Abstract:
|
The theory of record values originally focused on tracking outstandingly large (or small) observations in sequences of independent and identically distributed (i.i.d) random variables. A broad spectrum of related record models have been developed, with special emphasis on accounting for the observed superabundance of records in many real world data sequences. Perhaps the most obvious scenario in which records are established and broken at a phenomenal rate is the Olympic Games. However, major league baseball, with its penchant for keeping statistics on all aspects of the game, is another fertile source of records. In baseball, as in the Olympics, it is generally the case that records are too numerous to allow us to accept the basic record model (based on i.i.d. observations) as an appropriate description of the underlying stochastic record-generating mechanism. Preliminary analysis of some of the more highly publicized baseball record sequences (homeruns, batting averages, stolen bases, etc.) will be presented in the current paper. In some cases, evidence supports improving population models. In other cases, more complex explanations must be sought.
|