I have been keeping a disinterested eye on the Indian Premier League. I should admit upfront that the format doesn’t excite me much, but I guess that’s just old age (as admitted earlier). The cricketer I like the most - Sachin Tendulkar - hasn’t played a single game, and that has contributed to my continued lack of interest as well. Not that I want him to play in this format (or perhaps even the ODIs) in the first place, but that’s his choice.
Had I been interested in the IPL, I would had a conflict of interest in deciding which team to support. Mumbai Indians would have been a strong contender, given my admiration devotion to Tendulkar (and having spent 4 years of my life there), but then Rajasthan Royals would also stake a claim, given that I belong to that State.
You can find good coverage of IPL elsewhere, so I would spare you an overview. After reading GreatBong’s take on the series so far - and seeing him diss Agarkar very strongly - I decided to take a look at the stats coming out of IPL. It turned out that Agarkar had done quite well till date, and while that is intriguing enough, something else interested me more: the nature of the statistics on offer.
Batting
The batting stats for IPL, for example, have not changed their parameters at all. First of all, they should stop reporting the 100s and 50s. In Test Cricket, the number of centuries (and other derived data like centuries per game etc.) is a great indicator of how good the batsman is. If you make a list of batsmen with most Test centuries, you’ll come up with a defensible list of the best batsmen to have played Test Cricket. Does the same apply to ODIs? Not in the same measure, actually. You would be better off counting the number of 50+ scores (and some other stats like strike rate etc.), or risk missing out on batsmen like Michael Bevan. The number of centuries would still give you a good list, but less accurate than for Test Cricket. Counting only the 100s in ODIs, like counting only the 200s or 300s in Test Cricket, is likely to miss out on a number of great ODI batsmen.
In my opinion, counting the number of centuries in the T-20 format is essentially meaningless. Firstly, there usually isn’t enough time to hit a 100. A team score of 200 is considered pretty good in T-20. By that measure, a good batsman would need to play about 50% of the deliveries bowled to get a 100. That’s a very difficult thing to do in any format. In contrast, a good batsmen only needs to play about 33% of the deliveries to get a 100 in ODIs (the figure was higher till people started playing ODIs differently from Tests). If you apply this standard to T-20s, the number to watch out for is 67. Lets round it off to 70 and start counting the number of times a batsmen crosses that figure in T-20s. The other figure, naturally, would then be the number of 35+ scores that a batsman managed to put up on the scoreboard. As you can see, I have derived these numbers rather crudely. Skilled statisticians would do better, but the magic number is certainly going to be well below 100. Secondly, doesn’t sheer luck play a much larger role in getting a 100 in T-20s than it does in ODIs (leave alone Tests)?
Also, lets have the number (and percentage) of dot balls played by a batsman. He might have a good overall strike rate, but dot balls create a lot of pressure and are worth a lot more in T-20s than in any other form of Cricket.
It might also be useful to look at the structure of a batsman’s innings (what fraction of his runs are scored in ones and twos etc.). This is already done partially by giving the number of 4s and 6s hit by a batsman, and one can easily get the per inning figures. However, it would perhaps be a good idea to evolve these measures and make them a standard feature of T-20 stats.
Bowling
Take a look at the Bowling stats so far. See those last two columns with figures for 4 or 5 wickets/inning? Notice anything? With a third of the tournament over, only one bowler has managed to take 4 wickets in an inning. Of what use is such data? With a maximum of 4 overs to be bowled by every bowler, wouldn’t it make much more sense to look at bowlers that took 2 or more wickets in an inning? This isn’t nitpicking. The reason for this is the same as the one to look at 4 wickets/inning in ODIs as opposed to 5 wickets/inning in Tests (personally, I would prefer to look at 3 wickets/inning in ODIs).
Other bowling stats that would probably be useful: number of 4s and 6s (per over) conceded by a bowler, and the number of dot balls bowled per over.
There has been one notable change in the bowling stats here: the number of maiden overs bowled by a bowler. I think that’s something important to look at in the T-20 context. I wonder though, if they could come up with half-maiden-over (three dot balls in a row, kind of like a hat-trick) data as well.
All-rounders
There is no real data as of now about all-rounders in T-20s. However, I am sure this is something that needs to be evolved. The value of an all-rounder to the team is much more than in ODIs or Tests.
But why?
In ODIs and Tests, statistics are essentially for fans to obsess over. Australia can’t pick Tendulkar in their national side, no matter how much they like him. In IPL though, franchises can pick any player provided they are willing to pay the right price for him (the exception being the Icon Player nonsense, which should be done away with). Indeed, this is something that is going to happen when the next edition of IPL is played. The franchises are going to need a lot of data to make their decisions, and I think there is a good business opportunity here for companies like Cricinfo which already have a sound statistical framework to use.
The fans, too, deserve to look at something better than tall columns filled with zeros.
Your thoughts? Any statisticians reading this blog?
