The Cricket Abstract

Somewhat outraged, Pat writes in to note Shane Watson’s inability to make really big scores. And, like any good person with an opinion, he has some data to back him up.

Looking at players that have scored 1000 runs over the last ten years and at least one century, Watson’s conversion rate ranks among the very bottom and is the lowest of any Australian.

I recently wrote a paper on the effect of batting during the evening in one day cricket, I’ll save you the pain and skip to the conclusion:

This paper set out to measure the effect of batting during the evening in cricket. Both of our models provided strong evidence that batting during the evening had a negative effect on run scoring. The estimated effects do appear to be relatively small, at around 0.2 runs an over. However, over the course of a 50 over match these effects cumulate to a total of a loss of around 9.25-14.65 runs. In our complete data set, there were 205 matches where the team batting second in a day-night match lost and of these 15 were within a nine run margin. This indicates that batting during the evening could have had a meaningful effect on the outcome of around 1.6% of the matches in our dataset.

First off, yikes

Again CricInfo presents an interesting idea but with somewhat less than awesome analysis. They’re attempting to measure ODI wins when the winning team have a bunch of resource to spare and the author makes up some dodgey measure of resources remaining. Particularly egregious is the assumption that overs all have the same value.

People, the Duckworth Lewis method was designed to measure resources remaining.

I was able to run an analysis measuring resources remaining at the end of the innings for teams who won batting second. It’s pretty straightforward as you just look at how many wickets and overs they have left when they surpass the team who batted firsts score. It’s a lot more complicated for team who batted first and won as they, by definition, use up all their resources (they’re either bowled out or bat the 50 overs). You would need to make up a new measure for teams batting first, something that might be done later, but not now.

Also, my data set isn’t comprehensive and only really covers 1998-2007

So here it is:

Biggest wins by teams who batted second and won

Continuing in their habit of being interesting but providing slightly annoying analysis, there was an article on cricinfo a few months back dealing with the effect of the doosra on bowling performance.

They give us this data based on test matches since 2002:

Now this is interesting because off spinners have developed the doosra while left arm othodox spinners haven’t so by looking at this data we might get some idea of how using the doosra has affected wicket taking. Unfortunately the analysis kind of stops at the accumulation of the data (even though they do go into looking at it in a bit more detail).

But people, we’ve got four groups here. I’m feeling a difference in differences coming on.

I’m going to wing this a bit, so please correct me if I’m wrong. We need to mix the normal DID up a bit as we’re dealing with mirror images here so we want to compare bowlers who are spinning in to the batter with those who are spinning away from the batter.

So first we’ll do bowlers spinning in to the batter:

(Offspinners bowling at Right-handers) – (Left-arm orthodox bowling at Left-handers) = x

then we’ll do bowlers spinning away from the batter:

(Offspinners bowling at left-handers) – (Left-arm orthodox bowling at right-handers) = y

Then we’ll do x-y and we should, maybe, get an isolated effect of the doosra?

Running the calculations gives an isolated effect of -12.86. That is, the doosra has allowed bowlers to take a wicket for 12.86 less runs, on average.

Finally, I have figured out a way to get the AFL player data into a somewhat transferable package!

This file contains data for each AFL player in each match between 2000 and 2008. The data is in a Microsoft Access database which is then rared.

The data we have is:

• Round or finals series
• Round number
• Home team
• Away team
• Score (broken into goals, behinds and total)
• Match result
• Venue
• Attendance
• Date
• Time
• Season
• Goals scored by each individual in each match