NFL Betting Biases, Profitable Strategies, and the Wisdom of the Crowd.
The wisdom of the crowd hypothesis suggests that the decision of a group of individuals will outperform decisions from a single expert (Hastie & Kameda, 2005; Larrick & Soll, 2006; Soll & Larrick, 2009; Sunstein, 2006). This hypothesis is derived from mathematical principles that argue that the more individuals in the crowd will cancel out the noise and extract a more precise signal (Hogarth, 1978; Makridakis & Winkler, 1983). Individuals who receive identical information will perceive it differently depending on personal preferences, expectations, and experiences (Vasile et al., 2012). Therefore, by combining all individual estimates, individual errors and irrational choices should disappear, such as with participants on "Who Wants to be a Millionaire" hoped for when they would ask the audience for questions they were unsure about.
For example, Lorge et al. (1958) find that when asking students the temperature of the classroom, the student's average guess was only 0.4 degrees from the actual temperature, which was more accurate than 80% of the individual's estimate. Additionally, Treynor (1987) asked students to determine how many jellybeans were in a jar. The average estimate was 871, which was close to the actual number of 850, and better than 98% of the students' personal guesses. Similarly, Galton (1907) asked people at a regional fair competition to guess the weight of an ox. The average estimate was 1,197 which was only one pound away from the actual weight of 1,198 of the ox.
Additionally, Yaniv and Milyavsky (2007) find that individuals can make better estimates after consulting with a group of advisors. As it relates to finance, Pelster and Breitmayer (2017) find that crowd analyses of stocks from a social trading platform can provide explanatory power for stock returns, demonstrating the importance of the crowd in financial decision making. Following the wisdom of the crowd hypothesis, as the number of informed bettors increases, there should be less biased decisions, which should result in a better ability to predict outcomes in the NFL betting market.
In 2016, the American Gaming Association estimates that wagers on football games exceeded more than $90 billion for the second straight season, with approximately $4.7 billion wagered on the Super Bowl alone with about 97% of wagers done illegally outside of Nevada. Given the vast amounts of money involved, the efficiency of the wagering markets is of keen interest. In the wagering market, the bookmaker has a role similar to a stock exchange specialist, as both match buyers and sellers for a fee, with the normal fee, also called the vigorish, being 10%. Due to the vigorish, a bettor must wager $110 to receive a profit of $100 if they win the bet. As such, a betting strategy that provides a profit of more than 50% of the time presents evidence that the wagering market is statistically inefficient. However, a betting strategy must be able to produce a profit of greater than 52.38% to provide evidence that the wagering market is economically inefficient due to the vigorish. (1)
The two most common bets in the National Football League (NFL) wagering markets are the point spread and the over/under. The point spread is the forecasted amount of points by which the favorite team is expected to defeat the underdog. Therefore, a wager placed on the favorite team only wins if the favorite team has a winning margin greater than the point spread. If the favorite team loses or wins by less than the point spread, a bet on the underdog is a winning ticket. In the over/under market, the bookmaker predicts the total points that are expected for the two teams combined. After the over/under figure is set, bettors can place wagers on if the total points scored by the two teams combined will be over or under the posted total. If the outcome in the points spread or over/under wager is a tie, all money is refunded.
Most individuals believe that the point spread and over/under wagers are calculated to create equal amounts of dollars wagered on both sides of the bet (i.e., equal amount wagered on the favorite and underdog and the over and under) (Avery & Chevalier, 1999; Dana & Knetter, 1994; Gandar et al., 1988; Gray & Gray, 1997; Lee & Smith, 2002; Snowberg et al., 2005). If the bookmaker can collect equal wagers on both sides of the bet, they guarantee themselves a risk-free profit due to the vigorish. However, Levitt (2004) finds that the bookmaker rarely has equal wager amounts on both sides of the bet. The bookmakers use their expertise to set the lines, as they are better at predicting game outcomes compared to the typical bettor, which provides greater profits to the casinos, which is also confirmed by Paul and Weinbach (2007, 2011). Therefore, any biased decisions by bettors could be used to exploit bettors further and result in higher profits for the bookmaker.
Previous research provides evidence of both statistical and economic inefficiencies in the sports betting market, such as Zuber et al. (1985) who provides proof that the NFL betting market is inefficient. Furthermore, research finds that betting on the home team when they are the underdog can provide economically significant profits (Amoako-Adu et al., 1985; Golec & Tomarkin, 1991; Gray & Gray, 1997; Shank, 2018). Additionally, Vergin and Sosik (1999) find that in games that have national focus (i.e., Monday Night Football and playoff games) the home team produces a win rate of nearly 60%, with the win rate increasing when the home team is the underdog. Research finds that in the NFL the weather advantage for the home team can provide economically significant profits for the points spread (Borghesi, 2007) and over/under wagers (Borghesi, 2008). Similarly, Paul (2017) finds that atmospheric conditions such as humidity and wind speed can cause the betting market to be inefficient. Wever and Aadland (2012) provide a strategy of betting on underdogs with large closing spreads can provide a profit rate of near 60%, while Nichols (2012) finds that teams that travel east and change time zones statistically increase the chance of the home team winning, although it is not economically significant. However, this home field advantage seems to only occur in the NFL as Gandar et al. (2001) find that the home field advantage is not evident in the National Basketball League or Major League Baseball.
Conversely, Sauer et al. (1988) Dare and MacDonald (1996) and Dare and Holland (2004) find that betting on home underdogs does not provide statistically significant returns. Research concludes that betting the under in the first week of the NBA (Girdner et al., 2013) and the NFL (DiFilippo et al., 2014) is a profitable strategy. Finally, Paul and Weinbach (2002) find a profitable strategy on betting the under on games with high over/under, while Shank (2018) finds a profitable strategy on betting the over when the total line is either very small or very large.
It is intuitive, given the previous inefficiencies, that using the wisdom of the crowd could create economic profits from the betting market. However, Griffith (1949) and McGlothlin (1956) originally introduced the term favorite-longshot bias, which describes horse racing bettors prefer to bet the longshot rather than the favorite. Snowberg and Wolfers (2010) explain the favorite-longshot bias as bettors are irrational and have a distortion of probability as they overestimate the chances of the longshot horses winning compared to the betting odds following prospect theory. However, in other sports betting markets, the favorite-longshot bias appears to be opposite as bettors have a biased preference to bet on the favorite in the NFL (Humphreys et al., 2013; Paul & Weinbach, 2011), NBA (Paul & Weinbach, 2005a), and MLB (Woodland & Woodland, 1994). Additionally, when examining the totals market, research demonstrates that bettors have a biased preference to bet the over in the NFL (Paul & Weinbach, 2011; Humphreys et al., 2013), college and arena football (Paul & Weinbach, 2005b), and European soccer (Paul & Weinbach, 2009).
Furthermore, Paul and Weinbach (2011) find a profitable strategy using a contrarian approach by betting on the underdog when over 70% of bettors are betting on the favorite, but do not find a profitable strategy in the over/under market. Simmons et al. (2009) use a sample of NFL fans who may or may not have any betting experience and find that if you change the spread of the game, and increase the number of points the favorite team must win by to cover the spread, NFL fans will still select the favorite. Similarly, research finds that bettors are likely to follow the "hot hand" bias, which is a strategy derived from betting on teams that have performed well relative to the spread in previous weeks in the NBA (Camerer, 1989; Brown & Sauer, 1993; Paul & Weinbach, 2005a; Paul et al., 2011). Finally, Paul et al. (2014) find that as money wagered increases, so does the percentage bet on the favorite in the NFL and NCAA football.
This study examines the betting decisions of individuals on both the points-spread and over/under betting market in the context of the wisdom of the crowd. The results confirm previous literature that shows that NFL bettors prefer to bet the favorite and the over. Furthermore, this study adds to the literature by demonstrating that the biased decision to place wagers on the favorite increases as more bettors place wagers. Additionally, bettors prefer to bet against the line movement to gain better odds, and on the favorite when the underdog does not have the hot hand. Moreover, the results indicate that bettors are more likely to bet the over if the home team has the hot hand in the over/under market. Furthermore, a nonlinear betting preference based upon the spread of the game is found where bettors are less likely to wager on the favorite when the spread is small or large. Finally, as the percentage of bettors betting on the same side of the point spread or the over/under market increases over 60%, gamblers are more accurate in their wagers.
Data on bettor information is collected from Covers.com for the 2015-2016 and 2016-2017 seasons with the odds for the spread and over/under market purchased from oddswarehouse.com for the regular season and playoffs provided a sample of 527 games. Covers.com is a sportsbook simulation that has previously been referenced in publications such as USA Today, New York Times, and ESPN The Magazine where users can join leagues to make wagers and compete to win prizes, which provides a dataset rich with knowledgeable sports bettors. Covers.com claims that,
Covers Public Money data is a unique way to evaluate the money spent on individual teams throughout the season by the betting public. Although our numbers come from Covers free League Contests, with over 50,000 contest players participating this past year; these numbers provide an accurate comparison of where the 'real' betting public is laying their cash.
Additionally, while it is free to join leagues, participants can earn money in prizes for good performance. Simmons et al. (2009) argue that the wisdom of the crowd only works when the crowd's judges are knowledgeable, motivated to be accurate, independent, and diverse, which is provided by the sample from Covers.com. For each game, the number of bettors on both sides of the wager for the spread and over/under market are collected to find the percentage of individuals on each side of the bet. (2)
Table 1 provides definitions for all variables used in the study. Table 2 presents the summary statistics for these variables. I show that on average 56% of bettors wager on the favorite and the over. Additionally, the average number of individuals placing bets for each game is 2,720 against the spread and 1,848 for the over/under.
Figure 1 displays the betting behavior based upon the spread and the posted total in the over/under. For each, the spread and the total are separated into seven quantiles to achieve a similar amount of observations in each group. The results reveal a nonlinear effect of the posted spread on betting behavior as the percentage bet on the favorite and the percentage of times the majority bet on the favorite consistently rises from the first septile when the spread is -1 to when the spread is between -3.5 and -4 with the majority betting the favorite over 85% of the time. However, as the spread increases after 4, the percentage bet on the favorite and the times the majority bet on the favorite slowly decreases. Significant differences are found between the spread categories using dummy variables for each of 7 quantiles. These results are not shown to save space. Therefore, the following analysis will examine if there are possible non-linear betting preferences on the favorite. Additionally, Figure 1 shows that as the posted total in the over/under market increases, more bettors wager on the over.
A simple regression model is employed to examine the impact of the spread on betting behavior similar to Paul and Weinbach (2011)
% bet on the favorite or majority betting on favorite [dummy.sub.i] = [[alpha].sub.0] + [[beta].sub.1]Point [Spread.sub.i] + [[beta].sub.2] Road [Favorite.sub.t] + [[beta].sub.3]# of [Bettors.sub.i] + [[beta].sub.4] Spread [Movement.sub.i] + [[beta].sub.5] Favorite Hot [Hand.sub.i] + [[beta].sub.6] Underdog Hot [Hand.sub.i] + [[beta].sub.7] Point [Spread.sup.2.sub.i] + [[epsilon].sub.i] (1)
Where the dependent variable is either the percentage of bettors who place a wager on the favorite or a dummy variable that equals 1 if the majority of bettors bet on the favorite, or 0 otherwise. The independent variables are: The point spread for the game (presented as a negative number with greater favorites having a more negative figure); the road favorite which is a dummy variable that equals to 1 if the favorite is the road team; the number of bettors that placed a wager; the spread movement defined as the difference between the opening odds spread and the closing spread; the hot hand of the favorite team measured by the number of times the favorite has covered the spread in the last 5 games; the hot hand of the underdog measured by the number of occasions the underdog has covered the spread in the last 5 games; the point spread squared.
The dummy for a road favorite is based on research showing that road favorites are commonly overbet (Paul & Wein-bach, 2011; Golec & Tamarkiin, 1991; Gray & Gray 1997; Humphreys et al., 2013). The number of bettors variable is included because the wisdom of the crowd hypothesis suggests that the more bettors that place a wager, the less biased the group's decision should be. Paul et al. (2014) find that when the volume wagered on football games increases, so does the percentage bet on the favorite; however, instead of examining the amount wagered, this study examines the number of bettors, which gives each bettor an equal share regardless of the amount wagered.
In the betting market, an opening spread is created days or weeks prior to the game, as well as the closing spread, which all bets placed compete against regardless of when they are placed. The bookmaker often changes the odds because, from the bookmaker's perspective, too much money is placed on one side of the bet. Therefore they move the line in accordance. There are betting theories based on the line movement. Some believe that bettors should follow the line as they believe that the wisdom of the crowd provides additional information about the game (i.e., if the spread moves from -3 to -5 bettors should place a wager on the favorite as more people are betting on the favorite). This is similar to the ADR underpricing in that IPO's that receive a price upgrade exhibit greater returns (Hanley, 1993) Conversely, some believe that bettors should wager against the line movement, also known as contrarian betting, (i.e., if the spread moves from -3 to -5 bettors should place a wager on the underdog) to exploit sports bettor tendencies to select the favorite and to gain better odds (i.e., betting on the underdog when the spread is -5 provides a better wager than when it was -3).
In finance, research finds that momentum strategies like buying stocks that have recently increased can be profitable (Jegadeesh & Titman, 1993; Carhart, 1997; Jegadeesh & Titman, 2001). In the betting market, this is called the "hot hand," which research shows that bettors are more likely to bet on teams with the hot hand in the NBA (Camerer, 1989; Brown & Sauer, 1993; Paul & Weinbach, 2005a), but has not been examined in the context of the NFL to the best of my knowledge. Therefore, the study includes variables to detect the hot hand of covering the spread in recent games for both teams.
If bettors prefer to bet on the favorites, [[beta].sub.1] will be negative and significant. If bettors prefer to place wagers on the road favorites, [[beta].sub.2] will be positive and significant. If more individuals who place a bet on the game create a less biased group decision following the wisdom of the crowd, [[beta].sub.3] will be negative. [beta].sub.4] will be negative if bettors follow the spread and bet on the favorite after the line moves towards the favorite. Conversely, [[beta].sub.4] will be positive if bettors favor the contrarian strategy and bet on the underdog when the betting line moves toward the favorite, as this improves the odds of the underdog. [[beta].sub.5] will be positive if bettors prefer to bet on teams that have hot hand of covering the spread in recent games. [[beta].sub.6] will be negative if bettors prefer to bet on the favorite when the underdog does not have the hot hand of covering the spread in recent games. [[beta].sub.7] will be negative if bettors prefer to bet less on the favorite when the spread is smaller or larger.
To examine betting behavior in the over/under market a regression model similar to equation 1 is employed % bet on the over or majority betting on over [dummy.sub.i] = [[alpha].sub.0] + [[beta].sub.1] Posted [Total.sub.i] + [[beta].sub.2]# of [Bettors.sub.t] + [[beta].sub.3]Spread [Movement.sub.i] + [[beta].sub.4]Home Over Hot [Hand.sub.i] + [[beta].sub.5]Away Over Hot [Hand.sub.i] + [[epsilon].sub.i] (2)
Where the dependent variable is either the percentage of bettors who place a wager on the over or a dummy variable that equals 1 if the majority of bettors bet on the over, or 0 otherwise. The independent variables are: the posted total defined as the closing total points for the game created by the bookmaker; the number of bettors; the home over hot hand defined as the number of times the home team covered the over in the recent 5 games; the away over hot hand defined as the number of times the away team covered the over in the recent 5 games. (3) [[beta].sub.4] and [[beta].sub.5] will be positive if bettors prefer to bet on teams that have covered the over in recent games.
Research examines the impact of wager volume (Paul et al., 2014) and hot hand (Paul et al., (2011) in the context of increased biased betting on the favorite. However, they do not examine the impact it has on the over/under market.
Table 3 presents the results for bettor preferences in the point spread market in Panel A, and over/under market in Panel B using robust standard errors to control for heterosce-dasticity. In the point spread market, a significant nonlinear effect is found on the spread as bettors are more likely to place wagers on the favorite when there is a medium spread and less likely to wager on the favorite when the spread is large or small. Bettors especially like to bet on the favorite if they are the road favorite. Moreover, the results show that as more bettors place wagers, the bias of betting on the favorite increases. The coefficient of 0.133 demonstrates that for every 100 more bettors that place a wager, the percentage of bettors that wager on the favorite will increase by 0.133%, which contradicts the wisdom of the crowd hypothesis.
Additionally, the spread movement is positive and significant with a coefficient of 0.774 demonstrating that for every point that the line moves against the favorite, 0.77% more bettors will bet on the favorite in order to gain better odds. This result also goes against Simmons et al. (2009) who find that bettors continue to wager on the favorite even after the line moves disadvantageously; however, in their study, their participants were unaware of the disadvantageous movement. Finally, the results show that bettors are more likely to bet on the favorite if the underdog does not have the hot hand, which is consistent with previous research (Camerer, 1989; Brown & Sauer, 1993; Paul & Weinbach, 2005a). The probit regression in column 2 provides similar results. These results are robust to using different lags of the hot hand, such as the past three or seven games.
Panel B presents the results for betting behavior in the over/under market. Consistent with previous literature, bettors prefer to bet on the over, as the posted total increases by 1.178 points, the percent of bettors who place a wager on the over will increase 1 percent. Additionally, the results show that bettors are more likely to wager on the over if the home team has covered the over in more recent games, despite Paul et al. (2004) finding that the hot hand bias is not a profitable strategy in the NBA. Moreover, the results show that line movement and the number of bettors does not have a significant impact on the percentage that bet has on the over, unlike the point spread market.
Table 4 presents the results to examine profitable strategies based upon the percent of bettors on each side of the spread market in Panel A and over/under market in Panel B following Paul and Weignbach (2007). The idea of splitting the samples into different categories is based upon the wisdom of the crowd as more bettors wagering on the same side should be more accurate. While the percentages selected are arbitrary, they are chosen due to being simple rules for bettors to follow and approximately splitting into three groups of an equal number of observations. Column 1 presents the percentage selected, column 2 presents the number of wins for the category, column 3 presents the number of losses for the category, column 4 presents the win percentage for the for the selected percentage group, column 5 shows the earnings for bettors that bet with the public (that is, betting on the favorite/home team when the percentage is greater than 50% or betting on the underdog/away team when the percentage is less than 50%). Additionally, column 6 presents the earnings for a bettor who takes a contrarian strategy and bets against the public. Both betting strategies provide the returns to a hypothetical bettor who wagers $110 based on the strategy and wins $100 due to the 10% vigorish.
The results show that, when between 50% to 60% of bettors bet on the same team in the point spread market and the same side in the over/under market they are largely inaccurate, as using a contrarian betting strategy can be very profitable consistent with Paul and Weinbach (2007, 2011). However, when more than 60% of bettors are on the same page and wager together, they are somewhat accurate, as they win more than 50% of the time. However, the results are slightly profitable in the point spread market, and unprofitable in the over/under market.
This paper examines betting preferences in the NFL point spread and over/under markets using betting data from odd-swarehouse.com and bettor data from covers.com. The results show that betting preferences go beyond the previously known betting biases of wagering on the favorite and the over. This paper adds to the literature by finding that the percentage of bettors who wager on the favorite increases as the number of bettors increases. Additionally this is the first paper to demonstrate that bettors prefer to bet against the line movement in the point spread market to increase the odds on the favorite team. Both of these results conflict with the wisdom of the crowd hypothesis. Furthermore, the results show that bettors prefer betting against the underdog with the hot hand in the point spread market and wager the over when the home team has the hot hand. Finally bettors in the point spread market display nonlinear preferences as they are less likely to bet the favorite when the spread is low or high.
The wisdom of the crowd hypothesis argues that as more individuals boast their opinion, a better group decision can be made. The results show that when over 60% of bettors bet on the same team in the point spread market and the same side in the over/under market, they are much more accurate than when the percentage of bettors on each side of the bet is close. These results show that when there is the least amount of noise in the group decision, the wisdom of the crowd is more accurate.
These results have important implications for both sports bettors and bookmakers. First, sports bettors need to understand their biased decisions to make more rational decisions in an attempt to maximize profits. Second, as point spread bettors prefer betting on the favorite, especially if it is the road team, against spread movement, and against the hot hand of the underdog playing, the bookmaker can take advantage of this by setting the point spread to take advantage of bettor-biased decisions. Similarly, the bookmaker can increase the over/under for games, especially when the home team has the hot hand for covering the over to get more people betting the over while simultaneously increasing the odds that the final score will be under the posted total. Overall, this paper finds that while it is commonly known that betting biases typically result in more losses, bettors continue to make irrational decisions, similar to investors who consistently display overconfidence and commit investment biases despite the research showing they lead to suboptimal results.
(1) If bettors wager $110 and win they will collect $100 profit and if they lose they lose all of their $110. Therefore $110 divided by $210 provides the 52.38% that bettors must average in order to have a profitable strategy after accounting for the vigorish.
(2) As this data provides data only on the number of bettors and not the amount they wager there is no endogeneity as the bookmakers make changes to the spread and line based upon wagers not number of wagers.
(3) The hot hand of the home and away team are used instead of the favorite and underdog as every game has a home and away team, while in games where the point spread is 0 there are no favorites or underdog which would decrease the sample size.
Amoako-Adu, B., Marmer, H., & Yagil, J. (1985). The efficiency of certain speculative markets and gambler behavior. Journal of Economics and Business, 37, 365-378.
Avery, C, & Chevalier, J. (1999). Identifying investor sentiment from price paths: The case of football betting (Digest Summary). Journal of Business, 72, 493-520.
Borghesi, R. (2007). The home team weather advantage and biases in the NFL betting market. Journal of Economics and Business, 59, 340-354.
Borghesi, R. (2008). Weather biases in the NFL totals market. Applied Financial Economics, 18, 947-953.
Brown, W., & Sauer, R. (1993). Does the basketball market believe in the hot hand? Comment. The American Economic Review, 83, 1377-1386.
Camerer, C. (1989). Does the basketball market believe in the "hot hand?" The American Economic Review, 79, 1257-1261.
Carhart, M. (1997). On persistence in mutual fund performance. The Journal of Finance, 52, 57-82.
Dana, J., & Knetter, M. (1994). Learning and efficiency in a gambling market. Management Science, 40, 1317-1328.
Dare, W., & Holland, A. (2004). Efficiency in the NFL betting market: Modifying and consolidating research methods. Applied Economics, 36, 9-15.
Dare, W., & MacDonald, S. (1996). A generalized model for testing the home and favorite team advantage in point spread markets. Journal of Financial Economics, 40, 295-318.
DiFilippo, M., Krieger, K., Davis, J., & Fodor, A. (2014). Early season NFL over/under bias. Journal of Sports Economics, 15, 201-211.
Galton, F. (1907). Vox populi (The wisdom of crowds). Nature, 75, 450-451.
Gandar, J., Zuber, R., & Lamb, R. (2001). The home field advantage revisited: A search for the bias in other sports betting markets. Journal of Economics and Business, 53, 439-453.
Gandar, J., Zuber, R., O'Brien, T., & Russo, B. (1988). Testing rationality in the point spread betting market. The Journal of Finance, 43, 995-1008.
Girdner, C., Davis, J., Fodor, A., & Kirch, D. (2013). Early season NBA over/under bias. Journal of Prediction Markets, 7, 1-9.
Golec, J., & Tamarkin, M. (1991). The degree of inefficiency in the football betting market: Statistical tests. Journal of Financial Economics, 30, 311-323.
Gray, P., & Gray, S. (1997). Testing market efficiency: Evidence from the NFL sports betting market. The Journal of Finance, 52, 1725-1737.
Griffith, R. (1949). Odds adjustments by American horse-race bettors. The American Journal of Psychology, 62, 290-294.
Hanley, K. (1993). The underpricing of initial public offerings and the partial adjustment phenomenon. Journal of Financial Economics, 34, 231-250.
Hastie, R., & Rameda, T. (2005) The robust beauty of majority rules in group decisions. Psychological Review, 112, 494
Hogarth, R. (1978). A note on aggregating opinions. Organizational Behavior and Human Performance, 21, 40-46.
Humphreys, B., Paul, R., & Weinbach, A. (2013). Bettor biases and the home-underdog bias in the NFL. International Journal of Sport Finance, 8, 294-311.
Jegadeesh, N, & Titman, S. (1993). Returns to buying winners and selling losers: Implications for stock market efficiency. The Journal of Finance, 48, 65-91.
Jegadeesh, N, & Titman, S. (2001). Profitability of momentum strategies: An evaluation of alternative explanations. The Journal of Finance, 56, 699-720.
Larrick, R., & Soll, J. (2006). Intuitions about combining opinions: Misappreciation of the averaging principle. Management Science, 52, 111-127.
Lee, M., & Smith, G. (2002). Regression to the mean and football wagers. Journal of Behavioral Decision Making, 15, 329-342.
Levitt, S. (2004). Why are gambling markets organized so differently from financial markets? The Economic Journal, 114, 223-246.
Lorge, I., Fox, D., Davitz, J., & Brenner, M. (1958). A survey of studies contrasting the quality of group performance and individual performance, 1920-1957. Psychological Bulletin, 55, 337-372.
Makridakis, S., & Winkler, R. (1983). Averages of forecasts: Some empirical results. Management Science, 29, 987-996.
McGlothlin, W (1956). Stability of choices among uncertain alternatives. The American Journal of Psychology, 69, 604-615.
Nichols, M. (2014). The impact of visiting team travel on game outcome and biases in NFL betting markets. Journal of Sports Economics, 15, 78-96.
Paul, R. J. (2017). The Impact of Atmospheric Conditions on Actual and Expected Scoring in the NFL. International Journal of Sport Finance, 12, 14-31.
Paul, R., & Weinbach, A. (2002). Market efficiency and a profitable betting rule evidence from totals on professional football. Journal of Sports Economics, 3, 256-263.
Paul, R., & Weinbach, A. (2005a). Bettor misperceptions in the NBA: The overbetting of large favorites and the "hot hand." Journal of Sports Economics, 6, 390-400.
Paul, R., & Weinbach, A. (2005b). Bettor preferences and market efficiency in football totals markets. Journal of Economics and Finance, 29, 409-415.
Paul, R., & Weinbach, A. (2007). Does sportsbook.com set point spreads to maximize profits? The Journal of Prediction Markets, 1, 209-218.
Paul, R., & Weinbach, A. (2009). Are behavioral biases consistent across the Atlantic?: The over/under market for European soccer. Journal of Gambling Business and Economics, 3, 89-101.
Paul, R., & Weinbach, A. (2011). NFL bettor biases and price setting: Further tests of the Levitt hypothesis of sportsbook behaviour. Applied Economics Letters, 18, 193-197
Paul, R., Weinbach, A., & Humphreys, B. (2011). Revisiting the "hot hand" hypothesis in the NBA betting market using actual sportsbook betting percentages on favorites and underdogs. Journal of Gambling Business & Economics, 5, 42-56.
Paul, R., Weinbach, A., & Small, K. (2014). The relationship between sportsbook volume, forecast accuracy and market efficiency: The NFL and NCAA football. Journal of Prediction Markets, 8, 29-42.
Paul, R., Weinbach, A., & Wilson, M. (2004). Efficient markets, fair bets, and profitability in NBA totals 1995-96 to 2001-02. The Quarterly Review of Economics and Finance, 44, 624-632.
Pelster, M., Breitmayer, B., & Massari, F. (2017). Swarm Intelligence? Stock Opinions of the Crowd and Stock Returns. Working Paper. Retrieved from https://ssrn.com/abstract=2787744 or http://dx.doi.org/10.2139/ssrn.2787744
Sauer, R. (1998). The economics of wagering markets. Journal of Economic Literature, 36, 2021-2064.
Shank, C. (2018). Is the NFL betting market still inefficient? Journal of Economics and Finance,. 42, 818-827.
Simmons, J., Nelson, L., Galak, J., & Frederick, S. (2011). Intuitive biases in choice versus estimation: Implications for the wisdom of crowds. Journal of Consumer Research, 38, 1-15.
Snowberg, E., & Wolfers, J. (2010). Explaining the favorite-long shot bias: Is it risk-love or misperceptions? Journal of Political Economy, 118, 723-746.
Snowberg, E., Wolfers, J., & Zitzewitz, E. (2005). Information (in) efficiency in prediction markets. In L.V. Williams (Ed.), Information Efficiency in Financial and Betting Markets. (pp. 366-386). Cambridge: Cambridge University Press.
Soll, J., & Larrick, R. (2009). Strategies for revising judgment: How (and how well) people use others' opinions. Journal of Experimental Psychology: Learning, Memory, and Cognition, 35, 780-805.
Sunstein, C. (2006). Infotopia: How many minds produce knowledge. New York: Oxford University Press.
Treynor, J. (1987). Market efficiency and the bean jar experiment. Financial Analysts Journal, 43, 50-53.
Vasile, D., Sebastian, T, & Radu, T. (2012). An introduction to behavioral corporate finance. Annals of the University of Oradea, Economic Science Series, 21, 471-476.
Vergin, R., & Sosik, J. (1999). No place like home: an examination of the home field advantage in gambling strategies in NFL football. Journal of Economics and Business, 51, 21-31.
Wever, S., & Aadland, D. (2012). Herd behaviour and underdogs in the NFL. Applied Economics Letters, 19, 93-97.
Woodland, L., & Woodland, B. (1994). Market Efficiency and the Favorite-Longshot Bias: The Baseball Betting Market. The Journal of Finance, 49, 269-279.
Yaniv, I., & Milyavsky, M. (2007). Using advice from multiple sources to revise and improve judgments. Organizational Behavior and Human Decision Processes, 103, 104-120.
Zuber, R., Gandar, J., & Bowers, B. (1985). Beating the spread: Testing the efficiency of the gambling market for National Football League games. Journal of Political Economy, 93, 800-806.
Corey A. Shank (1)
(1) Dalton State College
Corey Shank is an assistant professor of finance at Dalton State College. His major research interests are in the fields of behavioral finance, market efficiency, and asset pricing. Corey's research has been published in journals such as Review of Financial Economics, Review of Behavioral Finance, Journal of Economics and Human Biology, and Journal of Financial Markets and Portfolio Management.
Table 1. Variable Definitions Percentage Bet The percentage of bettors who bet on the favorite in the point spread market or the over in the over/under market. Point Spread The forecasted number of points from the bookmaker in which the betting favored team is expected to defeat the underdog where a larger negative number means more of a favorite. Over/Under The forecasted number of points from the bookmaker that are expected to be scored by both teams together. Road Favorite A dummy variable that is equal to 1 if the road team is the favorite. # of Bettors The number of hundreds of bettors who placed a wager on the game on covers.com Spread/Line Movement The difference between the opening spread (in the point spread market) or line (in the over/under market) compared to the closing line. Momentum The number of games that the team covered the spread (in the point spread market) or covered the over (in the over/under market) in the past five games. Notes: This table provides definitions for the variables used in this study. Table 2. Summary Statistics Variable Mean Standard Deviation Min Max Percentage Bet on Favorite 0.560 0.094 0.28 0.81 Point Spread -4.895 3.529 -17.50 0.00 Home Underdog 0.287 0.453 0.00 1.00 Spread Movement -0.878 1.923 -8.00 7.50 Favorite Spread Momentum 2.962 1.333 0.00 5.00 Underdog Spread Momentum 1.981 1.361 0.00 5.00 Spread # of Bettors 27.20 5.568 7.75 42.30 Over/Under # of Bettors 18.48 3.863 4.99 31.50 Percentage Bet on Over 0.560 0.084 0.11 0.76 Over/Under 45.50 3.640 37.50 60.00 Line Movement 0.118 1.622 -5.00 7.50 Home Over Momentum 2.359 1.078 0.00 5.00 Away Over Momentum 2.355 1.126 0.00 5.00 Notes: This table displays the summary statistics for all variables used in this paper for the point spread market (N=480) and over/under market (N=527) Table 3. Betting Percentages Variables % Bet on Favorite Majority Bet on Favorite Panel A: Point Spread Market Point Spread -1.084 (***) -11.000 (-2.642) (-1.562) Point Spread (2) -0.049 (*) -0.529 (-1.912) (-1.239) Road Favorite 8.102 (***) 94.053 (***) (9.260) (5.410) # of Bettors 0.133 (*) 3.285 (***) (1.909) (2.924) Spread Movement 0.774 (***) 9.233 (**) (3.624) (2.357) Favorite Cover Momentum 0.353 -1.536 (1.164) (-0.256) Underdog Cover Momentum -1.802 (***) -17.069 (***) (-5.888) (-2.948) Observations 480 480 R-squared 0.292 0.121 Variables % Bet on Over Majority Bet on Over Panel B: Over/Under Market Over/Under 1.178 (***) 19.856 (***) (12.396) (7.772) # of Bettors 0.013 -1.277 (0.117) (-0.699) Line Movement 0.282 4.589 (1.390) (1.002) Home Over Momentum 0.704 (**) 5.407 (2.558) (0.837) Away Over Momentum 0.339 4.696 (1.249) (0.813) Observations 527 527 R-squared 0.281 0.162 Notes: This table presents the regression results for bettor preference for the point spread market in Panel A and the over/under market in Panel B. The first column examines the percent of bettors placing a bet on the favorite (Panel A) or the over (Panel B) using OLS regressions, while the second column uses a probit model where the dependent variable is a dummy variable that is equal to 1 if the majority bets on the favorite (Panel A) or over (Panel B). Robust standard errors are employed, and T-statistics are listed below the coefficients which have been multiplied by 100 for ease of reading. Pseudo R-squared is shown on probit models. Significance is shown at the 10% (*), 5% (**), and 1% (***) levels. Table 4. Simulation Based upon Public Betting Percentage Percentage Range Wins Losses Win % Panel A: Percent Bet in Point Spread Market 50% - 55% 72 87 45.28% 55% - 60% 59 82 41.84% 60%+ 112 101 52.58% Percentage Range Wins Losses Win % Panel B: Percent Bet in Over/Under Market 50% - 55% 85 94 47.49% 55% - 60% 68 96 41.46% 60%+ 90 84 51.72% Earnings Earnings for Percentage Range betting with contrarian the public bettors Panel A: Percent Bet in Point Spread Market 50% - 55% $ -2,370 $ 780 55% - 60% $ -3,120 $ 1,710 60%+ $90 $ -2,220 Earnings Earnings for Percentage Range betting with contrarian the public bettors Panel B: Percent Bet in Over/Under Market 50% - 55% $ -1,840 $50 55% - 60% $ -3,760 $2,120 60%+ $ -240 $-1,500 Notes: This table presents the results of how bettors perform in the point spread market in Panel A and over/under market in Panel B based on their betting percentages. The results are shown when 50% to 55%, 55% to 60%, and more than 60% bet on one side of the wager. Two betting strategies are used to examine if betting with or against the public can provide economic profits after accounting for the vigorish based on betting $110 per game.
|Printer friendly Cite/link Email Feedback|
|Author:||Shank, Corey A.|
|Publication:||International Journal of Sport Finance|
|Date:||Feb 1, 2019|
|Previous Article:||NCAA College Basketball Television Viewership: Does Preference for Outcome Uncertainty Change Throughout the Season?|
|Next Article:||Factors that Shape the Demand for International Football Games Across Different Age Groups.|