Moneyball

In a great post, Vivek asks why IPL's Twenty-20 matches use the kind of statistics that were originally created for Test matches, and modified (only slightly) for One Day Internationals. He goes on to suggest a few metrics that can help assess more meaningfully a player's usefulness in T20 matchs. As I said, it's a great post on an interesting topic that will appeal to the academic, the non-playing captain, and the statistical junky in you. Go read it!

Now, let me use this opportunity to connect Vivek's post with what I have read about Moneyball: The Art of Winning an Unfair Game by Michael Lewis (of Liar's Polker fame). The book is about how

... a small group of undervalued professional baseball players and executives, many of whom had been rejected as unfit for the big leagues, who had turned themselves into one of the most successful franchises in Major League Baseball. [link]

The team in question is the Oakland Athletics which went on to win the top spot in the American League - West in 2002.

The secret of their success? A reliance on (new, unfashionable, or under-utilized) statistical measures of a player's sporting prowess that really matters to the team's performance, and using these measures to evaluate, bid for, and trade players. Clearly this approach places a premium on past performance rather than on potential. As Fabio points out, moneyball tactics are an example of evidence based management!

As cricket becomes more professionalized, with lots more players and lots more matches, and with the availability of ball-by-ball accounts and videos available in the archives, our ability to slice and dice a player's every move and extract all kinds of statistical measures will only improve. As of now, Twenty-20 is a very young format, with even experienced players having played just a few tens of games. But Vivek's larger point is valid: this format cries out for new, more meaningful statistical yardsticks.

I haven't read Moneyball, but one of the impressions I get from reading about it (Wikipedia, NYTimes review, and Failure Magazine) indicates to me that some of the unfashionable metrics used by the Oakland A's were popular among enthusiastic amateurs. There is no reason why such cricket-loving amateurs cannot come up with new and interesting ways of looking at T20 players' records.

[Aside: In any event, I don't expect IPL franchises to start using serious statistical techniques -- moneyball or otherwise -- simply because they are all too busy counting the dollars that are pouring in. ]

* * *

For what it is worth, here's my early nomination for the IPL Moneyball Team of 2008: Rajasthan Royals!

* * *

In business, I can think of at least one initiative that would qualify for use of Moneyball tactics: Adventnet, the company behind the Zoho suite of products. What is so 'moneyball' about this company? Let me give the mike to Sramana Mitra:

... [In[ India [Adventnet CEO Sridhar] Vembu's operation does not hire engineers with highflying degrees from one of the prestigious India Institutes of Technology, thereby squeezing his cost advantage.

"We hire young professionals whom others disregard," Vembu says. "We don't look at colleges, degrees or grades. Not everyone in India comes from a socio-economic background to get the opportunity to go to a top-ranking engineering school, but many are really smart regardless.

"We even go to poor high schools, and hire those kids who are bright but are not going to college due to pressure to start making money right away," Vembu continues. "They need to support their families. We train them, and in nine months, they produce at the level of college grads. Their resumes are not as marketable, but I tell you, these kids can code just as well as the rest. Often, better."

* * *

Let me round this post out by taking you back to the US. Here's the latest Freakonomics column by Steven Levitt and Stephen Dubner on how a statistician may have helped Boston Celtics in their turn around from an abysmal record of just 24 NBA wins last year to an impressive record 66 wins this year.