Factors affecting outcomes in test match cricket

Least squares regression is used to model the first innings performances of teams in test cricket in order to establish batting and bowling ratings, a common home advan- tage and a country effect. Logistic regression techniques are then used to model match outcomes based on a team’s first innings lead, innings duration, home advantage, bat- ting and bowling ratings and the country effect. It is shown that the factors that impact most significantly on the outcome of a match are a team’s first innings lead home team performance and innings duration. A team’s first innings lead is found to more likely shape a win rather than a draw or a loss whereas the longer the duration of the first innings the more likely a match will end in a draw. It is shown that the home team, on average, needs to establish a lead in excess of 93 runs to have a better than even chance of winning, whereas the away team needs to establish a lead in excess of 115 runs to have the same chance. There is a better than an even chance of a draw for a first innings duration in excess of 1165 minutes (or approximately 277 overs). It is also shown that the home team is more likely to win a match rather than lose or draw, which suggests that the home team has a distinct winning advantage over the away team. There is some evidence suggesting that teams gained an advantage by batting last.