Moneyball: Selecting the Australian Test team on stats only

Moneyball was a movie based on the Oakland Athletics’ Major League Baseball team’s success in building a team of undervalued talent through purely statistical and analytical measures.

In the movie, Oakland Athletics were suffering from the departure of star players Johnny Damon, Jason Giambi and Jason Isringhausen and had a limited budget.

Similarly, fresh off a first-ever series home defeat to India, the Australian Cricket team is suffering from the suspensions of Steve Smith and David Warner, and have experienced an overall decline in talent ahead of Sri Lanka’s upcoming visit.

How we long for the days of names like Langer, Hayden, Ponting, Waugh, Waugh, Martyn, Gilchrist, Warne, Lee, Gillespie and McGrath.

Recently, the decline of the Test side, and the interesting, if not bizarre, selections of Marnus Labuschagne, Mitchell Marsh and Aaron Finch, led me to ponder how the team would look if it was based entirely on statistics from Shield and Test cricket over the past four seasons.

Marnus Labuschagne of Australia looks dejected after being dismissed by Mohammed Shami. (Photo by Ryan Pierse/Getty Images)

Recent examples like Mike Hussey, Chris Rogers and Adam Voges have shown that successful Shield careers are more likely to indicate success at the Test level, rather than picking players based on white ball ability like Aaron Finch and Mitchell Marsh.

Currently, with the way the Australian team is performing, surely picking a stats based only team for the Sri Lanka series could not be any worse.

Team selection
Five batsmen
Three fast/medium bowlers
One slow bowler
One all-rounder
One wicket-keeper

Methodology and rationale for weightings
Wanting to incorporate for both form and class, I structured the weightings for both batting and bowling as follows:
2018-19 Performances are multiplied by 0.4
2017-18 Performances are multiplied by 0.3
2016-17 Performances are multiplied by 0.2
2015-16 Performances are multiplied by 0.1

This allowed me to get a Weighted Average for each player. For example, this is how NSW Batsman Kurtis Patterson’s numbers look:

2018-19: 428 runs / 9 dismissals
2017-18: 672 runs / 18 dismissals
2016-17: 668 runs / 15 dismissals
2015-16: 737 runs / 14 dismissals

Weighted Average: ((428 x 0.4) + (672 x 0.3) + (668 x 0.2) + (737 x 0.1)) / ((9 x 0.4) + (18 x 0.3) + (15 x 0.2) + (14 x 0.1)) = 43.29

All Test performances for both batting and bowling received a 25 per cent boost, meaning 50 runs in Shield cricket is the equivalent of 40 runs in Test cricket. I included this, as it is much easier bowling to Shield players on a green top at Bellerive Oval than it is on a Mumbai road against Virat Kohli and Cheteshwar Pujara.

Similarly, you’d much rather face 125km/h non-swinging Shield bowlers than Kagiso Rabada, Dale Steyn and Vernon Philander at Centurion.

In essence, the Weighted Average shows the expected average a batsman/bowler would experience if they played a full Shield season from today.

Minimum statistical requirements
Weighted Average matches played = 5
Weighted Average runs = 250
Weighted Average wickets = 5
Weighted Average innings kept = 10

Batsmen for Sri Lanka series

Minimum statistical requirements
Weighted Average matches played = 5
Weighted Average runs = 250
Notable players excluded: WJ Pucovski, AC Voges, EJM Cowan

*Denotes suspended player
Dark background denotes player would make XI

The Moneyball method sees Marcus Harris, Usman Khawaja, Shaun Marsh and Peter Handscomb retaining their spots, with Glenn Maxwell coming in for Travis Head. Marsh and Handscomb would lose their spots in the line-up if Smith and Warner were available.

The Harris selection shows the value of a weighted system, rather than simply looking at Harris’s first-class average of 35.57. This average has been deflated through his lack of relative success in his teenage years and early twenties. He is a far more accomplished player now, and makes the team on the back of bludgeoning 759 runs already in 2018-19 for state and country.

Marcus Harris of Australia. (Photo by Michael Dodge/Getty Images)

Interestingly, Moneyball argues for the inclusion of Maxwell, who it seems has developed an unfair reputation as an Asian specialist in the eyes of selectors, with all seven of his Test matches coming in either Bangladesh, India or the United Arab Emirates.

You would think that a century in India, over 700 runs at 50.5 in the 2017-18 Shield and the suspensions of Smith and Warner would give Maxwell an opportunity to play an inaugural home Test. But Justin Langer and co don’t think so.

Peter Handscomb might have to thank the rain for holding his spot in the Moneyball team for Sri Lanka, as a dismissal for 11 runs or fewer would have seen Joe Burns take his place.

Interesting to note was Matt Renshaw’s lowly position of 19th despite being a popular media and fan selection and making headlines for a triple century in Brisbane grade cricket. His form this Shield season has been far less impressive, scoring 199 runs at 19.9.

Fast/medium bowlers for Sri Lanka series

Minimum statistical requirements
Weighted Average matches played = 5
Weighted Average wickets = 5
Notable players excluded: JP Behrendorff, JA Richardson, JL Pattinson

Dark background denotes player would make XI

A bit of a surprise here as Chris Tremain ousts Mitchell Starc for a spot in the side.

The argument for Starc would not only be his better batting (weighted average of 23.99 compared to 14.51 for Tremain), but the fact he provides a left-arm fast bowling option. With 17 Test wickets at 36.82 in 2018-19, Starc has looked short of pace and failed to strike as much fear into the Indian batsmen as we’d have hoped.

Dropping him back to the Big Bash would give him rest and allow Tremain to make a well-deserved Test debut. He has been on a tear with 157 wickets at 20.96 across the past four Shield seasons, including a competition-high 51 in 2017-18, winning him player of the tournament. Whether his bowling would trouble world-class international batsman is uncertain, but he has earnt the right to find out.

(Photo: Paul Kane/Getty Images)

Slow bowler for Sri Lanka series

Minimum statistical requirements
Weighted Average matches played = 5
Weighted Average wickets = 5
Notable players excluded: Fawad Ahmed, AC Agar, A Zampa

Dark background denotes player would make XI

No surprises here as Nathan Lyon comfortably beats out Jon Holland and Steve O’Keefe.

The battle to be the second spinner came down to 0.01, showing the tough choice selectors will have when the next spin-friendly pitch comes around.

Fawad Ahmed (31.03), Ashton Agar (36.34) and Adam Zampa (45.09) are all below the required five weighted average matches played to be considered, but lose out to Holland and O’Keefe regardless.

It will be interesting to see if Lloyd Pope can make it onto this list in 2019-20.

(AP Photo/Themba Hadebe)

All-rounder for Sri Lanka series

Minimum statistical requirements
Weighted Average matches played = 5
Weighted Average runs = 250
Weighted Average wickets = 5
Notable players excluded: JS Lehmann, MC Henriques, GJ Maxwell

For all-rounders, I have multiplied the weighted batting average by two and then subtracted the weighted bowling average.

One school of thought is that the bowling average should just be subtracted from the batting average to find an all-rounder’s worth.

The other school of thought is that you should pick six specialist batsmen. By weighting the batting average as doubly important than the bowling average, I have aimed to strike a compromise between the differing viewpoints.

Dark background denotes player would make XI

A surprise selection for those not following Shield cricket too closely, Michael Neser has been a revelation. Opening the bowling for Queensland while batting number 8, Neser has proven himself to be a genuine all-rounder.

In fact, Starc and himself were the only two with a weighted batting average greater than their weighted bowling average from those who met the minimum requirements.

Pat Cummins (26.31 batting and 20.70 bowling), Steve Smith and Jake Lehmann were among those with positive weighted averages but failed to meet requirements (lack of runs mainly due to missing much of the 15-16 and 16-17 seasons for Cummins and lack of wickets for Smith and Lehmann).

With 312 runs at 52.00 and 13 wickets at 32.46 in the 2018-19 Shield, Neser’s form would seem impossible to ignore, yet somehow media attention seems focused on Labuschagne, Mitchell Marsh and Marcus Stoinis (sixth on this list, with 28.49 weighted batting and 34.97 weighted bowling).

The selection of Neser would provide Australia with an unusual opportunity to have a genuine fifth bowling option, who opens the bowling for his state, can swing the ball and took 39 wickets at 21.85 last season. His 52 wickets since 2017-18 are more than Labuschagne, Mitch Marsh, Head, Maxwell and Henriques’ 51 combined scalps.

There is a question mark around the sustainability of his batting, but his ability as a bowler, combined with Pat Cummins’ similar all-round ability in the 7 and 8 positions, make a debut selection for him against Sri Lanka a no-brainer.

Travis Head was the best of the batsmen that met the statistical all-rounder requirements but suffered due to a weighted bowling average of 63.90. Glenn Maxwell could not meet the requirements and has, in fact, taken just eight wickets since the start of 2015-16.

Australia’s Travis Head . (AP Photo/James Elsby)

Wicketkeeper for Sri Lanka series

For wicketkeepers, I have multiplied the weighted batting average by the weighted dismissals divided by innings kept.

There is a consensus idea that Tim Paine is a better keeper than Matthew Wade.

For this reason, I did not want to simply pick the keepers by their batting average. Unfortunately, I am unable to find Shield statistics for missed catches and stumpings.

Instead, to account for wicketkeeping, I have no choice but to rely on the statistic of dismissals/innings.

Minimum statistical requirements
Weighted Average matches played = 5
Weighted Average runs = 250
Weighted Average innings kept = 10
Notable Players Excluded: CT Bancroft, PSP Handscomb, SE Gotch

Dark background denotes player would make XI

Wade moved ahead of Paine following the latter’s five runs at the SCG against India.

Unfortunately, due to the difficulties in grading wicketkeepers from a statistical viewpoint, this is the selection that causes the most headaches.

Dismissals/Innings is a decent indicator, but is heavily skewed by the pitch, length of the innings as well as specific bowling and batting tendencies in terms of finding edges and drawing stumpings.

Wade makes the team on the back of a super Shield season so far with 571 runs at 63.44.

Meanwhile, Paine’s batting has dropped, averaging 24.5 in six Tests against Pakistan and India in 2018-19.

Moneyball squad for Sri Lanka

First XI: MS Harris, UT Khawaja, SE Marsh, PSP Handscomb, GJ Maxwell, MS Wade, MG Neser, PJ Cummins, CP Tremain, NM Lyon, JP Hazlewood

Reserves: JA Burns, MA Starc

Moneyball projected Ashes squad

First XI: MS Harris, DA Warner, UT Khawaja, SPD Smith, GJ Maxwell, MS Wade, MG Neser, PJ Cummins, CP Tremain, NM Lyon, JP Hazlewood

Reserves: SE Marsh, MA Starc, JM Holland, M Labuschagne, TD Paine