A recent article by a former Australian cricketer got me thinking about how Agent Based Models could help understand how bowlers and batsmen behave in cricket....
To try and keep things simple I've assumed there are 4 bowlers and 4 batsmen.
There are 2 innings for each side in each game and 5 games in the series, therefore 10 innings for each bowler-batsman combination in the series.
In each innings I model, for each batsman, the following parameters:
- A baseline estimate for how likely each bowler is to get the batsman out in each innings (the baseline varies between batsmen, but is kept constant for each combination for each innings)
- Whether or not the bowler got the batsman out in this innings
- The number of times the bowler has got the batsman out in previous innings
