The pace v spin story of the last two India-South Africa series

South Africa's fast bowlers bowled 90% of the team's overs in the home series, while India's spinners bowled 75% of their team's overs in the series in India AFP / Getty Images

A review of the ball-by-ball control record for India's Test series in South Africa earlier this year showed that the hosts won a bowlers' series because they were able to create uncertainty with greater consistency. The visitors had a bad day first up and lost. This essay is intended as a follow-up to the conclusions of that review. It offers a comparative look at South Africa's four-Test series in India in 2015-16 and India's three-Test series in South Africa in 2017-18.

The overall comparison between the two series is summarised in the table below. ESPNcricinfo's control measure classifies each delivery according to whether or not the batsman was in control (IC) or not in control (NIC). The data in the table is organised by bowling team. BB is the number of balls bowled by each bowling team. IC_PER_NIC is the number of in-control deliveries per each not-in-control delivery. IC_ECON is the economy rate conceded from the in-control deliveries. %NIC_RUNS is the percentage of the total runs conceded from the not-in-control deliveries. NIC_BB_PER_W gives the number of not-in-control deliveries required on average for one wicket. %NIC provides the same information as IC_PER_NIC as a percentage. Some readers may find it easier to interpret uncertainty as a percentage rather than as a frequency. The frequency measurement is useful if one keeps in mind that an over includes six deliveries. So IC_PER_NIC above 5 means that uncertainty may not be created every over, while IC_PER_NIC under 5 means that uncertainty is created every over.

The series in India was dominated by spin, while the series in South Africa was dominated by pace. The comparative extent of this domination is notable. Pace played a far larger role in South Africa than spin did in India. In India, the host spinners bowled three out of every four overs overall. In South Africa, the host pace bowlers delivered nine out of ten overs. For comparison, in the 2017-18 Ashes, England's fast bowlers bowled 73% of England's overs, while Australia's fast bowlers bowled 69% of Australia's overs.

The series in South Africa was dominated by the ball to a far greater extent than was the series in India. Both sides used occasional bowlers in India, while all the overs were bowled by specialists or allrounders in the South African series. The overs delivered by non-specialist bowlers (JP Duminy, Stiaan van Zyl, Shikhar Dhawan, Stuart Binny, M Vijay, Virat Kohli and Cheteshwar Pujara) on both sides have been ignored in the data below. Overall, a wicket cost 19 runs in the 2015 series, and 21.6 runs in the 2018 series. But a wicket fell every 48 balls in the 2015 series, compared to one every 46 balls in the 2018 series. The batsmen were not in control 17.2% of the time in 2015 and 24.5% of the time in 2018. Whatever complaints they might have had about the pitches in India in 2015-16, South Africa returned the favour with interest in 2017-18.

Taken together, the data from both series reinforces the view that the difference in quality and depth between the Indian and South African spin attacks is far greater than the difference in quality and depth between the Indian and South African pace attacks. The South African pace attack is better than India's, while India's spinners are better than South Africa's. India's specialist batting in South Africa was better than South Africa's specialist batting in India, but this difference is arguably fully explained by the difference in bowling attacks.

As might be expected, the series in South Africa, dominated as it was by pace, produced a higher share of inadvertent runs (i.e., runs from deliveries where the batsman was not in control) than the series in India. More interestingly both sides found wickets easier to come by in India when they beat the bat than they did in South Africa. In general, pace could be said to create greater all-round uncertainty. Spinners beat the bat less often than fast bowlers, but when they do, they are more likely than fast bowlers to get a wicket.

Note that for the purpose of this essay, "beat the bat" may be considered synonymous with "not in control". It refers to the bowler beating the middle of the bat.

A comparison of the two pace attacks in India and South Africa is instructive. It underlines the overall superiority of the South African attack. This can be seen from the greater uncertainty in India's selection of fast bowlers. But mainly, it is evident from the fact that South Africa's fast bowling created more trouble for India's batting in India than India's fast bowlers did for South Africa's batting in India.

Similarly, India's spin bowling was superior to South Africa's both home and away. Though R Ashwin and Keshav Maharaj played a limited role in the series in South Africa, Ashwin troubled the South African batting far more than Maharaj did India's batting. In India, the difference was decisive. Ashwin, Ravindra Jadeja, and to a lesser extent Amit Mishra, allowed India to dictate proceedings in India in a way that Imran Tahir, Simon Harmer, Dane Piedt and Dean Elgar were unable to match, either in terms of creating uncertainty or in controlling the scoring.

In the series in India, the superiority of the Indian spin attack was decisive in the first innings of the matches. This left South Africa playing catch up, often leaving them with no option but to play for a draw. In South Africa, this was not the case. India stayed in the games over the first innings. However, they were unable to control the scoring as well as South Africa did. With the pitches wearing later in the game, the extra pace and height of South Africa's bowlers proved decisive.

Four bowlers - Ashwin, Ishant Sharma, Kagiso Rabada and Morne Morkel played at least two Tests in both series. The comparison between Rabada and Ishant is especially instructive. In India, the former beat the bat 54 times for two wickets, while the latter beat the bat 46 times for a solitary wicket. But Rabada created far greater uncertainty than Ishant, beating the bat 18% of the time, while Sharma beat the bat only 12.1% of the time. Morne Morkel beat the bat 83 times for nine wickets in India, and 150 times for 13 wickets in South Africa. He created about the same amount of uncertainty as Rabada did. Despite collecting only a couple of wickets in India, Rabada troubled the batting consistently.

Ashwin beat the bat 198 times in India for 31 wickets, and 72 times in South Africa for seven wickets. His most significant effort came in the second Test in South Africa, in Centurion. In the first innings there, he beat the bat 31 times in 233 deliveries and returned took wickets. In the second, he beat the bat 33 times in 177 deliveries for a solitary wicket. In contrast, on the opening day of the series, he beat the bat six times in 43 balls and took two wickets. Much like Rabada in India, Ashwin created significant uncertainty for less than ideal returns in South Africa. Even though the conditions did not favor his style of bowling, the control numbers reveal Ashwin's class. Over a longer series, Ashwin's returns would probably have improved and been commensurate with the amount of difficulty he created for South Africa's batting.

The records of the batsmen who played at least two Tests in each series suggest that the batting records are largely determined by the quality of the bowling (and consequently, the match situation) each batsman is likely to face. In India, AB de Villiers and Hashim Amla were led by the match situation into a tactical decision to defend. In the first innings of the matches in India, de Villiers scored 174 runs from 224 deliveries where he was in control (4.6 runs per over). In the second innings, he made 57 runs from 281 deliveries where he was in control (1.2 runs per over).

Similarly, in South Africa, Pujara and Vijay made the tactical decision to defend. Pujara made 84 runs from 275 balls where he was in control, and Vijay 81 runs from 291 balls where he was in control. For South Africa, Elgar adopted a similar approach, making 159 runs in 409 balls where he was in control. In contrast, both Kohli (237 in 383) and de Villiers (196 in 272) chose to attack the bowling. Elgar could be considered the most fortunate batsman in the series. He survived 134 deliveries where he was not in control and was dismissed only five times.

Finally, consider these respective core batsmen (I include Ajinkya Rahane to the list for this next data set) against pace and spin in India and South Africa respectively. This is the core contest in each series, the main batsmen against the specialist bowlers. In India, India's five core batsmen conceded a wicket every 86 balls to the South African pace attack. In South Africa, despite conditions being 50% more difficult (21% not in control as opposed to 14% in India), they managed to keep the South African pace attack at bay for 64 balls per dismissal.

South Africa's problem in India was not that their spinners were not ably supported by their pace bowlers. Rather, it was that their pace bowlers were not ably supported by their spinners. Tahir, Piedt, Harmer and Elgar were unable to control the scoring in India any better than Maharaj was able to in South Africa. Given how spin-friendly conditions in India were, this does not reflect well on Tahir and company. South Africa would probably have been better off playing three or four fast bowlers and the solitary spinner in Tests in India, much like India in the past found success overseas by fielding their best bowlers, even if two of those were spinners. As it happened, South Africa ended up having to use their fast bowlers far more than they planned to in that series in India. They might well have won a Test somewhere had they played to their strengths and been prepared to bowl 70-75% of their overs with pace.

India's bowling options are not similarly lopsided. Their spinners are currently world-class, and their fast bowling has greater depth and experience than it has ever had, with the depth to compete and stay in the game. With a little bit of luck, they can be competitive in a Test match anywhere in the world thanks to this depth (they now have four competent bowlers, even if none of them is currently of the quality of a truly elite international bowler like, say, Ryan Harris).

Control is not a retrospective measurement. It is recorded ball by ball. The overall picture it creates is therefore significant, since it is not coloured in any sense by the outcome of the game, or by any other summarised conclusion. It successfully separates performance from outcomes and provides a picture of the game that is not available simply by watching.

Watchers are prone to cognitive biases. For a classic example of this, consider two cases. In the first, the batsman scores 110 not out in a run chase and his team wins. In the second, the batsman scores 98 and his team loses by 25 runs. The first 98 runs in each innings are remembered differently, and more importantly, are characterised as being of different quality after the result. This despite the fact that for those first 98 runs, player and watcher alike were equally unaware of what the final result might be. This is not to say that the ball-by-ball team that creates the control record is not prone to biases. Rather, it is to say that the discipline imposed by having to consider one ball at a time localises these biases.

A look at the ball-by-ball record also allows the relative importance of different aspects of the contest to be considered. For instance, dropped catches tend to stick in the memory. But consider a session in which a team created 20% uncertainty (or 36 balls in 180), and dropped a catch. Consider another team, which created 12% uncertainty (or 22 balls in 180), and dropped two catches. The team that dropped two catches is likely to be considered by observers to have (a) done worse, and (b) created more trouble than the team that dropped just one catch. Among mistakes (or chances), a dropped catch is a loud, obvious one, which sticks in one's mind. Other chances, like beating the outside edge four or five times more, are not similarly memorable, even though the batsman has the same amount of control when the result is a the play-and-miss as he does when the result is an edge or miscue, and plays and misses happen more often. But it is difficult to convince observers that creating chances is more significant (albeit less obvious) than converting each chance. Things that are obvious are not always more significant than things that are not. Bowling sides that create more uncertainty are more successful in the long run (such as the course of a Test match or series). But not all of this uncertainty is equally obvious. The control measurement, by systematically recording uncertainty, can help make this point to observers who are otherwise learned watchers of the game.

The order in a contest can be identified from the control record. This is not always possible from the scorecard. Control should become a mainstream measurement in the cricketing record as soon as possible.