wasted in the vain pursuit of the returns to streak management for which only a very
few managers have demonstrated talent.
The work here is better than it would have been thanks to comments by Andrew Gill, Elizabeth Gustafson, Wayne Joerding, Tong Li, and James Quirk. We acknowledge the able
research assistance of Jennifer Schultz.
Finding the streaks are random in professional basketball and football, past works seek to
determine whether participants in sports betting markets treat streak as random. The empirical
evidence is mixed. Camerer ( 1989) finds no evidence of betting market belief in the so-called
"hot hand" phenomenon in professional basketball. However, Brown and
Sauer ( 1993) disagree
and Badarinathi and
Kochman ( 1994) find that betting against the hot hand in professional
football is profitable. Essentially, streaks effect betting markets but it remains to be seen whether
steals effect score differences.
If there had been managerial changes during a given year, an argument could be made that
wins and losses should be observed across manager, rather than teams. However, there only
were four changes in each of 1989 and 1990. None of the changes in 1989 involved teams
where non-random streaks were detected at the 5% level. In 1990, the Yankees had a change
of managers and non-random streaks. But Stump Merrill managed 113 of their games (70%)
to Bucky Dent's 49 (30%); we attribute the outcome to Merrill.
It is entirely possible that we are understating the ability to manage streaks. Suppose that
a team with a +3 streak meets a team with a +4 steak. The +3 team wins, but the other team
(after its loss) proceeds to win two more games. Isn't this really a +6 streak, interrupted by
having run into a better streak manager? The time required in order to define streaks in this way
proved prohibitive, especially since evidence of non-randomness seems so apparent.
Camerer ( 1989) includes the longest of the streaks in a given game as the observation; if
team i is +6 while team j is -3, the observation is a win streak of +6. "To use the same
observation as evidence for streaks of +6 and -3 would cause statistical dependence across streak
categoris" (p. 1258). The dependence claim is obvious (one team's win is another team's loss)
if looking across all teams at streaks. However, we examine a given team's streak history, so that
such dependence is beside the point. Brown and
Sauer ( 1993) define a streak as a run of games
where the team beat the spread or a run where the team failed to beat the spread. Their streak
variable measures whether the home team is on a short (2-3) or long (4+) streak and, similarly,
for the visitor.
Questia, a part of Gale, Cengage Learning. www.questia.com
Book title: Sports Economics:Current Research.
Contributors: John Fizel - Editor, Elizabeth Gustafson - Editor, Lawrence Hadley - Editor.
Publisher: Praeger Publishers.
Place of publication: Westport, CT.
Publication year: 1999.
Page number: 133.
This material is protected by copyright and, with the exception of fair use, may
not be further copied, distributed or transmitted in any form or by any means.