Archive | statistics RSS for this section

Falling short of 200, again and again

two hundred178 152 161 148 94

If Indian cricket fans needed a 24 hour telephone help-line, this could be the number. It’s the sequence of five innings scores that saw the optimism of leading the series after Lord’s get swung and spun into humiliating defeat.

Five consecutive sub-200 run completed innings in Test cricket is a rarity. It’s 55 years since India’s only ever previous sequence of this kind – also in an away series against England. In 1959, they faced Trueman and Statham. But just as in 2014 it wasn’t always Anderson and Broad who took the wickets, so Tommy Greenhough (a Lancashire leg-spinner), Harold Rhodes and Brian Close prospered in that earlier series.

Scoring fewer than 200 runs in a completed innings is itself not unusual. There have been 1,478 instances in the 2,137 Test matches played to date. 28% of all completed innings fail to top 200. What is rare, in Test history, is the run of these run-shy innings.

The longest sequence suffered by a Test team stretches over 11 Test matches and four years. From July 1886 to August 1890, Australia were dismissed by England 21 times in succession for scores as low as 42, but never higher than 176. Every match, but one, was lost. In July 1888, scores of 116 and 60 by Australia were sufficient for a victory at Lord’s with the comfortable margin of 60 runs.

The late 1880s were a period of exceptional low scoring in Test cricket. In four of the Tests that feature in Australia’s skid, England set their own record of eight consecutive innings without making 200, while winning three of the matches.

South Africa own the equal second longest streak of this kind, which began in the late 1880s and continued until the second half of the next decade. These were the African colony’s first six Test matches, all played at home against England and lost.

Bangladesh have also endured a run of 12 innings of sub-200 scores. It started 13 months after their Test debut, when they had not only topped 200 but reached 400. Dizzy heights. Across seven Tests, in four countries on three continents, Bangladesh went from December 2001 until October 2002 making 12 scores between 108 and 184.

The West Indies were another side to find 200 too steep for a period of their early days in Test match cricket. In their first series away in Australia in 1931, the West Indies had six consecutive sub-200 scores. They broke that sequence at Sydney in February 1931 with their first victory away from home, losing only 11 wickets in the match.

Two other Test teams have had runs of six completed innings below 200. New Zealand in England in 1958 didn’t even make it to three figures four times out of the six innings. Australia in 1979, rebuilding without most of their established stars who were playing World Series Cricket, finished the Ashes with five scores below 200 and started a series away in Pakistan with a sixth.

Then one notch down comes the India team of 2014. Not a nation at the outset of its Test career; not battling in tense encounters on uncovered nineteenth century pitches; not deprived of its finest players by defection or selection.

India’s streak is alive, although its prognosis is poor. Their next Test match starts on 31 October at the Hyderabad (Deccan) ground [note 1], where India’s lowest completed innings score in three Tests is 438. The opponents are the West Indies who conceded 453 and 495 in India’s two innings in Tendulkar’s farewell series last autumn. It will take something truly wretched for the India team of 2014 to go one step further and become holders of the joint fourth longest streak of sub-200 Test match innings scores.

Note 1: India did not play their next Test in October, because of the abandonment of the West Indies series owing to the tourists’ dispute with their Board. The start of India’s next series, in Australia, was delayed for compassionate reasons. Their opportunity to end this streak will take place at Adelaide, where India have only once been dismissed for less than 200. Their most recent innings at Adelaide (January 2012), however, only amounted to 201.

(updated 7 December 2014)

Back-to-back Tests

Embed from Getty Images

The first Test at Lord’s finished after 6pm on Monday night. Three days, 16 hours later, England and Sri Lanka return to the field at Headingley on Friday. If the second Test lasts as long as the first, the teams will have played ten days of cricket in 13 days.

Later in the summer, the five Test series between England and India involves two pairs of back-to-back matches, separated by the third Test that starts six days after the scheduled end of the second Test and is itself due to complete just seven days before the opening day of the fourth Test.

Back-to-back Test matches – borrowing the term from the design of labourers’ homes, thrown up during the Industrial Revolution close to factories and mills, with nothing but an alleyway separating the row facing one road from the next row looking out on a parallel street. In international cricket, the alleyway is the two, three or four days for players to rest, travel and recuperate before throwing themselves into another five day contest.

I think of back-to-back Tests as pairs of matches where the second begins in the same week as the first ended. And on occasions when the first contest has been compelling, with a thrilling denouement, the second Test is under way before I have fully absorbed the performances and the significance of the match just finished. The game at Headingley starts with my mind still reflecting on the trail of shots, wickets and captaincy decisions that took the Lord’s game to the final ball of its final over.

Back-to-back matches are a feature of the modern Test tour. The programme of international fixtures, by historical comparison, turns over rapidly. Pre-series warm-up matches have been jettisoned or compromised as proper cricket matches. The international encounters themselves have been distilled in many cases to two Test encounters. Days between matches incur cost and generate no revenue. Get the players back on the pitch. Back-to-back.

So common has this pattern of play become (see graph below) that I propose we drop the term ‘back-to-back’. It has become redundant, an unnecessary qualifier. A more interesting term would be one coined for the rare Tests that are spaced more generously. We could keep the domicile reference: semi-detached; borrow another term that confers judicious distance: arms length; or just refer to the calendar: fortnightly.

back to back tests 2010s-page-001

But before we allow the back-to-back Test to sate our impatience for more cricket, there are questions to be asked. Does it make for good Test cricket? Can players flourish when asked to climb again the mountain they’ve just scaled (or trudge through the valley, into which they’ve sunk)? Are contests skewed by events in the first match, from which the opponents have insufficient time to recover? Do we see the best of the best players, or just those whose bodies are resilient, or know how to play within themselves?

The major Test nations collect and keep data on the physical condition of all of their players and so must know the toll, or otherwise, of playing Test cricket one week after another. The data they keep, they keep to themselves. I have seen one published study on the recovery time for a professional cricketer. It simulated the exertion of scoring a century in a one day international and found that the batsman took a week to regain the level of fitness exhibited at the start of his innings. Assuming the Test match innings is similarly depleting, only the few players scoring heavily on days 3-5 of the first match will be below par at the start of the second of the pair of matches.

It’s the fast bowlers that are of more concern. And we know this concern is felt by the authorities by their selection decisions. The Australian policy of rotation, practiced most notoriously in 2012/13, is the highest profile example. The decision to rest Peter Siddle and Ben Hilfenhaus who bowled 33 and 34 fourth innings overs respectively, from the next and deciding Test, which began fewer than four days later, stirred public discontent.  One month later, Mitchell Starc was left out of the Boxing Day test, which began a leisurely week after he had bowled Australia to victory over Sri Lanka. National Selector, John Inverarity, was quizzed on the policy and responded “I presume you mean informed player management.”

The authorities that brought you back-to-back Tests and compressed international tours were, it was perceived, rationing the public’s view of the players they wanted to see. Scientific advisers cautioned that the players, who make the sport, weren’t able to cope with the sport’s intensity.

Darren Lehmann distanced himself from the policy on taking charge a few months later and Australia recommitted itself to playing its strongest team in all Tests. Over 14 weeks last southern summer, Lehmann flogged through eight Tests a 34 year old quick with dodgy knees, as well as a mercurial 32 year old, with an exceptionally stressful action – but one who bowled in four over spells, even if he had taken a wicket in the fourth over.

Rather than fatigue, it has since been said that each of Starc, Hilfenhaus and Siddle were carrying injuries that caused them to miss these matches. This is probably the real risk that back-to-back Tests pose to the contests being the best against the best: abbreviated turnaround times giving bowlers insufficient opportunity to recover from minor knocks and strains – or risk those injuries becoming graver.

Another question about the desirability of back-to-back Tests is whether running one Test so soon after another prevents the contest developing and favours the team that gets ahead in game one. My sample of recent Tests was so heavy with back-to-back matches that I did not have enough control data. To find this I went back twenty years and looked at Tests from the 1990s. That recently, the spacing of matches within a Test series was very different.

back to back tests compared-page-001

To get a view of whether back-to-back Tests influence outcomes I have (holding my hands up to acknowledge that it is methodologically suspect) combined my sample of Tests from the 2010s with the sample from the 1990s. Looking at pairs of matches where the first game had a decisive outcome, I have calculated how often that outcome was repeated or reversed.

back to back test results-page-001Surprisingly, the pairs of matches with longer break periods had a higher proportion with both matches having the same outcome. Back-to-back Tests were, however, less likely to produce a pair of games where the loser of the first game bounced back to take the following match. Perhaps the key figure, although one that is not easy to interpret, is that back-to-back Tests were more than twice as likely to follow a decisive result with a drawn match. This could be evidence of fatigue affecting both teams, or of less enterprising play, or simply a feature of this sample.

Back-to-back Tests don’t merit comment because of their rarity, but because they have, relatively recently, become the common currency of Test cricket. Along with lengthy ODI series, back-to-back Tests are where the pressure of a full international calendar seems to erode the standard that the highest form of the game should be of top quality and played by players in full fitness. Whether the hustling of one match so soon after the first actually affects the outcome of those matches is far less clear and probably deserves a more comprehensive review. It’s the sort of question one would hope, but sadly not expect, the ICC to investigate.

 

Graeme Smith declares

Embed from Getty Images

Graeme Smith’s Test career ended in defeat and South Africa’s first series loss in five years and 15 series. He must have drawn satisfaction from his final, albeit unsuccessful, day of Test cricket as his middle and lower order battled and came so close to eking a draw from a game in which they were profoundly outplayed.

Smith’s records as captain – most matches, most victories – have been celebrated. There’s another record he holds, less well known, but which makes the denouement of his final Test ironic. Australia won at Cape Town with time almost, but not quite, up after declaring in the third innings. Smith is the most prolific declarer in Test history, having called time on South Africa batting in the third innings of a match 25 times. Yet, only eight of those declarations were converted into victories – producing a success ratio of 32%, below the average for the game and a long way short of the proportion achieved in recent years of Test cricket.

South Africa have become known for being a team hard to beat and one that finds it hard to speculate to accumulate if the victory isn’t coming comfortably their way. Smith’s statistic of only converting 32% of declarations – made when ahead in the game – into victories gives fuel to this notion of a risk averse South African side. The purpose of this post is to assess whether, from the perspective of the target-setting declarations, this is a fair assessment of Smith’s career, which in so many respects deserves to be remembered for the effectiveness of his leadership.

Smith’s third innings declarations spanned from his third Test as captain (v England at Edgbaston in July 2003) to his 107th and penultimate Test (v Australia at Port Elizabeth). In total, these declarations yielded eight victories, 16 draws and a single defeat. How many of the draws were likely victories spurned through over-caution?

Three of the draws can be discounted immediately. These weren’t ‘target-setting’ declarations, but decisions to close the innings to end a match that hadn’t progressed beyond the third innings. Adjusting for these, Smith’s win rate increases to 36%. A further adjustment, to take out two games where the fourth innings was prevented by rain from running its course, lifts the success rate to 40%.

The chart below shows the target set and estimated number of overs remaining in the match for each of the 22 ‘live’ declarations (including the two rain-affected games). Matches that were won by South Africa are green, defeats in red and draws in blue. The dotted turquoise line indicates the current record fourth innings chase in Test cricket (418). On ten occasions, Smith has challenged opponents to set a new Test fourth innings record.

G Smith decs 1-page-001

Smith’s teams have won each of the matches where they have given themselves over 150 overs (five sessions) to bowl out the opposition. They have won only one-third of the nine games when they had 100-150 overs. Not a single game has been won with the declaration leaving a day’s play (90 overs) or less.

To test for examples of over-cautious captaincy, it makes sense to start with the four matches where Smith set a target of over 400 and gave his team over 100 overs to secure the victory (a fifth, when the final day was rain affected is excluded). Should he have declared earlier?

In only one of the matches (v England, Cape Town – Jan 2010) does it appear, with the benefit of hindsight, that more overs could have delivered a victory. England were nine wickets down and 170 short of the target after the 141 overs were delivered. In the other matches, the opposition were three, four and five wickets down at the close of play. In each case, bringing forward the declaration by an hour would not have substantially increased the chance of a South African defeat, but may have helped a victory push, although that seems unlikely given the position of the game at the end of day five.

Teams batting third create time to bowl the opposition out by batting, if not aggressively, then enterprisingly. The chart below shows the run rate per over achieved by South Africa for each of the 22 ‘live’ declarations. The colour of the bar indicates the match result: green – win; blue – draw; red – defeat.

G Smith decs 2-page-001

The picture is mixed. Three of the four third innings with run rates below 3 per over presaged draws. But three of the five third innings with the highest run rates were also in matches that concluded in draws.

Smith has earned praise for several of his declarations. The defeat against Australia at Sydney in 2006, when South Africa were 1-0 down in the final Test of the series, came about following a “sporting declaration” by Smith – and tremendous attacking innings by Ponting. His next declaration, against India at Durban in December 2006, showed no signs of being scarred by defeat. Smith set the visitors 354 at a rate of almost one run per over slower than his team had amassed their third innings total. He was rewarded with a comfortable victory and a squared series. At Headingley six years later, Smith was credited with “happy daring” when 1-0 up in the series with two to play, he set England 253 in 39 overs.

Taken together, these observations point to criticism of Smith’s negative approach as a captain when setting the opposition a target being overstated. There are examples where he could have ventured more in pursuit of victory. But there are also examples of bold declarations, just as there are of frustration in the face of stubborn, unanticipated fourth innings salvage jobs.

Where Smith’s team appears to vary from the norm in Test cricket declarations is the failure to convert the majority of situations where 100-150 overs remain in the game into victories. In an earlier analysis of target-setting declarations in Tests between 2009-11, All the time in the world, I found that only four of 17 (24%) declarations made with 100-150 overs left in the game resulted in draws. Smith’s career record is 67% (six of nine).

The key to Smith’s apparently low conversion of declarations into victories has not been the timing of his declarations, nor has it been the urgency with which his side has batted in the third innings of the match. If there is a deficiency it lies with the concoction of factors that have made South Africa relatively ineffective at dismissing sides in the fourth innings. In that mix may be: the lack of top quality spin bowling, unhelpful wickets, unadventurous captaincy in the field and, of course, ill-fortune. What do you think accounts for Smith’s mediocre record of driving home match advantage into victory?

Making up for the mistake

randall run out“That is tragedy for Derek Randall.” (Richie Benaud on commentary)

Geoff Boycott clipped the ball straight and set off for a single. Jeff Thomson intercepted the ball before it passed the stumps on the non-striker’s side. Randall, back in his crease, looked up and saw the striker’s end vacant. When he set off he was racing Thomson’s flicked throw, which Marsh collected and completed the run out.

Randall was making his first Test appearance at his home ground, Trent Bridge. A very popular local man, made heroic by his quirky, marathon 174 in the Centenary Test at Melbourne earlier in the year.

Boycott was playing his first Test in four years. A combination of personal, public, cricketing and domestic issues had kept him away from the international game. Aged 36, he was making his comeback. When he took off for the run that sacrificed Randall, he had yet to make 20. In an unusually public display of contrition – for any cricketer, not just the stereotypically pragmatic Yorkshireman – Boycott covered his face as Randall trotted off the field. What could he do to make up for the mistake? Bat on and keep on batting.

The following day, five hours of batting later, Boycott reached a century, acknowledged in subdued fashion by the still bruised Nottingham crowd. His application took England to a first innings lead of 121 which was to become a victory, and a 2-0 lead in the series with two to play.

Boycott never lacked the incentive to score runs, but he has acknowledged the mix-up with Randall made this an especially challenging and important innings. Is there any generalisable truth from this story? Do batsmen, culprits in their partner’s dismissal, become more focused and committed to playing a telling innings and make up for their mistake?

I have taken as my sample the 117 run outs in Test cricket since January 2011. I further refined the sample (footnote 1) and using Cricinfo’s ball-by-ball commentary identified 21 instances where a top-order batsman was responsible for the run out of a partner and had the opportunity to build an innings.

In that sample are two batsmen, who on the face of it, emulated Boycott by scoring centuries after running out a teammate. However, neither are achievements of the same order. Virat Kohli had scored 65 when his tardily retracted call led to Cheteshwar Pujara’s dismissal. He went on to make 119 against South Africa at Johannesburg in December 2013. Hashim Amla had already recorded his hundred when Ashwell Prince was called for a single that only Amla could see, also at Johannesburg, against Australia in November 2011. Amla lost his wicket a handful of runs later.

The most runs scored by the guilty party following a run out mix-up, in this sample, is 76 by Angelo Matthews against New Zealand at Colombo in November 2012. The distribution of runs scored after contributing to a teammate’s run out is depicted below. The large proportion (48%) who score fewer than ten additional runs suggests that the experience can act to unsettle, not galvanise the surviving batsman.

runouts grouped-page-001

Isolating the effect of the run out on the subsequent innings is impossible. I have, however, attempted to compare the runs that were scored with the runs that might have been expected to have been scored. This involves looking at the runs the surviving batsman had scored at the time of the run out and calculating that batsman’s average score in Test cricket in innings when that score had been reached.

Taking the Angelo Matthews example – the Sri Lankan was on six when Samaraweera was run out. In all Test innings where Matthews has reached six (as at January 2014) his ‘situation average’ score has been 43 (NB this is a calculation of average innings score, not average runs per dismissal as is conventionally used in batting average calculation.) So, Matthews’ total of 84 far surpasses the average that stands as a proxy for what he might have been expected to have scored.

Each of the 21 innings in the sample is plotted below. The runs scored by the surviving batsman after the run out are the blue columns; the difference between the situation average and the total achieved is shown in orange.

runouts&score.pdf-page-001(1)

16 of the 21 batsmen in the sample scored fewer runs after the run out than their ‘situation average’. The average of these 21 innings is a 15 run shortfall from what might have been expected given the runs the surviving batsman had scored at the time of the run out.

I am ready to acknowledge that this analysis does not control for match situation, conditions or opposition strength. Moreover, the sample is small. However, the findings do not encourage me to look further for evidence of batsmen rising to the task of making right their error in causing the dismissal of a teammate. If anything, the data suggest that surviving batsmen are vulnerable, they under-achieve and perhaps even compound the error. In this, as in so many other things, Boycott is atypical.

————————————-

Footnote 1: The initial sample of 171 run outs was reduced by omitting cases where:

  • the surviving batsmen batted at number seven or lower in the order
  • the innings was close to its conclusion and so the surviving batsmen would have had limited opportunity to develop his innings
  • the cricinfo ball by ball commentary placed the blame on the departing batsman for his run out.

Dead rubbers

dead rubbersThe Ashes series is decided. Australia have overwhelmed England at Brisbane, Adelaide and Perth. England’s squad is gathering itself for Melbourne and Sydney depleted by the withdrawal, retirement and impotence of first choice players.

Those of us protective of Test cricket laud the extended narrative of long series. We don’t hesitate to point out the unsuitability of the increasingly frequent two-off Test format that leaves so many contests undecided (two Test series = oxymoron). Yet, with the luxury of lengthy series comes the risk of one-sided contests rendering later matches irrelevant to the series outcome. It’s worth considering whether this lessens the significance and impact of the matches played. Firstly, though, a few numbers to evaluate how common the dead rubber is.

I looked at the progress of the 99 most recently completed series up until September 2013. 43 of these were two Tests in length; 41 lasted three Tests; ten of four Tests; and five of Ashes-length five matches. I extended the sample of four and five match series to 25 by adding the next ten most recent longer contests. It is worth noting an important bias in the data: longer (four/five Test) series are unequally distributed across Test playing nations. England contested 16 of the 25 series in the sample; Australia 11; New Zealand, Sri Lanka, Zimbabwe and Bangladesh 0.

The frequency of dead rubbers is shown in the table below.

Series length Series (no.) Dead rubbers Dead rubber %
3 tests 41 10 24%
4 tests 18 11 61%
5 tests 7 2 29%
4 or 5 tests 25 13 52%

Over half of longer (i.e. four or five) Test series ended with matches that had no bearing on the series outcome, compared to one-quarter of three Test series. Only one series had the ‘double dead rubber’ we are about to experience – England’s last but one visit to Australia in 2006/07.

One step back from the dead rubber in terms of predictability of outcome, is the series which reaches the stage where only one team can win and the other draw (‘2 result rubbers’). This analysis draws in the two-off contests.

Series length Series (no.) 2 result rubbers 2 result %
2 tests 43 37 86%
3 tests 41 19 46%
4 tests 18 11 61%
5 tests 7 5 71%
4 or 5 tests 25 16 64%

The vast majority of two-off contests saw a positive result in the first Test meaning that only one team could win the series. A much lower proportion of three Test series had one team going into the final match with a lead of a single victory.

Many Test rivalries are played out for a trophy. The convention that the most recent winner only has to draw the current series to retain the trophy means that encounters where only one team can win the series and the other draw it have the potential for great excitement. The Oval Test of 2005, with England needing a draw to regain the Ashes and Australia a victory to retain the urn, was an exemplar of this kind from recent years.

Series that enter their final chapter with all three outcomes possible can be the most prized.

Series length Series (no.) All 3 results poss %
2 tests 43 6 14%
3 tests 41 12 29%
4 tests 18 2 11%
5 tests 7 1 14%
4 or 5 tests 25 3 12%

This eventuality is as equally unlikely a culmination to a five Test series as it is to the abbreviated two-off contest. The three Test series most frequently delivers the ‘everything to play for’ finale. Cricket is a confounding sport, so it’s worth recording that not all series that enter that last match with everything at stake should not be cherished. Two high scoring bore-draws can presage an equally turgid third and indecisive Test match.

To consider whether matches in dead rubbers are a dead loss or are fought to the death, I’ve looked at the actions and outcome of the 23 dead rubber matches in the sample.

The majority (13) were won by the team that had already secured the series victory (included here was the match awarded to England in 2006 when Pakistan refused to take the field after being penalised for altering the condition of the ball). While that provides evidence that the winning team remains motivated to pursue the win, it does suggest that the dead rubber adds little to what we already know about the relative strengths of the two teams.

Four of these matches were won by the team that had lost the series. In early 2009, Australia and South Africa played home and away three match series – virtually a six match series across two continents. In an unexpected symmetry, the host lost the first and second matches of both series before recording a home win. The home teams also gained consolation victories in the other examples: England defeating South Africa in 2008 at the Oval in Kevin Pietersen’s first match as captain; and a thrilling victory for India in Mumbai in 1994 when Australia were skittled for 93 and a 13 run defeat.

The remaining six dead rubbers produced draws. There were some notable matches, including: Lara’s 400* for the series losers; Sri Lanka running out of time 101 runs shy of a world record 4th innings target with six wickets in hand. There is also the strong rebuttal to the notion that dead rubbers don’t matter from November 2011 in Mumbai. India, series victors, were set 243 to win the third and final Test. The match went to the last ball of the final over with India’s tenth wicket pair managing a single to bring the scores level.

While dead rubbers have been the occasion of some very notable cricket, on the whole, the sport would be better without them. In this sample, series comprising three Test matches provided the most sustained uncertainty over the series outcome. Nearly half of longer (four and five Test) series, fulfilled the final fixture with honours already awarded. My recommendation would be that three Test series should be the norm. Longer series should be reserved, as they are now, for the traditional marquee series, but also any contest between, say, two of the top three ranked teams in the world. This would require some flexibility in scheduling, something the BCCI has shown in 2013 is very feasible. Two-off contests should be limited to match-ups between teams separated by five or more places in the rankings.

Later this week, the first of two Tests starts with the Ashes already decided and the rubber, strictly, dead. The onus is on England to breathe some life back into the contest.

___________________

Do leave a comment, or tweet me, with your views of preferred Test series lengths and also of memorable dead rubbers.

Catches rhymes with matches

Folkestone's Cheriton Road ground

Folkestone’s Cheriton Road ground

The outfield at Folkestone was bone hard and sun seared. Hot and weary we made our way across it to the pavilion. Tea, the innings break and shade were all welcome. As our fielders funnelled together over the last 20 metres, a teammate spoke at me, over my shoulder: “Catches win matches.”

It was an accusation, not acclaim. Early on I had dropped the opener at second slip. But that had been a good effort. Four runs saved. An over or three later, the same batsman had got a leading edge, sending the ball spiralling up and in my general direction at point. I shuttled to my left, backwards, turned, stretched and got the barest scrape of the fingers of one hand on the ball. I thumped the ground, picked up the ball and flung it to the keeper. A teammate pointed at the sun and nodded. Yes, the sun had got in my eyes, as it inevitably would at some point when you do a pirouette with head tilted skywards on a clear day. The batsman went on to score 80, playing barely another false shot and providing the backbone of the Folkestone 2nd XI total.

The exertions of fielding and bowling had wilted us. We never challenged the total, but took the game deep before losing. I don’t remember my innings, but it must have been brief. Ready for an early night, I was tied to my lift and eventually made it back to London at 11pm.

Contributing little; taking no pleasure in the company of my team; and seeing a whole Saturday pass without reward; that day, 19 years ago, sealed my disenchantment with regular club cricket. I played the last couple of league matches of the season and didn’t return.

Even now, if I hear the phrase, ‘catches win matches’, it triggers uncomfortable associations with that Kent League fixture. Putting aside the discomfort, however, the phrase intrigues me. It sits at the centre of the great unresolved quandary of cricket selection: how does a player’s fielding ability balance against his or her batting and bowling contribution?

An answer (not necessarily ‘the’ answer) is provided in a piece of research, ‘Do catches win matches?’ (1) carried out by Seamus Hogan, economist at the University of Canterbury, New Zealand. Hogan’s work looks at one day internationals and analysed every opportunity for a fielder to make a dismissal in 122 matches, using Cricinfo ball-by-ball commentary. Fielders are scored for their performance. A strong fielder is defined as one with a score one standard deviation above the average. Their contribution is compared to that of the strong batsmen and bowlers – each defined as performing one standard deviation above their respective discipline’s average. The superior fielder is found to contribute less than two runs per innings, well below the equivalent bowler (six) and batsman (eight).

After identifying some caveats to the findings, Hogan concludes:

the “catches win matches” cliche should be put to bed.

Then in response to a comment to his piece, Hogan placed the cliche in its context:

it would also be true that “groundsmen win matches”, “tosses win matches”, “boundaries win matches”, “singles win matches”, etc. I do think there is something about a brilliant catch or a horrible drop that sticks in the mind more than any single cover drive or or even a seaming jaffa that earns an LBW, leading to the importance of catches being overstated in people’s intuition.

And catches happens to rhyme with matches.

Nearly two decades on and 300 miles north-west and I am back in club cricket. Drawn into the world of club administration, I find the work continues after the playing season has finished.

I attended a local club forum last week. The ECB’s club cricket survey results were the headline item. The presenter, from the county cricket board, noted that participation rates stayed constant for players at all ages between 26 and 56. It was in the ten years up to the mid-20s that saw a steady decline as youngsters left cricket. The presenter regretted that the full results of the survey were not yet available, but he was anxious to see the feedback from this crucial age group. Match duration, start time, travel distance, pitch quality, competitiveness of fixtures, umpire reliability and that malleable notion, the spirit of the game, were all independent variables that could be evaluated and changes made to accommodate the game’s younger players.

The discussion took me back to my withdrawal from the game, that hot day in Folkestone, the ball looping out of my reach, the teammate pointing the finger of blame at me and the late return home after a day wholly wasted.

Just as the result of a cricket match cannot be distilled into something as simple as which team takes its catches, the players’ survey results won’t be able to single out just one step that will keep more young men in the game. But there is a factor, in the hands of the players rather than the administrators, that my experience suggests does determine whether members return year after year. It’s not the format of the game, the competition, where or how it is played. The key ingredient is that teammates enjoy each others’ company.

——————————-

Footnote 1: Do Catches win Matches (UPDATED) was published by Seamus Hogan on the Offsetting Behaviour website on January 31, 2013. http://offsettingbehaviour.blogspot.co.nz/2013/01/do-catches-win-matches.html

Neutering the intelligent cricketer

morkelIn cricket, the physically strong can be undone by the weak who are technically gifted; and the clever can prevail over the skilled. It does now, however, appear that the advantage of the intelligent cricketer is being eroded.

That was my reflection on listening to the recent Couch Talk interview with CKM Dhananjai, Performance Analyst with the Indian national team. The interview began like a bad day at work with the interviewee talking about ‘performance enhancement’, ‘SWOT analyses’ and data ‘delivery models’. Subash Jayaraman probed in his courteous way and out came the evidence for there being an active programme to neuter the intelligent cricketer.

Before I substantiate that charge, I will clarify my position on cricket and analytics. Readers of Declaration Game will know that I like to play with numbers, test hypotheses, find associations and contrasts. I do it because I find mainstream coverage of cricket lacking in insight and reliant upon assumptions, cliché and inherited beliefs. I don’t think there’s a secret formula to winning cricket games that can be found if only we conduct enough regression analyses. But I do sense that a sport with so many numbers has done little to understand the probabilities of outcomes for players and teams, and the actions and conditions that affect those probabilities. It’s also harmless fun.

Back on ‘The Couch’, CKN Dhananjai started to give examples of the information he would make available to Indian players.

to play a Morne Morkel, a batsman is already given information about what he does, his instances of bowling a bouncer every three or four balls, and if he is hit for a boundary in a particular ball, what is his follow-up ball, and all that stuff.

This is granular, highly specific information. The technically skilled batsman, capable of absorbing and applying that information, is now on a par when facing Morkel, with the intelligent player, who through his own observation has discerned the pattern in Morkel’s bowling, or perhaps can detect from the South African’s run-up and delivery stride when the bouncer is coming.

Dhananjai’s second example is for the fielding team.

There are many cricketers in the world today who like to hit and run, and we have analytics on that, so you know that if they hit and run, there is an opportunity for a run-out.

The cover point who studies the new batsman’s body language to detect the nervousness that will lead to a poorly judged run has no advantage over the fielder who has listened to the analyst’s briefing and has the ‘hit and run’ batsman pointed out when he arrives at the crease. It’s hard to imagine the creative and cunning tactic of the young Jack Hobbs being tolerated – Hobbs would gift new batsmen a run or two to him in the covers before swooping and running out the complacent batsman.

It’s not just a player’s intelligence that is being neutered in this data-led approach to coaching and match preparation, but individual responsibility; the desire for self-determination that would have a batsman either study a bowler from the pavilion or quiz him over drinks after the game to identify and absorb his opponent’s variations. Those lessons are received passively now in video presentations about the opposition.

It was CKM Dhananjai’s response to the final question of the Couch Talk interview that made me want to distance myself from this analytical approach to the game. He was asked: ‘do you actually get to enjoy a particular game of cricket?’

That’s an interesting question, and a tough one, actually. As a fan… I don’t think I can ever watch a cricket game as a fan, to be honest. There lies the answer. Even if I am watching something on TV sitting at home, it is very difficult to watch it as a normal fan because of the fact that you have been immersed in this day in and day out for more than ten years now.

His dedication to stripping the game down to probabilities and predictive analysis, have left this former cricketer unable to watch the game purely for fun.

What happens when a partnership ends

Jaya & Mahana4 August 1997: Roshan Mahanama walked out to the middle with Sanath Jayasuriya at the start of the third day of the Test.  It was the beginning of the second wicket partnership – the first wicket of the innings had fallen to the last ball of the second day.

5 August 1997: The pair walked out together again on day four, having batted throughout the third day – a very rare occurrence.

6 August 1997: An unprecedented feat – batting unbeaten through two full days of Test cricket – the pair started day five still in partnership. At this point they had amassed 548 runs together, surpassing the previous record for a partnership for any wicket in Test cricket by a clear margin.

Sri Lanka batted through day five to reach 952-6, setting the record for the highest total in Test cricket. Opposing captain, Sachin Tendulkar, was full of praise for the Sri Lankan batsmen, but branded the pitch unfit for Test cricket.

Jayasuriya and Mahanama had taken their partnership to 576 on the fifth morning when the latter was dismissed lbw by Anil Kumble. Two balls later Jayasuriya was out caught at silly-point. After twelve and a half wicket-less hours, India had prised out both batsmen in a matter of minutes.

It is a commonly made observation in cricket that after a long partnership, the two batsmen fall in quick succession. Jayasuriya and Mahanama are the archetypes of this phenomenon.

I find the idea that two well-set batsmen will lose their wickets suddenly, one after the other, intriguing. There are conventional cricket explanations – wickets falling to the new ball; an old ball beginning to reverse; or some other change in conditions. But what interests me is that the idea is suggestive of a range of psychological explanations.

There is the simple notion that the disruption in concentration caused by the first wicket makes the remaining batsman vulnerable. More complex is the possibility that the two batsmen during a long partnership have developed an equilibrium, where each has an understood role and has become secure in their relationship. The loss of one wicket and replacement of the established partner with a new batsmen, upsets that equilibrium, making the surviving partner struggle to adapt to a new role and so prone to dismissal. An even more extreme psychological explanation could be that the surviving batsman is grieving for the loss of the established partner and subconsciously wants to re-join them – analogous to the elderly couples whose deaths follow fast upon the other.

But before I indulge in any more psychology, there’s a more basic question to be asked: is there any truth in this cricketing received wisdom that the end of a lengthy partnership often results in both batsmen falling quickly?

To test the theory, I have taken as a sample all Test partnerships of 200 or more since July 2007. I have recorded whether the surviving batsman was the next, the second, third, etc wicket to fall after the partnership ended; how many runs were scored and overs elapsed between the dismissal of the two partners.

I have also taken a control sample of other partnerships (96) to compare the aftermath of a large partnership with randomly selected partnerships (footnote 1).

There were 98 double-century partnerships in this period, 8 of which were unbeaten and so excluded. The fate of the surviving batsman after a 200+ run partnership ended is shown below. 24% (control: 21%) remained not out. 44% (control: 41%) were the next batsman to be dismissed, which is less than the one in two chance of a random result and lower still than the received wisdom of partners falling in quick succession after a lengthy alliance. The results of the control sample are similar, suggesting that the surviving batsman is not much more likely to be the next wicket to fall after a large partnership than any other partnership.

partnerships1-page-001

How long was the portion of the surviving batsman’s innings that remained? 85% (control: 82%) were still batting four overs after the partnership ended; 62% (control: 62%) when ten overs had elapsed. In terms of runs, 84% (control: 86%) added at least 10 runs; 62% (control: 65%) were still batting with the score 30 runs on. The majority of surviving batsmen are not falling hard on the heels of their erstwhile partner. The length of their stay after a major partnership does not differ significantly from the control sample of partnerships

partnerships2-page-001

These figures point to the received wisdom about partners falling one after the other not being supported by the evidence.

I had sought a psychological explanation for a batsman falling soon after his long-standing partner was dismissed, when I should have been looking for a psychological explanation for cricket viewers believing this to be a commonplace. That phenomenon is known as ‘confirmation bias’ and it comes about because our memory is biased to remember the ‘hits’ (Jayasuriya and Mahanama) but not the ‘misses’.

——————————–

Footnote 1: the control sample was drawn from some of the same matches as those involving the double-century partnerships. If the large partnership was for the first wicket, the control figures were those of the other first wicket partnerships in the match.

Acknowledgement: Gabriel Smith (no.1 son) assisted in the collection of data for this post.

Home advantage and the First Test

clarke and warnerGoing into the first Ashes Test at Nottingham, the challenges facing the Australian team appear considerable: a fragile batting line-up, whose star performer has a chronic back condition; inexperienced and injury-prone bowlers; a recent history of divisions within the playing staff; one player serving a ban for indiscipline; and a team manager installed only days before the start of the series.

I was interested to see whether, based on recent Test history, the first test of a series could present even greater difficulties for Australia as the touring team and whether warm-up matches help a touring team achieve better results.

I selected as my sample every series from (and including) the last Ashes series in England four years ago. That gives a total of 63 bilateral series, including one- and two-off contests and a total of 163 Test matches (footnote 1).

The chart shows that visiting teams performed worse in first (and only) tests than in 2nd – 5th tests. They were more likely to lose the opening match (15 percentage point differential) than they were all subsequent matches in series, with their chances of getting a draw dented slightly more than chances of a victory.

new home advan 1-page-001

I refined that analysis to focus on the top eight teams (i.e. excluding any contests involving Zimbabwe or Bangladesh). The remaining results draw from 48 bilateral series and 138 Test matches.

In contests between Test cricket’s top eight teams the difference between first tests and 2nd-5th tests is more pronounced with the visiting team almost 20 percentage points more likely to lose the opening fixture than the matches that follow.

new home advan 2-page-001

A potential cause of the slow start to series suffered by visiting teams is the packed international calendar that does not allow tourists much time to acclimatise. Going back to 1989 when Australia began their 16 year dominance of the Ashes, the touring team played five first class fixtures, three ODIs and had been in the country for over a month before the first Test of the series. The 2013 Australians have two first class matches, although almost all of the squad were present playing cricket in the Champions Trophy, Australia A tour or in county cricket. This is unusual as very few touring teams have the benefit of an international tournament, domestic cricket or an A team tour to get used to playing conditions ahead of a series.

Of the 48 touring teams in this sample, 46% played no first class cricket prior to the first test of the series. No tourists had more than three first class warm-up matches.

warm up 1-page-001

This distribution of warm-up matches allowed me to test whether there is an association of this kind of preparation with Test match results. Counter-intuitively, teams that had first-class match preparation fared worse in first Tests and overall than those without first-class match preparation. Perhaps all this shows is that first-class matches are only one part of a range of preparatory work that needs to be done before a Test series and too much emphasis can be placed on it. It is also possible that international cricketers do not take the opportunity of first class warm-up matches sufficiently seriously to get their full benefit.

warm up 2-page-001

Should Australia be concerned about the challenge of the initial test as a touring side? Australia’s own record away in this period is strong: 43% (3/7) of first tests and 33% (7/21) of all tests won as the away side. Their players have had a good opportunity to become acclimatised. Given their other difficulties, I don’t think they should be too preoccupied by thoughts of a ‘First Test effect’.

————————————————————————————

Footnote 1: the Pakistan v Australia series of 2010 in England was excluded as both sides were playing away. All Tests involving Pakistan in the UAE are treated as home fixtures for Pakistan.

In at first drop: batting at 3 in ODIs

sangakkaraBatting order or batting situation? I believe cricket pays far too much attention to the former and too little to the latter in deciding whose turn it is to bat. I’ve described it before as a personal hobby-horse.

The number three slot in one day international cricket provides a useful testing ground. It’s the only batsman in the order who faces the uncertainty of starting against the second ball of the innings, or the 300th. And in England, there is a debate of passion and parody about Jonathon Trott’s value to the team at number 3.

With the Champions Trophy upon us, I start with a review of which batsmen have the best records at 3 in the last two years of ODI cricket – i.e. since the last World Cup. Three regulars at number three head the table. Sangakkara is the most prolific, with Kohli and Trott clear of a following pack.

Innings Runs Average
K Sangakkara 40 1606 42.26
V Kohli 27 1173 46.92
J Trott 28 1041 45.26
DM Bravo 13 509 46.27
M Samuels 16 497 33.13
B McCullum 11 421 42.10
M Clarke 10 369 36.90
H Masakadza 11 334 30.36
Asad Shafiq 11 325 36.11
Younis Khan 15 311 22.21
G Gambhir 10 306 34.00

The next chart adds the dimension of scoring rate, plotting it against the batting average for the top number three batsmen.

No3 batsmen ODIs-page-001

Kohli’s record is very impressive as only one of only three batsman with a strike rate over 80 and the highest batting average. Trott is one of the slower scorers, but not by a significant margin. If he quickened to the median rate (76 runs per 100 balls), in an innings of average duration (51 deliveries) Trott would only score an additional two runs. If he matched Kohli’s pace – the ICC’s ODI player of the year in 2012 – he would score an additional six runs in an average innings.

The charge against Trott is that his game isn’t versatile enough. We can get an insight into the situations a number three faces at the outset of an innings by looking at opening partnership totals and durations in matches since the last World Cup.

The average opening stand is 33, and median score at the fall of the first wicket 19. 31% of opening stands end before the total reaches double figures. Over three-fifths are finished before 30 is on the board. The number three has plenty of opportunity to build an innings and shape his team’s total.

fall of 1st wkt runs-page-001

The number three comes to the wicket in the first five overs in 55% of innings and has begun batting before the end of the initial ten over power-play in more than three-quarters of instances. In the majority of situations, therefore, there is an advantage in having the solidity of a conventional top-order player batting at number three.

fall of 1st wkt overs-page-001

Looking at average opening partnership duration by team in this period, there is a significant variance (50%) between England at one extreme (51 deliveries) and India (34 deliveries), the lowest of the major nations. Perhaps this is where some of the criticism of Trott gains a little traction: despite beginning his innings after his team has had relatively good starts, his run rate remains on the low side.

fall of 1st wkt countries-page-001(1)

Is there any evidence that ODI sides are sticking inflexibly to their batting order – at least in terms of the number three? To test this notion, I’ve looked at the 21 instances of the first wicket falling after more than 120 deliveries have been bowled and compared who batted at three in that match with the order in the other matches in that series.

In 12 of the innings, the number three was unchanged compared to other games in series. Brendon McCullum, Sangakkara, DM Bravo, Shane Watson and Trott were amongst the batsmen retained at three even when the innings was well under way when the first wicket fell. Pakistan appeared to be the team most willing to shuffle their order according to circumstance, relegating Younis Khan and Azhar Ali behind the likes of Umar Akmal, Abdul Razzaq and Shahid Afridi. New Zealand also used Jacob Oram to add some vim to a couple of solid starts. Of the five longest opening partnerships, three saw promoted number threes, with Trott once being supplanted by Eoin Morgan.

In conclusion, I recognise that my perception of inflexibly applied batting orders, in the case of number three, is not well supported by the evidence. Teams do shuffle the stodgier ‘threes’ when the opportunity arises. However, with opening partnerships rarely providing the innings with a solid base, the continued presence at number three of players such as Trott is justified.