Morne Morkel's stats in the recently concluded four-Test series against England were impressive enough - 19 wickets at 26.36 - but the general consensus among experts was that he was desperately unlucky and didn't get the rewards he deserved: several times, he beat the bat or forced an edge, but the batsmen managed to survive. Do ESPNcricinfo's ball-by-ball stats bear this out? Let's check the numbers.
Among the parameters on which each ball is tracked is one called the control factor, which records how a batsman responded to each delivery. If he middled a ball or left it alone, it is recorded as a ball that the batsman was in control of. If, on the other hand, he was beaten or edged the ball or mistimed the shot, it is recorded as a ball that the batsman wasn't in control of. For a batsman, the aim is obviously to get the control factor as high as possible (though there are instances of batsmen getting lucky and making big scores even with a relatively low control factor). Similarly, a bowler whose not-in-control percentage is high is one who has troubled the batsmen and induced several false shots.
In the England-South Africa, series, England's bowlers had the better numbers: they took 79 wickets at 24.50, compared to South Africa's 77 wickets at 28.77. (When comparing the pace attacks, though, there was no difference: South Africa's pacers averaged 27.51 to England's 27.67.)
However, in terms of the control factor, South Africa's bowlers were marginally better: 21.64% of the deliveries they bowled induced false shots from England's batsmen, compared to 19.57% for England's bowlers against South African batsmen. England's bowlers, though, on average took a wicket every 9.1 times they induced a false shot, while for South Africa that factor was 10.95, which means they had to force almost two extra false shots per English wicket.
The pace-spin break-up further illustrates that there was little to choose between the pace attacks of the two teams. South Africa's fast bowlers took more wickets and induced a higher percentage of false shots, but Moeen Ali had stunning success as a spinner, and his contributions went a long way in establishing England's bowling supremacy in the series. Moeen took a wicket every sixth time he induced a false shot, which was much better conversion rate than that of South Africa's lead spinner, Keshav Maharaj (10.41).
Among the England bowlers, James Anderson, who was outstanding with 20 wickets at 14.10, induced a false shot off 22.2% of the deliveries he bowled, but there were three South Africans who did even better than him on that count. The not-in-control per cent against Kagiso Rabada was 24.1, while against Morkel and Vernon Philander it was 23.58 and 23.54. Among the England bowlers, Toby Roland-Jones had a higher NIC percentage of 23.24.
Interpreting these numbers can be tricky, though. First, there is the question over the line of attack. In the overall series numbers, Rabada comes out as the bowler against whom batsmen played the highest percentage of false shots, but some of those balls were fairly harmless ones sliding down leg side, which the batsmen missed when attempting to flick. Of the 195 balls from Rabada that the batsmen weren't in control of, 31 finished down the leg side, which is a fairly significant percentage of nearly 16%. For Anderson and Philander, the corresponding contribution was 8%, and for Morkel 13%. It is true that teams sometimes deliberately bowl that line to strangle batsmen down the leg side, but Rabada clearly had problems controlling his line against a batting line-up that had an even mix of right and left-hand batsmen.
So, looking only at deliveries that finished either in line with the stumps or outside off, Rabada's percentage drops to 21.93, which is still impressive, but three bowlers did better than him: Philander (22.91), Roland-Jones (22.54) and Morkel (22.37). All four bowlers induced a slightly higher NIC% than Anderson (21.51), and these five bowlers were the only ones with NIC percentages of more than 20. Also, Morkel beat the bat 95 times when the deliveries were directed at the stumps or outside off, the most by any bowler in the entire series - Rabada was next with 86, while Anderson got past the bat 76 times.
* Excluding balls drifting down the leg side
The other aspect that needs to be considered when interpreting these numbers is the area where the bowlers pitch the ball, especially the length. In English conditions, where the overcast skies and the Duke ball combine to generate significant swing and seam movement, length is often key. Pitching it slightly short of the ideal length can cause plenty of plays-and-misses, but will seldom result in edges and wickets. Overseas fast bowlers often struggle to hone in on that perfect length, but that is hardly ever a problem for Anderson, who has perfected the art of bowling in home conditions. Moreover, his ability to swing the ball late also means he induces more edges even when he bowls a fuller length. With other bowlers who aren't as adept at late swing, a full length draws fewer errors from batsmen.
The break-up of lengths from which Morkel induced false shots (excluding deliveries which drifted down leg) reveals that from the 102 good-length balls when he drew false shots, he picked up 11 wickets, an average of one wicket every 9.27 such deliveries. From back-of-a-length balls, though, he only managed one wicket every 26 such deliveries, which is probably why he picked up the "unlucky" tag. Off fuller balls, though, his chances of picking up wickets increased when he induced false shots. Anderson's conversion rates are much better even against the short balls.
As mentioned at the start of the piece, Morkel's series numbers of 19 wickets at 26 are excellent, but if his conversion rate of taking wickets off not-in-control deliveries had matched Anderson's, Morkel would have ended up with 27 wickets in the series. Morkel bowled 225 deliveries that induced false strokes, taking a wicket 11.84 times, while Anderson took a wicket every 8.35 balls when he induced a false shot. However, as explained earlier, that might be too simplistic a way to look at control stats.