“Hot Hand” in the National Basketball Association Point Spread Betting Market: A 34-Year Analysis

Benjamin Waggoner,Brian Soebbing,Daniel Wines,Jean Martinez,Chad Seifried

doi:10.3390/ijfs2040359

Benjamin Waggoner, Brian Soebbing + Show 3 more

Open Access

https://doi.org/10.3390/ijfs2040359

Copy DOI

Abstract

Several articles have looked at factors that affect the adjustments of point spreads, based on hot hands or streaks, for smaller durations of time. This study examines these effects for 34 regular seasons in the National Basketball Association (NBA). Estimating a Seemingly Unrelated Regression model using all 34 seasons, all streaks significantly impacted point spreads and difference in actual points. When estimating each season individually, differences emerged particularly examining winning and losing streaks of six games or more. The results indicate both the presence of momentum effects and the gambler’s fallacy.

Highlights

Sports betting markets have been compared to simple financial markets, which allowed researchers to examine financial phenomena difficult to observe in other markets [1]
The “R2” reported in both models is consistent with previous research where Equation 1 explains more of the observed variation in the point spreads than in the actual difference in points
Team momentum effects are a popular area of analysis with particular emphasis looking at National Basketball Association (NBA) games

Summary

Introduction

Sports betting markets have been compared to simple financial markets, which allowed researchers to examine financial phenomena difficult to observe in other markets [1]. Early research regarding sports betting markets focused on the efficiency of these markets (see Sauer [2]) through the rationality between the opening and closing betting lines [3]. Sauer’s [4] review of the sports betting markets outlined three different types of market efficiency: weak, semi-strong, and strong. Within these forms of market efficiency, numerous other studies looked at biases such as the favorite/longshot bias [5,6,7,8], reverse favorite/longshot bias [5,9,10], racial bias [11], and sentiment bias [12,13,14,15,16]

Objectives

Methods

Results

Conclusion