At Globogym we’re better than you. And we know it!
Dodgeball – 2005
In last week’s blog, the data showed how poorly some overseas players performed in First Class cricket compared with their Test performances.
Looking at overseas players, surprisingly they perform 21% worse in Division 1 than their Test average. Contrast that with England players who do 28% better. Two examples jump out: Pujara scoring 172 runs at 14, Kane Williamson scoring 260 runs at 26. How can we explain those scores?
As there have been only 20 non-England Test players in Division 1 over the last three years, the sample size is too small for meaningful analysis. To get more insight, I’ve combined Division 1 and Division 2, which increases the sample size to 331 completed innings. I then found 3 factors which influence performance:
- SA / NZ / Australian players outperform other nations (probably because these are the countries with conditions most similar to those in England).
- Test players will average more in Division 2 than Division 1.
- Top order (1-3) batsmen are most affected by English conditions (this makes sense – they will face lengthy spells against the best County bowlers with the ball swinging and seaming more than they are used to). Middle order players (numbers 4-7) are unaffected, while tailenders get a boost to their average.
I created a model to quantify this behaviour, combining these factors. The best fit to the data is as follows:
- SANZAR +10%, others -10%
- Top order -25%, Middle order +3%, Lower order +25%
- Division 2 +10%
Applying this makes Pujara’s performance less of an outlier, and more a function of being a number 3, and therefore the wrong type of overseas batsman to go for. Using my model, his expected average in D1 is just 36, and while he underperformed this, it’s no longer an outlier. Similarly, Azhar Ali (Test Avg 48) would be expected to average 33, and averaged 34.
But – the current iteration of the model has arbitrary cut-offs (why should a number 4 outscore a number 3 by 25%?) and the above table has a high standard deviation. I’ll enhance it once it can be tested against 2019 data.
What the current model can do is make predictions:
Poor 2019 Overseas Player selections
Azhar Ali will be playing for Somerset next season. He’ll be 34 by then, and will be expected to average 30. I hope they aren’t paying him too much. Next season could be the one where Somerset’s batting frailty bites.
Bancroft at Durham and Joe Burns at Lancashire should struggle at the top of the order.
Top 2019 Overseas Player picks
1. S.Marsh better hope Glamorgan bat him below 3 – he could do well if he avoids the new ball.
2. Temba Bavuma isn’t the strongest Test batsman, but as a 28 year old he’ll be at or near his peak, and Division 2 cricket with Northamptonshire should suit him. It helps he doesn’t start until 14th May.
3. Bowlers! Abbas, Worrall, and Siddle should be far more valuable than top order batsmen. That said, I’ve not done the analysis of bowlers yet. Watch this space.
Red Ball Data 