Archive | February 2013

Infernal machines

Sitting by my Dad’s hospital bed, I recognise a familiar sense of distraction. A monitor, displaying live graphs of my father’s vitals, keeps drawing my attention away from the man himself. What it reminds me of is watching cricket on television. The Sky (or other local) production, sets its image on a platform of game data. My eye is drawn repeatedly away from the match image to the game data to note, in particular, the speed recorded of the delivery just bowled.

(My Dad is recovering well from a surgical procedure known as the Whipple – a name waiting for an improvised T20 shot to be attached to it – although he has put some more distance beyond his days of taking the first nine wickets in an innings or even sharing a century opening stand with me.)

This afternoon, one of the monitoring machines started bleating. It displayed a message: ‘air in tube’. I went to get a nurse who reset the machine, squeezed a tube, and explained, “there was air in a tube.” We nodded and remembered the story of my Mum being pre-opped for some minor surgery. The nurse tried to take her pulse, but the machine’s read out was stubbornly off-kilter and so the nurse shoved it away to be sent for repair. Several weeks after the surgery, my Mum’s doctor took her pulse, trusted his reading and had her booked in for a pace-maker that month.

Do we trust the machines that interpret the cricket we watch, or do we prefer to trust what we think we know?

In recent days, my cricket viewing has been distracted by the speed-gun’s read-out for the world’s finest fast bowler and a young English all-rounder. Dale Steyn’s profile invariably features the descriptor RF. Chris Woakes is generally understood to lack the pace to hold his own in Test cricket and will tend to have his F qualified with a preceding, not succeeding, M.

Steyn was bowling South Africa to a strong position in the second innings of the second Test against Pakistan. He was keeping to a tight off-stump line, registering 132-136kph. Woakes was bowling some economical middle-inning overs against New Zealand in the first one day international. His range was 130-137kph. Neither set of commentators mentioned that the bowler was operating at a pace unusual for themselves. Steyn remained RF and Woakes RMF.

This disjuncture between a bowler’s reputation and their recorded speed is fairly regular. Courtney Walsh continued to be called a fast bowler when delivering at a Trott-like 128kph. In the eyes of English cricket commentators, Sri Lanka (Malinga excepted) and Bangladesh only ever have medium pacers. These commentators find the on-screen data less distracting then I do, because I notice the numbers indicating that they tend to bowl at around the same pace as England’s quick bowlers.

How reliable are the ‘speed gun’ results? To be comparable, bowler by bowler, match by match, ground by ground, the camera must record the velocity of the ball at the point of release from the bowler’s hand – the point of maximum speed. Mohammed Irfan (left arm over, 7’1” tall), Tim Bresnan (right arm over, wide on the crease, 6′ tall) and Fidel Edwards (right arm over, slingy action, 5’8” tall) have very different release points. I can imagine technology capable of capturing all three accurately, but I don’t know if that’s what we have in place at every international cricket fixture.

mirfan    fideledwardstim bres

Watching Steyn, I pondered a) for the thousandth time how speed in a bowler is over-rated, that control and movement must be the key to effectiveness; and b) whether his narrow features and niggling moustache make him look more like a hill-billy car mechanic or a 1940s black-marketeer.


Younis Khan was subjected to verbals from the niggling moustache and, very much more impressively, an old-fashioned working-over, peaking with a bouncer clocked at 149kph. Not long after, he played on, trying to force a fast, short delivery through the off-side. We didn’t need the speed-gun to tell that Steyn had upped his velocity. If we didn’t feel we could trust our own eyes, we only had to look at Steyn’s.

Close match-ups and one-sided matches

Australia’s cricketing decline at the end of the last decade opened up Test cricket to one its most evenly competitive eras. A pack of well-matched teams have traded blows and consistently defeated a second group made up of weaker teams. This period may be about to close as South Africa are now pushing ahead of the pack.

The group of well-matched teams numbers six: Australia, England, India and South Africa – all of whom have maintained ICC rankings of above 100 since 2008 –  Sri Lanka and Pakistan – the former in decline, the latter ascendant.

This post considers the contests between these well-matched teams and assesses the sort of match outcomes produced. The aim is to test whether this era of relatively evenly matched teams has led to competitive test cricket.

The six teams provide a possible 15 bilateral contests. 12 of these match-ups had two or more series (including two test contests, aka two-offs). Two had just one series (South Africa v Pakistan and SA v Sri Lanka) and no test matches were played between Pakistan and India.

The chart below shows the outcomes of the 12 contests with two or more series. Five of the 12 contests saw each side winning a series. Only two of the contests were one-sided – one team winning all series. At this level of analysis it seems to have been an era of competitive, close match-ups. South Africa provide something of an exception, losing a series in only one of the contests. All the other teams lost series against three opponents.

test match ups-page-001

The six Test teams contested 33 series in this period. The chart below shows the distribution of series results. Two-thirds were closely contested, using the indicator of a drawn series or a one test margin. Almost half saw both sides winning matches. But there were some one-sided affairs.

close series-page-001

This post ends with a match-level analysis of how close the contests were. To do this I have introduced the concept of the ‘major’ victory, where the margin of victory is significant. I settled on two measures:

  • a victory by 6 wickets or more (using median partnerships, 75% of a team’s runs in two completed innings would be scored with the loss of 14 wickets)
  • a victory margin in runs equivalent to 25% or more of the runs scored by the winning team in the match.

In total, 100 tests were played in these 33 series. 25% were draws; 49% were home wins; 26% away wins. The incidence of major victories, home and away is shown below. Playing at home or away, 73% of victories were by margins that I have considered to be ‘major’.

victory sizes-page-001

Included in the major victories are three where the winning team scored 40% more runs than its opponent; 23 innings victories and 9 victories by nine or ten wickets.

(I have not attempted to assess teams’ relative positions in drawn matches, relying instead on the notion that those games are indeterminate. I acknowledge that a detailed review of drawn games would show some that one team was odds-on to win but for weather interruption.)

The impression of how closely contested these fixtures were shifts as the analysis moves from the high level analysis of series results down to individual matches. At the higher level, there were few one-sided contests as most involved both teams winning a series or one team victorious in a series but drawing the return series. Looking at individual series results two-thirds had teams drawing or separated by a single result. But at the individual match level, almost three-quarters of victories were by significant margins.

This is the paradox I call close match-ups and one-sided matches. In a future post, I will look in more detail at the matches to see how two even teams end up with uneven results.

The ICC and the International Year of Statistics

2013 is the International Year of Statistics. On the balance of probability it was bound to get its own year sooner or later. Whether it is found to be significant for the whole 12 months, time will tell.

On hearing of this year’s appellation, I thought it an opportunity to explore some issues with data recorded at international cricket fixtures that intrigued me. The answers I sought weren’t on the ICC website, so I emailed a query to

I noticed that 2013 has been designated International Year of Statistics, which prompted some thoughts about the statistics that I care most about: cricket statistics. I would be grateful if you could help me with these queries:
1. Is the ICC planning any developments to mark the International Year of Statistics?
2. Who owns the data from which international cricket statistics are drawn? Is it the ICC, the associations of participating teams, the host?
3. Are the detailed data of international matches collated and held centrally? Match scorecards are widely published, but is there a repository of the underlying ball-by-ball data? If so, how can an individual access it?
4. Is there a database of non-match cricket data held by the ICC (e.g. players’ dates and places of birth)?
5. Most international matches have high standards of television coverage, including ball-tracking and ball speed cameras. Who owns the data from these devices? Are these data collated centrally? Are these data held in a way they can be cross-matched with the ball-by-ball score data?
I look forward very much to hearing from you.

I sent that email on 8 January; requested an acknowledgement a week later and then having stumbled across the name of the ICC’s Official Statistician, David Kendix, I re-sent it to his attention on 27 January. You may have guessed that I’ve not had a reply.

I am more interested in the answers than in criticising the ICC. It would be wrong to be too harsh on their customer service. I would imagine they get hundreds of queries daily, many scurrilous, many others asking for services they don’t provide or pitching sales. And they have had a busy month: the build-up to an international tournament, managed in a ‘last-minute Larry’ fashion; a Board meeting where one of the game’s most contentious issues – DRS – was kept off the formal agenda. So, I can see why my questions dithered in somebody’s in-box before disappearing to the trash.

Anyway, the ICC have done me the honour of following me on twitter. I double-checked this after I told a cricket-playing colleague, who looked alarmed that the International Criminal Court was taking an interest in me.

Back to my questions. I make use of cricinfo’s statsguru function for exploring cricket statistics. It’s an admirable application, pretty flexible, free and authoritative. But, increasingly I am finding that the features of the game I want to explore are not easily queried.

For example, a friend (@ghdunn1) asked whether the new ball was a more potent feature of Test cricket now than in the past. It took me over an hour, working through scorecards, to generate the analysis depicted below on last summer’s England v South Africa series. A database of ball-by-ball data could return an answer to that very important question about one of the variables in the game in less time than it took me to analyse a single series manually (NB clearly the availability of ball-by-ball data is a limitation).

new ball analysis-page-001

While it would be wonderful for ball-by-ball data to be available for a hobbyist such as myself, I don’t pretend that is sufficient reason for time to be invested developing such a database. But I do think there is a justification.


England are probably the world’s best-resourced test team. One of the methods they have employed to gain competitive advantage, is the detailed analysis of their own players’ and opponents’ performance. The exact manner they do this, and the resources used, are not made public.

One of the concerns supporters of test cricket wrestle with is the polarisation in performance amongst the small number of test playing nations. The ability to carry out detailed analysis has not created that polarisation, but it reinforces the competitive advantage of the richer nations. A free database would counter that. Even if a team lacked the money to employ analysts, I reckon they could crowd-source the analysis they needed from the many part-time and hobbyist statisticians across the world.

I emphasise that the database should be free to use. It may be that I misunderstand the ownership of cricket data (see question 2 to the ICC), but it ought to belong to all of us who follow the game.

You may have the answers to the questions I posed the ICC. If so, please let me know. If you don’t, but think they are important or interesting, perhaps you would email or tweet the ICC?