Northern Kentucky University
Estimating NBA Playoff Success Probabilities
Institution
Northern Kentucky University
Faculty Advisor/ Mentor
Joseph Nolan; David Agard
Abstract
The National Basketball Association (NBA) is a multi-million dollar industry in the United States. Success on the court can make a substantial difference in the amount of profit for each team. In 2011, the Miami Heat brought together several “All-Star” players in an attempt to win the NBA Championship. This project investigated, statistically, how NBA team success may be related to individual player successes. Team success for each year was measured by depth of playoff advancement on an ordinal scale (0 - did not make the playoffs, 1 - lost in first round, 2 - lost in second round, 3 - lost in third round, 4 - lost in fourth round (this is the runner up), 5 – Won Championship). Individual player success was measured by looking at a player’s history of All-Star appearances. For any particular year, each previous All-Star appearance of a player’s career is assigned a weight (several different weighting schemes were considered). These weighted appearances are summed to create a player’s individual contribution; the contributions of all players on a team are summed as a measure of Team All-Star Composition. Results indicated statistically significant, positive associations between playoff success and each measure of Team All-Star Composition. Binary logistic regression procedures were applied to estimate team chances for simply making the playoffs. Additionally, ordinal logistic regression procedures were applied to estimate probabilities of different levels of team playoff advancement based on their Team All-Star Composition. A comparison of the different weighting schemes also indicated that the most recent timeframe of All-Star appearances is of greatest importance.
Estimating NBA Playoff Success Probabilities
The National Basketball Association (NBA) is a multi-million dollar industry in the United States. Success on the court can make a substantial difference in the amount of profit for each team. In 2011, the Miami Heat brought together several “All-Star” players in an attempt to win the NBA Championship. This project investigated, statistically, how NBA team success may be related to individual player successes. Team success for each year was measured by depth of playoff advancement on an ordinal scale (0 - did not make the playoffs, 1 - lost in first round, 2 - lost in second round, 3 - lost in third round, 4 - lost in fourth round (this is the runner up), 5 – Won Championship). Individual player success was measured by looking at a player’s history of All-Star appearances. For any particular year, each previous All-Star appearance of a player’s career is assigned a weight (several different weighting schemes were considered). These weighted appearances are summed to create a player’s individual contribution; the contributions of all players on a team are summed as a measure of Team All-Star Composition. Results indicated statistically significant, positive associations between playoff success and each measure of Team All-Star Composition. Binary logistic regression procedures were applied to estimate team chances for simply making the playoffs. Additionally, ordinal logistic regression procedures were applied to estimate probabilities of different levels of team playoff advancement based on their Team All-Star Composition. A comparison of the different weighting schemes also indicated that the most recent timeframe of All-Star appearances is of greatest importance.