Has fast bowling in men's test cricket ever been this good?
Welcome to the first edition of Plot the Ball — a newsletter where I will offer data-driven answers to interesting questions I have about the world of sport.
This month — after spending lots of time following the recent Australia vs. England and South Africa vs. India series — I’ve been thinking about fast bowling in men’s test cricket, and wondering how the seamers of this era stack up historically.
Has fast bowling in men's test cricket ever been this good?
Ask anyone who follows men’s test cricket in its current era, and they’ll tell you the same thing: it’s a bowler’s game.
As journalist Tim Wigmore put it recently in a piece for The Telegraph:
“In the age of T20, the challenge of how Test cricket markets itself is greater than ever. Cut-throat contests in recent years suggest that the answer might be that Test cricket is the format where bowlers are ascendant. The unique selling point for Test cricket in the era…is: how do you score runs against bowlers this good and this relentless?”
In fact, it’s even worth reformulating that question to make it more specific: how do you score runs against pace bowlers this good?
As recently as March 2017, the average number of runs scored per wicket taken by pace bowlers in men’s test matches sat — on a 100-match rolling basis — at just over 32.
Since then, according to this metric, the performance of batters against seamers has declined precipitously.
In August 2020, the rolling average reached a new post-war low of 26.0, and has remained below 28 all the way through to the beginning of 2022 — markedly lower than the level at which it has sat for much of the preceding 75 years.
(The 1957-60 period — which bottomed out at an average of 26.1 runs per wicket in March 1959 — is the only stretch really comparable to this current one.)
It currently sits at 27.0, with that average covering the 100 tests played between March 2019 and January of this year.
Obviously, there are a number of reasons why the dynamic between batters and pace bowlers could have changed so significantly in such a short space of time.
Principally, the quality of batting could have declined of its own accord. However, the comparative resilience of global batting averages against spin — as the chart above displays — suggests that this isn’t the case.
It’s worth noting that research performed by analytics firm CricViz — and discussed by Ben Jones, their Head of Media Services, on a recent podcast — has apparently found that test batsmen are being dismissed more frequently when they play defensive shots on a per-delivery basis.
But, according to Jones, much of that has to do with the quality being delivered from the other end. On the same podcast, he said:
CricViz points to its ‘expected average’ figures — “an advanced metric that uses ball-tracking to illustrate what we would expect a team to average based on the quality of deliveries faced” — in support of this conclusion.
In the recent men’s Ashes, for instance, Australia recorded one of the best series of work in the company’s database — which currently goes as far back as 2006.
And it was primarily their seamers doing the work — led by the unerring Pat Cummins and supported by a formidable cadre of quicks including unlikely mid-series debutant Scott Boland, who slotted in almost immediately as one of the most accurate bowlers in the men’s game.
According to CricViz:
Further analysis of test bowling data shows that their seamers’ 30-game rolling average dropped below 25 runs per wicket during the series for the first time since 2002.
Australia’s performance by this measure has been remarkably consistent in the post-war period, however, rarely straying north of 30 runs per wicket for very long.
Of the four teams that contested major test series over December and January, it is India who have seen the most remarkable change in recent years.
As a testament to the transformation of their seam-bowling attack, consider the following.
In the post-war period, India have only had seven pace bowlers who have taken more than 20 test wickets at an average of lower than 30 runs conceded per wicket.
Four of those seven — Shardul Thakur (26 wickets at 21.0), Jasprit Bumrah (113 at 22.9), Mohammed Shami (209 at 27.1) and Mohammed Siraj (36 at 29.6) — took part in the recent three-match series against South Africa.
But even that exceptional attack wasn’t enough to win away from home against a resurgent Proteas side — another team led by a pace-bowling unit on a positive trajectory.
Over the course of the series, the shorter stature of Bumrah and his teammates meant that they struggled to extract as much from the pitch as South Africa did — particularly in the second half of games.
ESPNCricinfo Stats Editor S Rajesh showed in his analysis of the series that the home team — especially 2.06m-tall newcomer Marco Jansen — got much more out of shorter-length deliveries than India could.
The back-and-forth nature of those three tests in South Africa exemplified how the sport has become captivating viewing once more — in large part because of the renewed ascendancy of fast bowlers.
In a sense, it’s fitting that the most iterative of cricketing skills is again leading to success in a format of the game that — more than any other — rewards consistency and application.
What Wigmore said of Cummins at the beginning of the Ashes is more broadly applicable to many of his peers in this new world:
Ben Jones and Patrick Noone of CricViz on Australia’s unprecedented performance with the ball, the remarkable start of Scott Boland’s test career, Mark Wood’s toil Down Under and what their metrics tell us about this Ashes series on the whole
Kartikeya Date of A Cricketing View on India’s desire to take 20 wickets every match under the captaincy of Virat Kohli and measuring the depth of bowling attacks in this era of men’s tests
S Rajesh of ESPNCricinfo on how South Africa’s seamers ultimately came out on top of India’s
You can find the code for this piece on GitHub here
The main question I grappled with while doing the analysis for this piece was where to draw the line on rolling averages. There were a number of issues to consider: the increasing frequency of play over time (there were upwards of 40 men’s tests a year throughout the 2010s, compared to fewer than 20 annually for much of the initial post-war period), for instance, and the pandemic — which caused a change in the pattern of fixtures. I eventually settled on a 100-game rolling average for the aggregate charts — on the basis that, over the entire period, this worked out to somewhere in the range of two to four years of matches being considered at any given plotted point — and a 30-game average for the individual team charts. (This applied a similar rule of thumb: over the entire 76-year period, teams played an average of 593 tests each; 30 tests at a rate of 7.8 per year comes out near the top of the two-to-four-year range.) However, there’s still something arbitrary about such an approach — and, if I was repeating this analysis in future, I’d think hard about following football analyst Ben Torvaney’s advice and experimenting with exponential weighting.
For this first edition, I spent a fair amount of time creating and refining a custom theme in R — and I’m pretty happy with the results. I’m always forcing myself to try and cut back the amount of information I’m communicating in my charts so that their key messages are easily discernible, and I think the look I landed on does a decent enough job of providing structure without obscuring the story I’m trying to tell with the data.
Communicating colour encodings through titles and subtitles is something I’m a big fan of, and I think it works well in these charts. As useful as
ggtext) is for this purpose, however, I did miss the range of styling options that Datawrapper (the visualisation tool I use most in my day job) offers; for line charts specifically, the ability to build in a legend by underlining subtitle text with the variable colour (rather than changing the text colour itself) is a personal favourite.
Being able to provide the colour key as part of the subtitle in the small multiples chart was a big help, though — particularly as I struggled with customising text colours for the strip of each facet. South Africa’s period of exile between 1970 and 1991 also forced me to confront the problem of adding different annotations to different facets; I was initially planning to shade this area of the chart (with
geom_rect) as well as label it, but eventually admitted defeat and just used
geom_text. Getting better at dealing with these nuances is something I’ll focus on when I inevitably return to small multiples in the future.
Next month — to coincide with the return of the men’s Champions League — I’ll be looking at whether one of the most storied clubs in European football has changed its approach to squad-building in recent seasons.