued by the bettors. After this first investigation of the betting behavior on a race track, [McG56] was another psychologist who investigated the behavior at a race track. McGothlin wanted to look at the stability of risk-taking behavior over a series of events. Do bettors show the same betting behavior at the first race as at the last race of the day? He investigated this by looking at the expected value of a 1 dollar bet in the 8 track odds categories made. Also he distinguished between the first and last (eighth) race per day. This study included eleven race tracks from two states (New York and California) of the United States. The results concerning the relationship between the expected winning probabilities calculated in each odds group and the socially determined odds were the same as found by [RMG49]. So the high probability horses showed a higher expected value than the low probability horses, again indicating an undervaluation of favorites and overvaluation of underdogs. Secondly McGothlin found that the behavior of bettors changed at the end of the day. Betting on the favorite becomes more valuable and even on average profitable when the track take is taken into account. Furthermore the heavy underdogs are less overvalued, but the average low probability bets are overvalued. So the extremes were estimated more accurately at the end of the day, in contrast to the horses with more average probability. Following the two previous authors, [Muk77] also investigated a horse racing track setting. But he investigated the existence of the favorite longshot bias for a much larger dataset. Even ten times larger than the dataset [RMG49] and [McG56] used. The Sample Ali used was 20507 horse races at three different New York horse race tracks in the years 1970 to 1974. Ali had a new feature where the allocation to categories were based on. Previously Griffith based the allocation in groups purely on odds. In this way there was a chance that two competing horses were in the same group. Therefore an average is biased, because only one horse can win the race. So a representative average horse cannot be captured. Ali made sure that in each category there were no horses competing in the same race. The results showed that on every track in each year there was a favorite longshot bias. So in a pari-mutual setting on every track the bettors overvalued the chance that an extreme underdog would win the race, while the extreme favorites in the races were on average undervalued. Another study done in the United States in a pari-mutual market was done by [Pet82]. They investigated a dataset containing horse races from the Atlantic City racetrack. In this paper they came to the same conclusions as in [RMG49]. Firstly, they found that the odds resulting from the betting behavior of the bettors are a really good indication of how the horses will finish. The other main conclusion is that on average the favorites are ‘good’ bets and the least favorites are ‘bad’ bets. Arsch & Malkiel are one of the first authors discussing the link between betting and financial markets. They noticed the similarities: in both situations one does not know what ones future earnings are with certainty. There are a lot of participants involved in these markets. There is a lot of information available and this information can come with advice of professionals. Involvement in these markets also provides knowledge of what others in the market do. [MRu85] investigated a financial market and found an indication for the favorite longshot bias. He investigated the put and call market. One of the main conclusions in this paper was that short- maturity out-of-the-money calls are priced significantly higher than the Black-Scholes model would predict. This means that call options that have a striking price higher than the market price are valued higher by the option buyers than the Black-Scholes model predict. In the paper of Rubinstein this finding is the only significant deviation of the valuation of the Black-Scholes model and the true value of the options. This could be evidence that the longshot in option valuation is overvalued, indicating a same finding for the longshots as in the sports betting literature and maybe the financial market exhibits the same biases as the betting market. Rubinstein was not the only one noting similar behavior in sport betting markets and financial markets. Another paper investigating the favorite longshot bias in a financial market has been written by [Hod]. They examined this for the S&P500 futures, the FTSE 100 futures and the British Pound/US Dollar futures in the period March 1985 to September 2002. A future is an agreement to buy or sell an asset or a financial value at some time in the future that has to be supplied by the seller to the buyer. Firstly for call options on the FTSE 100, they found a relationship between probabilities and true mean returns that is very similar to the favorite longshot bias in sport markets for call options. For these FTSE 100 3- and 1month call options a long shot bias exists if these options are deep out-of-the-money. Secondly, for the S&P 500 there was even a profit for buying deep-in-the-money calls and increasingly bigger losses as the calls were going out-of-the-money. In the options on the British pound/US Dollar trading market they found no evidence concerning the favorite longshot bias. They concluded that the patterns they observed in both S&P 500 and FTSE 100 were analogous to the favorite longshot bias in sport betting markets. Directory: pub pub -> Project Scoring Template Pre-Construction Phase New Core/Branch Campus Projects pub -> Acm word Template for sig site pub -> The german unification, 1815-1870 pub -> Baltic Olympiads in Informatics: Challenges for Training Together pub -> Preparation of Papers for ieee transactions on medical imaging pub -> Forthcoming meetings and conferences pub -> Asilomar Conference on Circuits, Systems & Computers, Pacific Grove, ca, November 1978, pp. 55 58 pub -> Forthcoming meetings and conferences pub -> Harmonised compatibility and sharing conditions for video pmse in the 7 9 ghz frequency band, taking into account radar use Download 419.07 Kb. Share with your friends:
|