Softball Run Expectancy using Markov Chains and Simulation
Description
Implements two methods of estimating runs scored in a softball
scenario: (1) theoretical expectation using discrete Markov chains and (2) empirical
distribution using multinomial random simulation. Scores are based on player-specific input
probabilities (out, single, double, triple, walk, and homerun). Optional inputs include probability
of attempting a steal, probability of succeeding in an attempted steal, and an indicator of whether
a player is "fast" (e.g. the player could stretch home). These probabilities may be
calculated from common player statistics that are publicly available on team's webpages.
Scores are evaluated based on a nine-player lineup and may be used to compare lineups,
evaluate base scenarios, and compare the offensive potential of individual players.
Manuscript forthcoming. See Bukiet & Harold (1997) for
implementation of discrete Markov chains.