Academia.eduAcademia.edu

Figure 6 – uploaded by Deniz Igan

See full PDFdownloadDownload figure
Notes: The regressions are estimated using probit. Difference in black starters is calculated as the number of black starters on the home team minus the number of black starters on the visiting team. Difference in popularity: televised games is calculated as the difference between the number of nationally televised games of the home team during the season and the corresponding value for the visiting team. Difference in popularity: attendance is the average home attendance of the home team during the season minus the corresponding value for the visiting team. Difference in all-star players is calculated as the difference between the number players selected for the all-star team in the season for the home team and the corresponding value for the visiting team. Difference in foreign players is calculated as the difference between the number of foreign players on the roster of the home team and the corresponding value for the visiting team. Robust standard errors are in square brackets. ™ Statistical significance at 5% level. “ Statistical significance at 1% level. Robustness: alternate explanations.

Table 6 Notes: The regressions are estimated using probit. Difference in black starters is calculated as the number of black starters on the home team minus the number of black starters on the visiting team. Difference in popularity: televised games is calculated as the difference between the number of nationally televised games of the home team during the season and the corresponding value for the visiting team. Difference in popularity: attendance is the average home attendance of the home team during the season minus the corresponding value for the visiting team. Difference in all-star players is calculated as the difference between the number players selected for the all-star team in the season for the home team and the corresponding value for the visiting team. Difference in foreign players is calculated as the difference between the number of foreign players on the roster of the home team and the corresponding value for the visiting team. Robust standard errors are in square brackets. ™ Statistical significance at 5% level. “ Statistical significance at 1% level. Robustness: alternate explanations.

Related Figures (10)

Notes: Point spread is the quoted spread on a game as of the closing time for bets, expressed from the home team’s perspective. Realized margin is the actual difference between the home team score and the visiting team score at the end of the game. Probability of beating the spread is a dummy that is 1 if a bet on the home team wins. Black starters is the number of black players in the starting line-up. Black players on the roster is the number of black players on the team roster. Black minutes is the proportion of minutes played by black players to the total minutes in the game, calculated over the past five games the team has played. These measures of blackness of a team refer to the home team. Difference in black starters is calculated as the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes) of the home team minus the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes) of the visiting team. Summary statistics at team-game level. Table 1
Notes: Point spread is the quoted spread on a game as of the closing time for bets, expressed from the home team’s perspective. Realized margin is the actual difference between the home team score and the visiting team score at the end of the game. Probability of beating the spread is a dummy that is 1 if a bet on the home team wins. Black starters is the number of black players in the starting line-up. Black players on the roster is the number of black players on the team roster. Black minutes is the proportion of minutes played by black players to the total minutes in the game, calculated over the past five games the team has played. These measures of blackness of a team refer to the home team. Difference in black starters is calculated as the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes) of the home team minus the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes) of the visiting team. Summary statistics at team-game level. Table 1
Notes: The dependent variable in the upper panel is the realized margin in the game, computed as the home team score minus the visiting team score. The dependent variable in the lower panel is the probability of winning, which is a dummy that is 1 if the home team won the game. The regressions are estimated using ordinary least squares for the realized margin, and using probit for the probability of winning. Difference in black starters is calculated as the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes over the past five games) of the home team minus the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes over the past five games) of the visiting team. Robust standard errors are in square brackets. “ Statistical significance at 1% level. Winning the game. Table 2
Notes: The dependent variable in the upper panel is the realized margin in the game, computed as the home team score minus the visiting team score. The dependent variable in the lower panel is the probability of winning, which is a dummy that is 1 if the home team won the game. The regressions are estimated using ordinary least squares for the realized margin, and using probit for the probability of winning. Difference in black starters is calculated as the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes over the past five games) of the home team minus the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes over the past five games) of the visiting team. Robust standard errors are in square brackets. “ Statistical significance at 1% level. Winning the game. Table 2
Notes: The dependent variable in the upper panel is the point spread quoted on the game, expressed from the home team’s perspective. The dependent variable in the middle panel is the difference between the realized margin (the final score of the home team minus the final score of the visiting team) and the point spread. The dependent variable in the lower panel is the probability of beating the spread, which is a dummy that is 1 if a bet on the home team wins. The regressions are estimated using ordinary least squares for the point spread and the difference between the realized margin and the spread, and using probit for the probability of beating the spread. Difference in black starters is calculated as the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes over the past five games) of the home team minus the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes over the past five games) of the visiting team. Robust standard errors are in square brackets. * Statistical significance at 10% level. i a mg Pe Beating the spread. Table 3
Notes: The dependent variable in the upper panel is the point spread quoted on the game, expressed from the home team’s perspective. The dependent variable in the middle panel is the difference between the realized margin (the final score of the home team minus the final score of the visiting team) and the point spread. The dependent variable in the lower panel is the probability of beating the spread, which is a dummy that is 1 if a bet on the home team wins. The regressions are estimated using ordinary least squares for the point spread and the difference between the realized margin and the spread, and using probit for the probability of beating the spread. Difference in black starters is calculated as the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes over the past five games) of the home team minus the number of black players in the starting line-up (number of black players on the roster, proportion of black minutes over the past five games) of the visiting team. Robust standard errors are in square brackets. * Statistical significance at 10% level. i a mg Pe Beating the spread. Table 3
Robustness: referees. Notes: Ihe regressions are estimated using probit. Difference In Diack starters 1s calculated as the number of Dlack starters on the home team minus the number of black starters on the visiting team. The race composition of referees are taken into account by splitting the sample by the proportion of blac! referees in the 3-person crew. Alternatively, the proportion of white referees is included as a control variable. Robust standard errors are in square brackets * Statistical significance at 10% level. * Statistical significance at 5% level. “” Statistical significance at 1% level. Table 4a
Robustness: referees. Notes: Ihe regressions are estimated using probit. Difference In Diack starters 1s calculated as the number of Dlack starters on the home team minus the number of black starters on the visiting team. The race composition of referees are taken into account by splitting the sample by the proportion of blac! referees in the 3-person crew. Alternatively, the proportion of white referees is included as a control variable. Robust standard errors are in square brackets * Statistical significance at 10% level. * Statistical significance at 5% level. “” Statistical significance at 1% level. Table 4a
Notes: The regressions are estimated using probit. Difference in black starters is calculated as the number of black starters on the home team minus the number of black starters on the visiting team. Difference in records is calculated as the difference between the number of wins the home team had ina particular season prior to the game under consideration and the corresponding number for the visiting team. Difference in recent performance is calculated as the difference between the number of wins the home team had in a particular season over the five previous games before the game under consideration and the corresponding number for the visiting team. Margin on the teams’ last match is computed as the difference between home team’s score and the visiting team’s score obtained the last time the two teams played against each other (irrespective of the location and expressed from the home team’s perspective). Spread on the teams’ last match is the point spread quoted on the last game the two teams faced each other (irrespective of the location and expressed from the home team’s perspective). Note that the margin/spread is equal to the margin/spread from the last match-up in the previous season when the game under consideration is the first time the two teams face each other in a given season. Robust standard errors are in square brackets. ” Statistical significance at 5% level. YY Ct et aa olan t sannane to baal Robustness: history of teams. Table 4b
Notes: The regressions are estimated using probit. Difference in black starters is calculated as the number of black starters on the home team minus the number of black starters on the visiting team. Difference in records is calculated as the difference between the number of wins the home team had ina particular season prior to the game under consideration and the corresponding number for the visiting team. Difference in recent performance is calculated as the difference between the number of wins the home team had in a particular season over the five previous games before the game under consideration and the corresponding number for the visiting team. Margin on the teams’ last match is computed as the difference between home team’s score and the visiting team’s score obtained the last time the two teams played against each other (irrespective of the location and expressed from the home team’s perspective). Spread on the teams’ last match is the point spread quoted on the last game the two teams faced each other (irrespective of the location and expressed from the home team’s perspective). Note that the margin/spread is equal to the margin/spread from the last match-up in the previous season when the game under consideration is the first time the two teams face each other in a given season. Robust standard errors are in square brackets. ” Statistical significance at 5% level. YY Ct et aa olan t sannane to baal Robustness: history of teams. Table 4b
Fig. 1. Percent of money bet on the home team and the difference in black starters. Notes: The actual values of the difference in black starters are discrete. The line shown is continuous because it is constructed using locally weighted scatterplot smoothing (LOWESS). See Cleveland (1979) for details on the LOWESS methodology. Notes: The dependent variable is the difference between the closing and opening values of the spread on the game, showing how much the point spread moves from the start of betting until all bets close. The regressions are estimated using ordinary least squares. Difference in black starters is calculated as the number of black starters on the home team minus the number of black starters on the visiting team. Robust standard errors are in square brackets. * Statistical significance at 5% level.
Fig. 1. Percent of money bet on the home team and the difference in black starters. Notes: The actual values of the difference in black starters are discrete. The line shown is continuous because it is constructed using locally weighted scatterplot smoothing (LOWESS). See Cleveland (1979) for details on the LOWESS methodology. Notes: The dependent variable is the difference between the closing and opening values of the spread on the game, showing how much the point spread moves from the start of betting until all bets close. The regressions are estimated using ordinary least squares. Difference in black starters is calculated as the number of black starters on the home team minus the number of black starters on the visiting team. Robust standard errors are in square brackets. * Statistical significance at 5% level.
Moving the point spread. Table 5a
Moving the point spread. Table 5a
Notes: The dependent variable is the money bet on the home team, expressed as a percentage of the total bets placed on the game. The regressions are estimated using ordinary least squares. Difference in black starters is calculated as the number of black starters on the home team minus the number of black starters on the visiting team. Difference in recent performance is calculated as the difference between the number of wins the home team had ina particular season over the previous five games before the game under consideration and the number of corresponding wins for the visiting team. Difference in popularity: televised games is calculated as the difference between the number of nationally televised games of the home team during the season minus the corresponding value for the visiting team. Difference in all-star players is calculated as the difference between the number players selected for the all-star team in the season for the home team minus the corresponding value for the visiting team. Robust standard errors are in square brackets. * Statistical significance at 10% level. ™* Chatictical cionjificance at 1% level Bias in bets. Table 5b
Notes: The dependent variable is the money bet on the home team, expressed as a percentage of the total bets placed on the game. The regressions are estimated using ordinary least squares. Difference in black starters is calculated as the number of black starters on the home team minus the number of black starters on the visiting team. Difference in recent performance is calculated as the difference between the number of wins the home team had ina particular season over the previous five games before the game under consideration and the number of corresponding wins for the visiting team. Difference in popularity: televised games is calculated as the difference between the number of nationally televised games of the home team during the season minus the corresponding value for the visiting team. Difference in all-star players is calculated as the difference between the number players selected for the all-star team in the season for the home team minus the corresponding value for the visiting team. Robust standard errors are in square brackets. * Statistical significance at 10% level. ™* Chatictical cionjificance at 1% level Bias in bets. Table 5b
Chances of winning with simple strategies. Notes: The table shows the outcome of bets placed on a team when it has X (taking on values of 1, 2, 3, or 4) more white/black starters than the opposing team, as defined by the strategy. Bets show the number of games that satisfy the condition and the strategy would require an $11 bet being placed on the team. Win % is the proportion of bets that the betted-on team would beat the spread and the bettor would receive $21. Return % denotes the return on the betting strategy, computed as the total money earned on the bets as a proportion of the money spent on placing the bets. Bets are expressed in units; win % and return % are in percent terms. The first row reports the total number of bets and the simple average for the wins and returns over all the seasons in the sample. The last row reports the total number of bets and the simple average for the wins and returns for the strategy considered as a whole. “ The win % (return %) is significantly higher than 52.4% (0.0%), i.e., the win percentage required to break even, according to a one sample t-test at the 1% level. Table 6
Chances of winning with simple strategies. Notes: The table shows the outcome of bets placed on a team when it has X (taking on values of 1, 2, 3, or 4) more white/black starters than the opposing team, as defined by the strategy. Bets show the number of games that satisfy the condition and the strategy would require an $11 bet being placed on the team. Win % is the proportion of bets that the betted-on team would beat the spread and the bettor would receive $21. Return % denotes the return on the betting strategy, computed as the total money earned on the bets as a proportion of the money spent on placing the bets. Bets are expressed in units; win % and return % are in percent terms. The first row reports the total number of bets and the simple average for the wins and returns over all the seasons in the sample. The last row reports the total number of bets and the simple average for the wins and returns for the strategy considered as a whole. “ The win % (return %) is significantly higher than 52.4% (0.0%), i.e., the win percentage required to break even, according to a one sample t-test at the 1% level. Table 6
Notes: The table shows the outcome of bets placed on the home team only when the home team has X (taking on values of 1, 2, 3, or 4) more white starter: than the visiting team. Bets show the number of games that satisfy the condition in a given season and the strategy would require an $11 bet being placec on the home team. Win % is the proportion of bets that the home team would beat the spread and the bettor would receive $21. Return % denotes the return on the betting strategy, computed as the total money earned on the bets as a proportion of the money spent on placing the bets. Bets are expressec in units; win % and return % are in percent terms. The row before last reports the total number of bets and the simple average for the wins and returns ove! all the seasons in the sample. The last row reports the total number of bets and the simple average for the wins and returns if one bets anytime the home team has more white starters than the visiting team. “ The win % (return %) is significantly higher than 52.4% (0.0%) i.e., the win percentage required to break even, according to a one sample t-test at the 19 Chances of winning with a simple strategy: season by season.
Notes: The table shows the outcome of bets placed on the home team only when the home team has X (taking on values of 1, 2, 3, or 4) more white starter: than the visiting team. Bets show the number of games that satisfy the condition in a given season and the strategy would require an $11 bet being placec on the home team. Win % is the proportion of bets that the home team would beat the spread and the bettor would receive $21. Return % denotes the return on the betting strategy, computed as the total money earned on the bets as a proportion of the money spent on placing the bets. Bets are expressec in units; win % and return % are in percent terms. The row before last reports the total number of bets and the simple average for the wins and returns ove! all the seasons in the sample. The last row reports the total number of bets and the simple average for the wins and returns if one bets anytime the home team has more white starters than the visiting team. “ The win % (return %) is significantly higher than 52.4% (0.0%) i.e., the win percentage required to break even, according to a one sample t-test at the 19 Chances of winning with a simple strategy: season by season.
Related topics:
EconomicsApplied EconomicsSocial Science Research NetworkMarket efficiency
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved