Malcolm Knox’s pretty average use of batting stats


Malcolm Knox has invented his own metric. Oh dear.

Knox starts this piece with a bang: “WARNING: the following contains material that may offend those who don’t eat cricket statistics for breakfast, or those who think Test cricket only started in the 1990s.”

Well, that list of offended parties at least makes a start. Can we add those who may not eat statistics for breakfast or any other meal, but do have a passing understanding of them?

But before I get too self-righteous, I should at least acknowledge that the point that this article is trying to make – that a test cricket batting average is a metric without much nuance – is well made. It’s just that the substitute Knox proposes doesn’t solve the problem. (As for where Ponting ranks in the batting pantheon – that’s cricket journo territory – I don’t have enough data for an opinion.)

So what is the problem, in statistical terms? Broadly, it’s the assumption that all test innings provide a batsman with an equal chance of making runs. Regardless of the strength of the opposition bowling attack and your batting teammates, the state of the pitch, the state of the ball, the match situation, the speed of the outfield, the nearness of the boundaries, the prevailing rules of cricket, or the technological advancement of your batting equipment. In short, sampling bias.

So, Ponting got to use some pretty high-tech bats, but not for as long as Michael Clarke will in his career (but Clarke gets some second chances to be in – or out – via the DRS). Steve Waugh got to be part of a team that crushed everyone, and he never had to face Shane Warne; Kim Hughes played in a regularly crushed team and never had to face Bruce Yardley. Ian Chappell played in a team that got stronger each year, but faced the West Indies when helmets were “in development”. The rest of the world played test cricket in the absence of the world’s best players during World Series Cricket. And the further you go back, you get a mixture of all of the above, and throw in uncovered pitches, the second new ball after 200 runs rather than 80 or 85 overs, and a variety of rules on concepts as fundamental as LBW or no-balls, or as seemingly marginal as rest days or bowlers running on the pitch.

What Knox has attempted to do is factor in all of the above by measuring it against the skills displayed by teammates who faced the same conditions. This is not a bad start, but he doesn’t say how he has made his calculations. If all he has done is compare a batsman’s average to the raw figure for [team runs scored / team wickets lost], then we immediately have a problem, exemplified by the short career of fictional test player Adam Goodbat:

In his first innings Australia is skittled for 100, of which he contributes 10 – dead on the average. Luckily that test is rained out and the next match is against the Outer Hebrides on a better pitch: 5 (dec) for 500 this time, with Goodbat contributing an even 100, once again right on the team average, before rain arrives once again, this time robbing the Aussies of victory. So, by Malcolm’s measure, Goodbat should be right on average, 0% additional runs contributed, right?

Er, no. The team has lost 15 wickets for 600, an average of 40. Goodbat’s average is 55, his super-contribution a more than solid 37.5%.

In this example, Goodbat has been the beneficiary of getting the opportunity to bat on at least one good pitch, but it’s just as easy to imagine the opposite: overdosing on the local curry in the Outer Hebrides and not making the team, and therefore missing out on the easy pitch. His teammates build their batting stats, and Goodbat’s super-contribution goes down the toilet with the last of last night’s dinner… all the way down to minus 75%.

A teammate who didn’t play in the first match but made 10 in the second is also on minus 75%, but has obviously done poorly on the batman’s paradise.

But the best position to be in is to have skipped the first match, and matched the team average of 100 in the second: a super-contribution of 150%!

So, in the world of the Malcolm-metric, are matches in which Goodbat is sitting on the sidelines counted towards the new stat? I don’t know, but I’m guessing yes because the “Crunching the Numbers” figure makes vague reference to each batsman’s era, which could easily be shorthand for “I’m just going to use all add up all the runs made and wickets lost… that’ll be quicker”. I can’t really imagine a journalist going across the numbers for Ponting’s whole career (not to mention everyone else’s), carefully extracting any missed games to up the accuracy level and then not bothering to let us know how much work they’ve done.

All of which, whilst not making Malcolm’s article totally pointless, at least undermines his effort to use numbers to help his cause.

However, I’m not one to criticise without offering an alternative (well, sometimes I am, but this isn’t one of those times). But for that, you’ll have to wait for my next post, when I’ll explain why batting is like golf and why, just like a cricketer’s box, every batting performance needs a little adjustment.


Click here to like, or tell me whether this is genius or gibberish

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s