REVISITING THE “HOT HAND” HYPOTHESIS IN THE NBA BETTING MARKET USING ACTUAL SPORTSBOOK BETTING PERCENTAGES ON FAVORITES AND UNDERDOGS

Abstract

The “hot hand” hypothesis was first investigated in sports betting markets by Camerer (1989) and Brown and Sauer (1993), who examined if professional basketball teams truly could become “hot”, implying a change in their actual skill level, and if the betting market believes teams become “hot” and over bet the teams on winning streaks.   Both assumed that book makers operated a balanced book.  Recent evidence suggests that book makers do not set point spreads to balance betting on either side of games.  Book makers may price as a forecast or shade point spreads to exploit known biases.  The “hot hand” could exist, but closing point spreads may not reflect this bias due to an unbalanced book.   Using a 6 season sample of NBA betting market data, we show wagering against the “hot hand” does not win more than implied by efficiency.  However, OLS and two-stage least squares regression models show that bettors believe in the hot hand, as teams on streaks attract a significantly higher number of bets.  This illustrates that the public believes in the hot hand, reflecting an actual behavioral bias.  This bias exists even though the closing price serves as an optimal and unbiased forecast of outcomes.