Insights

Mohammed Shami and the Question of Luck

Ben Jones looks back at the Indian seamer’s 2018 series, analysing of the unluckiest campaigns in the modern era.

2018 was a strange year for Indian Test cricket. Their tour of Australia saw them become the first Asian side to ever win a series on Australian soil, a historic achievement which stands as the pinnacle of Virat Kohli’s tenure. At the time, the caveat of David Warner and Steve Smith’s absence was tossed around to diminish the victory; the subsequent victory on the following tour swept those caveats aside.

Yet the two other landmark series for India in 2018 – first in South Africa, then in England – both ended in failure, and defeat. The 4-1 loss in England was particularly chastening, with Joe Root’s side very much in transition, still recovering from a winless Ashes winter.

Plenty of the inevitable criticism for India’s performance turned on the batsmen. In a low-scoring series, too few of the batting order stood up and offered consistent resistance, with only three batsmen (Kohli, Pujara and Jadeja, the latter batting just twice) averaging over 30.

However, on the bowling side of things India’s attack left with their heads held high. The finest group of seamers the country had ever taken on tour were relentless throughout, England’s batsmen in constant peril and unquestionably flattered by the scoreline. The only bowler on the flight home met with anything other than overwhelming support, was Mohammed Shami.

It’s understandable. While Shami’s 16 wickets left him second on the wicket-taking chart for the visitors, his bowling average – 38.87 – was the worst of anyone on tour. Across five matches where batsmen rarely got on top for any length of time, Shami’s figures stood out.

2adoInaWb9MfYXaoUJPwVMT2xnikQI2Kq7gGNkeGbpHEO7Hy9ugsbJICyGbP9ex2nkZgQPAGRYSPviIdR72qIAAvPO

And yet, a closer look tells a rather different story. The basic facts of runs and wickets present Shami as the outlier in a series of brilliant bowling, the weak point in an attack that had the opportunity to beat up an England side very much there for the taking. But – as was apparent to anyone watching with any level of attention – this simply was not the case.

For one thing, Shami created chances for fun across that summer. 26% of his deliveries brought an edge or a miss, the most of any bowler English or Indian. In terms of forcing errors from batsmen, nobody on show did a better job than Shami.

46RThsdUhM5U5H86d2RcPmU9HU2 sB0g4Wj4jyLROm idQ

If 26% sounds like a lot of edges or misses to you, then you’d be right. Only a handful of bowlers in recorded history have managed to maintain that degree of attacking threat across a series. Indeed, 2018-Shami drew the fifth highest percentage of false shots for any series in the CricViz database. 

y5cE17S9GflktppIlHyllz3CNZ9evzOMs3oMRX kegG1fkau2GVXny3G0MN3ALS4rfuCxcFJVIo6 BPVAecKDJbwNzx8nwBLAWKT Z5pw2VXT oBMtUVjqeUBkmymH1houUC7

In this light, the image of Shami’s summer is clearer; a tale of blunt underperformance transformed into one of almost constant, historically notable levels of threat.

But in order to get from these numbers, to a bowling average of nearly 40, there’s a logical step. That step is all about luck.

Shami had five catches dropped off his bowling in that series, a figure ‘bettered’ only by Anderson. To an extent we would expect this, given that these two sent down the most deliveries of the bowlers in this series, but given that only 55% of the chances created from Shami’s bowling were taken – the lowest success rate for any bowler on either side – it’s fair to say he suffered unusually tough punishment at the hands of his own fielders. Bowl as well as you like, take the edge as often as you like, but if the catches don’t stick then you and your figures are done for.

In essence, what this points towards is that Shami bowled very well in the 2018 series and, more importantly, that he bowled significantly better than his bowling average would imply. Shami’s Expected Average (xA) in the 2018 series – the record we would expect given the balls he bowled, according to our Expected Wickets model built on ball tracking data for his deliveries – was 23.2, a figure much lower than his ‘actual’ average of 38.87. Typically, Shami’s bowling in that 2018 summer would have brought far greater reward than he received.

While we almost always see xA and ‘Actual’ Average differ to some degree – for any number of reasons from size of sample, to quality of batting, to luck – this is a historically significant gulf. Of all seamers to take 10 wickets in a single series since 2006 (something which has happened on more than 600 occasions), Shami’s gap of around 16 between average and xA, is the fourth largest.

o vjlGLzgxGEms75EgCBDZ6gmFrByvx5HlAfQCA81SluaEvtY HBeEv9H1S2OuMXAQhTYDG143tUY22gFYY5qMIungvg3XaY0W2Oe iQK3mS1IIl O1Ma4nH49x 5jWrVrFcmj5

Undoubtedly, there are elements of the game which the model does not consider. Currently, it does not consider the importance of ‘setting batsmen up’, albeit the importance of this is analytically disputed in some quarters. It rewards bowlers for bowling ‘good balls’ but not necessarily for targeting specific weaknesses of individual batsmen – a model trained on the ‘average’ batsman naturally assumes you’re bowling to the, well, average batsman. These are all things we are working to improve, and which will no doubt yield interesting results in the future. But in its current state, this model is the best available tool for assessing how well a bowler actually bowled, and it suggests that in 2018, Mohammed Shami actually bowled extraordinarily well.

zf3QWGqnu6JfwSZXREKXRx HCNrO1qiHE0GIN5UMLUkD511g8 ALogJRyeQ1Krf7Ct6VDVVQXgyN9VLpbBtU

As the World Test Championship Final looms, afforded a longer build-up by virtue of the IPL’s cancellation, discussion has turned to selection. In Indian cricket, this has primarily focused on two questions: one spinner or two spinners, and the identity of the seamers. The latter has particularly been about whether Shami should give way to the promise of Mohammed Siraj, or the experienced Ishant Sharma.

The arguments on either side are clear and persuasive, and are deserving of closer attention on another day. Both Ishant and Siraj are wonderful bowlers, albeit with differing skills, and both have reasonable claims on a place in the attack for this unique game in cricket history. The Shami who flies to England this summer is not the same man who flew out in 2018, a statement made without judgement but with an understanding of the toll age and injury can take on any sportsperson, particularly fast bowlers. The misfortune or otherwise that Shami experienced in 2018 does not make him a better (or worse) pick this time around, but dismissing him on the basis of how he bowled last time is foolish. Those performances belong to the case to include Shami, not to exclude him.

Prodding at received wisdom, or suggesting that a scorecard may not tell the whole truth, is a funny old thing. One reason why people push back against using numbers to comment on cricket, or sport more generally, is the assumption that those numbers – and those using them – are claiming objectivity. When introducing data to a discussion about a player or a team, sceptics assume an implicit arrogance on behalf of the analyst. Hence the criticism: “numbers don’t show everything”.

That is of course, partly true. No single number shows everything, and one of the issues with cricket’s relationship with numbers is that plenty have pretended this isn’t the case. By hoisting batting and bowling averages to a status where they speak volumes on every player, where their entire worth and quality is reduced to a single figure, cricket has wilfully created an environment where people can dismiss the value of statistics while being entirely beholden to averages. At various points – often within the same breath – people will disregard the importance of ‘stats’, while citing a player’s average as evidence of their quality. Batting and bowling averages are not seen as being on a spectrum with other more advanced metrics, but a separate and transparent vessel of truth, thereby ignoring the swathe of other measures which often provide clearer, more contextual, more specific information.

In the case of Shami in 2018, you can look at the averages and decide he bowled poorly – that is your prerogative. But you’re rewriting history as much as anyone else, and are making a claim of objectivity far greater than the other people in the room. If one metric is telling you Shami bowled poorly but three, four, five others are saying he bowled exceptionally well, then criticising his performance is placing a rather large amount of faith in one single number.

Which is dangerous, if numbers don’t tell you everything.

Ben Jones is an analyst at CricViz.

Share the Post: