A Markov chain approach to baseball

Bruce Bukiet, Elliotte Rusty Harold, José Luis Palacios

Research output: Contribution to journalArticlepeer-review

47 Scopus citations


Most earlier mathematical studies of baseball required particular models for advancing runners based on a small set of offensive possibilities. Other efforts considered only teams with players of identical ability. We introduce a Markov chain method that considers teams made up of players with different abilities and which is not restricted to a given model for runner advancement. Our method is limited only by the available data and can use any reasonable deterministic model for runner advancement when sufficiently detailed data are not available. Furthermore, our approach may be adapted to include the effects of pitching and defensive ability in a straightforward way. We apply our method to find optimal batting orders, run distributions per half inning and per game, and the expected number of games a team should win. We also describe the application of our method to test whether a particular trade would benefit a team.

Original languageEnglish (US)
Pages (from-to)14-23
Number of pages10
JournalOperations Research
Issue number1
StatePublished - 1997

All Science Journal Classification (ASJC) codes

  • Computer Science Applications
  • Management Science and Operations Research


Dive into the research topics of 'A Markov chain approach to baseball'. Together they form a unique fingerprint.

Cite this