Quote:
Originally Posted by ashley12
I was talking to a poker friend of mine about the IPL in general, and how there is certainly a lot of room for a "Moneyball" type of approach the auction rather then just blindly picking players like Maxwell for a casual million.
The actual discussion was what metrics would be the "On-Base %" in how important it is in the judging of players.
Strike-Rate sounds like the most important aspect, and probably is, but sure average, 4/6%, number of balls per innings ave, has to play into this magical formula to devise the worth of a player.
Its pretty obvious that the teams have in the past made a lot of mistakes (Royal Challengers Bangalore season 1 the obvious choice with a top order of Dravid and Kallis among others) but yeah, thoughts?
To analyze batting I would start with 2 basic metrics. One based on average, and the other on strike rate.
Balls per dismissal: How many balls faced per dismissal. Now this has to account for not-outs (i.e. batting position). Especially in T20 with so many not-outs.
Runs per ball: i.e. Strike rate.
Balls per dismissal is surely important even in T20. Especially for top order batsmen, because wickets are limited.
Six rate/Four rate: Number of sixes/fours per 100 balls. Obviously hugely important in T20s.
Running between wickets: This is interesting. With play-by-play data you could built metrics to measure how often batsmen "get the second run" etc. Where they pick up extra runs here and there based on good running. It's obviously complicated by having a running partner.
Over adjusted stats: With so much data available, it makes sense to look at time phased stats. Stats in the powerplay or death overs can't be fairly compared to middle over stats. Although in T20 I think in general it's best to bat aggressive all the way through (not "save" yourself until the death). One needs to weigh the value of the above stats based on the situation.
Finally I think you have to adjust for situation/leverage. Your stats should be normalized based on required run rate (second innings). I'm not sure how to normalize first innnings stats. But I guess something based on average score at that ground, or overall average score.