Player 1 v/s Computer

This isnt a Playstation game or the latest version of International Cricket Captains (ever tried this computer game .. it's amazing !!!). Just an extension of Devlin's piece on "A Game of Numbers". Devlin in this treatise has described how Oakland Athletics won the Western division of baseball in 2000, 2002 and 2003 with a team which had no 'A' cat super-stars ... a team which was just a summation of individuals with specialised skills.

Here's how they did it -
a) The initial premise was that unlike football (american style), baseball didn't require excessive teamwork. The pitcher pitches depending on his individual skills, the batter plays using his own skills and the out-fielder fields the ball unguided by other individuals. This is somewhat similar to what happens in a game of cricket.

b) Over time some 11 million batter-pitcher confrontations have occurred in 150,000 games. Thats a world of data to analyse. Coupling this data with random elements of pitches, weather, toss etc. you can create patterns w.r.t. which position to send each batsman, batsman can play a spinner or a faster bowler better, condition training schedules accordingly, individual tactics, patterns on how a bowler balls his first two balls and the last two balls (i've always noticed that the first two balls of any pace bowler are never the slower ball) etc.

Reading the article I came across some really wonderful notes, worth putting down here -

1. Batting averages are now generally regarded as a poor guide to performance - not least because they do not distinguish between a single, double, triple, or boundary; or how many players scored runs on a mis-hit or after being dropped or narrowly escaped being out.

Also, averages are rather misleading. Analyse this -
In the 1997 season, Jacques Kallis played 14 matches and scored 560 runs while Rahul Dravid played 35 matches to score 1350 runs. Assuming both got out on all occassions, the averages for the two batsmen are - Kallis - 40.00; Dravid - 38.57 ... So Kallis was better than Dravid for the 1997 season.

In 1998, Kallis played in 30 matches to score 1000 runs (at an avg of 33.33) while Dravid featured in 16 matches to score 500 runs (at an avg of 31.25). For the second time, Kallis beats Dravid with a better average.

Now, just combine the numbers. So,
Kallis ........ Matches = 44 (14+ 30); Runs = 1560 (560+1000)
Dravid ..... Matches = 51 (35+16); Runs = 1850 (1350+500)

Relook the averages .... Jacques Kallis is now at 35.45 runs/match while Rahul Dravid has an average of 36.27 runs/match. Hence Dravid is a better pick (and averages is not as powerful a statistic as is expected to be)

2. Extra credits - Peculiar to baseball, the batter is scored as per 'runs'. So you come out to bat and there is also a player standing on base 1. And you hit a 'home run' ... both you and the man at base 1 collect one run each. Howzzat !!! You do the fantastic work of hitting the home run and the free-rider also collects the same points for just being on base 1.

In cricket ... similar parallels in getting out on a run-out, mis-field off bowlers, missed chances etc. are never considered in the algorithms.

The entire attempt to bring the games down to numeric equations has many an advantage. The approach pins the relative importance of the contributions that players with different talents, can make to a team's overall performance.

Ofcourse, this doesnt mean there isn't a lot of skill involved in any sport, cricket or baseball. Nor does it mean that some players aren't better than others. Statistical tools are fantastic aids in guiding team fortunes and in predicting trends which can be rectified or improvised for success.