This paper uses the betting market for professional basketball games to address the issue of unexplained asset price volatility. A pricing model is presented which identifies two components in point spreads for professional basketball games. Both components—the market's estimate of relative team abilities and an idiosyncratic factor—are essentially unobserved, but can be identified ex post. The structure of this market enables tests of competing hypotheses about point spread variation. The tests reject the hypothesis that variation in the two components represents irrelevant noise. The hypothesis that unobserved fundamentals account for this variation is consistent with the data.