Test series analysis - beyond the scorecard
Summarised features of the Team performance analysis
Towards end of 2011 I had done an analysis of team performances in series. I had developed formulae, from scratch, to assign team performance points based on results and margins of wins/losses. My view is that I got the result matches fairly correct. However the draws were allocated on simplistic methods and were found wanting. There is no doubt that draws are very complicated to evaluate and play an important part in evaluation of team performances. It is significant that a third of the Test matches played have ended in draws.
Maybe inspired by this analysis, Milind and I embarked on our ambitious Test Performance Analysis project. We adopted a totally top-down approach and took our time at each level. It took more than six months and the results have passed all our stringent tests. We feel it is time to unveil part of the work. How we handle the complete set of tables is something we are yet to decide on.
The contribution determination is structured in the following manner.
Match >>> Teams >>> Innings >>> Functions >>> Players.
In this article I will cover the team performances in Test series using the data obtained from the Test Performance Analysis project. The first two levels are covered. These two levels cover about 10% of the total project contents.
Graphical representation of Test Contribution project
The player performance points can be combined at career, innings, function, Test, series and year levels to derive various tables.
First, let me give some important facts on the Test Performance Analysis project. The complexity level of the project is such that when I look at the code after a gap of about a few weeks, I take a day to get myself back on rails. So it is obvious that not everything will be clear to start with. We may also not be able to provide all clarifications.
Salient facts on Test Performance Analysis project: Top two levels
- These are non-contextual. Home/away, team strengths, period, series status etc. are not part of the equations. These are based solely on scorecard data, and nothing more. As such, these are not ratings.
- The cornerstone of these computations is the fact that the two tied matches are the only ones that will be accorded 50-50 performance points to the two teams. Everything else flows from this base.
- Wicket resources are calculated based on an in-depth analysis of the 2100+ Tests that have been played.
- There are draws and draws. Not all draws are equal. Some draws are more equal than others.
- All non-tied draws will be allotted points below 100. How close to 100 will be determined by how close to a result the match is at.
- The match is extended into the fourth dimension, so to speak, to get a handle on the nature of draw and points allotted.
- The total points for a drawn match depend on the extent of completion of the match. It could range from 0.38 points (Test #1907) to 98.36 points (Test #1420).
- A team that draws a match would always score less than 75 points depending on the scores. The highest points acquisition for a drawn match has been by South Africa, with 74.5 points, in Test #616 in which they missed a win very narrowly. Aus: 143, Saf: 332 and Aus: 148/8. Test #1420, mentioned above, follows closely.
- As already mentioned, only the teams in the two tied matches get equal points. Even among these two, the 1960-61 tie is the perfect one since all resources were exhausted. We cannot play a single ball more in any innings, in any dimension we choose.
- The 1986 tie is an imperfect one since Australia lost only 12 wickets.
- The Mumbai Test during 2011 was a draw with scores level. India get slightly more points since they had a very good chance of winning if the match continued for a few more balls. Also, the fact that they could not lose, come what may. The detractors of R Ashwin can change their mind and now pat him on the back for having the intelligence to play for this situation in the penultimate ball.
- All matches that ended in results will have 100 points allocated, to be shared between the two teams, based on a set of complicated formulae. There are two exceptions.
- The first is the contrived result based on the agreement between Nasser Hussain and Hansie Cronje. Only two innings were completed and the two teams share only 50 points (Test #1483).
- The other is the Test conceded by Pakistan because of ball-tampering allegations (Test #1814). This is worked out based on the condition at the time of the abrupt conclusion of the match. The situation then was favourable to Pakistan. This is the only Test in which the "winning" team gets fewer points than the "losing" team. For all practical purposes the "win" is just not there.
- Not all identical wins are equal. A 100-run win with scores of Abc: 150, Xyz: 70, Abc: 100 and Xyz: 80 will be considered to be far more emphatic than another 100-run win with scores of Abc: 250, Xyz: 200, Abc: 300 and Xyz: 250. The reason is obvious. The win margin of 100 is a much higher percentage of the winning team run totals: 250-40% and 550-18%.
- Similarly, two wins by innings and 50 runs could have different team performance points for the winning teams depending on their single innings scores. A win with scores of Abc: 80, Xyz: 200 and Abc: 70 will be considered to be far more emphatic than another similar margin win with scores of Abc: 250, Xyz: 500 and Abc: 200. The reason is obvious. The win margin is a much higher percentage of the single innings score of the winning team: 200-50% and 500-20%.
- It is not possible to predict a range for a win by runs since it depends on the scores and the final margin. The margins have varied between a one-run win (Test #1210-scored 50.08-49.92 for West Indies) to 675-run win (Test #176-scored 92.82-7.18 for England).
- A team that wins by runs could easily score more points than even a big innings win depending on scores: Eng: 85.95 pts (Test #47-Win by 288 runs) & 83.56 pts (Test #48-Win by inns & 197 runs). In fact the maximum points in a match as has been secured by England who won Test #176 by 675 runs, the target being 742. This has been referred to above.
- A team that achieves an innings win will always score more than 75 points since the winning team always played only one innings.
- A team that wins by wickets will always score below 75 points since the winning teams have always played two innings.
Readers might argue that the home/away data is available in the scorecard and can be applied. Very true. However home/away information means very little if we do not have the team strengths, which is an off-scorecard measure. Most teams are weak in the early days and hence cannon fodder to the established teams. South Africa in the 1890s, New Zealand in the 1920s, India in the 1930s, the early Bangladesh teams were all awfully weak. No great achievement in defeating these teams.
On top of that the winning teams should not be given any additional credit. Of course the English win over India in 2012 was a great achievement and stakes claims for additional credit. But I do not want to look at team strengths at this point.
Regarding series status, it is my firm belief that there is no such thing as a dead rubber. Try telling that to the hapless captains of the losing teams during the last Tests of the recent five whitewashes. Through the years, pride and careers were at stake. Of late money and rankings are also at stake. No team ever takes a single match lightly. Before the Perth ODI, Alastair Cook talked about possibly giving up the captaincy. Once the Perth match was won, he suddenly talked about leading until the World Cup. Let no one get fooled by the talk of dead rubbers.
I think Australia made a mistake, since they do not seem to be averse to the rankings system, by resting five key players before the Perth match when the #1 position was at stake, especially as they could very well have lost the Adelaide ODI. And, on this, I am not going to be unduly influenced by the player rotation policies of a team or two.
As I normally do, I have to set some qualification criteria. In this case it is simple. I will consider only series in which three or more Tests were played as genuine series. Out of the 658 series which have been played to date, 456 series qualify. I understand that many series involving Sri Lanka, Zimbabwe and Bangladesh will be excluded. But I cannot really include these since I am averaging the Performance points and without the lower limit of three, the tables will be diluted.
Anyhow, in the new regime, the "single A+ class and two A- class" teams, in a new imperial avatar, will play five-Test home and away series amongst themselves and two (even one)-Test take-it-or-leave-it filler-contests with the "C class" teams, as and when they see fit, probably once every five years, if the "C Class" teams do not unite now and stay united in the future, and avoid the temptation to grab the crumbs (read IPL) that are being thrown at them. (I wanted to see whether I could do this in a single sentence and can say to myself "well done".)
Any reader who has survived this long introduction deserves a pat or two on the back, and as a reward, let us now move on to the tables. I will be posting a summary table listing the most comprehensive series amongst these 456 Test series. I will also post tables listing the most comprehensive of home series and away series wins. Finally, I will post tables listing the most comprehensive of three-Test series, four-Test series and five/six-Test series wins.
The most one-sided series of all time was the three-Test series held in South Africa in 1895-96 against England's second string team, but with George Lohmann in the ranks. England won the series 82.2-17.8 and the matches by 288 runs, innings and 197 and 33 runs in low-scoring matches. The second series is recent vintage: the 2007 home series for Sri Lanka against Bangladesh, won 81.7-18.3. Wins by innings and 234, 90 and 193 runs meant a wipeout nearly as good as the one 110 years back.
The next one had Sri Lanka at the receiving end, when they travelled to India in 1994 and lost 22.2-77.8. India won by innings and 119, 95 and 17 runs. Sri Lanka have another similar dominating series, against Zimbabwe, during 2001. As expected this table is dominated by three-Test series.
An interesting series is the 2002 one between Australia and Pakistan played in Sri Lanka and the UAE. Australia won by 41 runs, innings and 198 and 20 runs. This included the shocking Pakistan collapses for 59 and 53. Australia won 72.9-27.1
This table lists the home series wins by wide margins. We have already talked about the two Sri Lanka series, one on either side. The third entry in this table is England's home defeat of West Indies, playing their first Test series, by margins of innings and 58, 30 and 71 runs, leading to a 77.3-22.7 series win. The 2011 whitewash of India by England just about gets in, at the 20th place.
Australia's 1931-32 demolition of South Africa by margins of innings and 163, 155 and 72 runs and by 169 runs and ten wickets, and series margin of 76.5-23.5, is in comfortable top-six positions in both tables.
Now let us see the top away series wins of all time. We have already discussed the 1896 South African home series loss and the 2002 UAE win for Australia (classified as away series for both teams, for this table only). The home loss sandwiching these two is the 1962-63 series between New Zealand and England. England drew the Ashes series quite comfortably by sharing two wins and came to New Zealand in good spirits. They found New Zealand to be a much weaker opponent and took them to the cleaners by 74.6-25.4: wins by innings and 215 and 47 runs and seven wickets.
Two series are of interest, both involving West Indies. First they toured India in 1958-59 and rolled over India by a margin of 69.1-29.2. At least India managed to draw two Tests, one by the skin of their teeth. West Indies toured England 25 years later and vanquished them 5-0 and 71.8-21.2. England and Australia have suffered heavy home losses. Also of interest is the fact that many losing teams have won a Test.
Since the previous tables were dominated by three-Test series, I have now prepared a set of tables based on the number of Tests played. First, the three-Test series.
All the top Test series have already been covered. Only pertinent comment is that there are couple of series in which the losing team has managed to draw a match. No series which has ended in a 2-1 scoreline features here. It is understandable since a lost Test means a loss of many performance points and it is difficult to make it up in the other two Tests.
These are the four-Test series. Finally, we come to something we can identify with. The most devastating four-Test series loss was India's rout by England during 2011. England won that series by a margin of 71.2-28.8 (wins by 196 and 319 runs and innings and 242 and 8 runs). India's loss to Australia during 2011-12 was by a margin of 69.4-30.6 but almost carbon-copied by India when Australia visited India during 2013 and India won by 67.8-32.2.
England's 3-1 win over Pakistan during 2010 is in second place with a winning margin of 69.6-30.4. England, with three huge wins was able to compensate for Pakistan's narrow win.
Finally the cream of Test cricket: the five/six-Test series. The one nice change from the previous tables is that there are a number of drawn matches in the picture. Margins of 3-0 and 4-0 wins are spread amongst this list.
The 1983-84 series between West Indies and Australia is an interesting one. The first two Tests were drawn but West Indies dominated both by 70-24 margins. Then West Indies won the next three Tests by huge margins winning the series by a 73.0-24.7 margin. The year 1984 also saw a 5-0 whitewash of England by West Indies, England losing the series 28.2-71.8. Let us salute the magnificence of West Indies. Between March 2, 1984 and August 14, West Indies faced the two top teams of the world at home and away in ten Tests and won 8-2-0 and 141.8-52.9. West Indies, please show some of those qualities now, off the field and do not meekly succumb to the money being offered.
Then comes the 1959 disaster of the Indian team in England. But for some good fight shown by Abbas Ali Baig and Polly Umrigar in the fourth Test, this 5-0 defeat might have finished on top. As it is England won the series 72.5-27.5. Considering that India lost the 2011 series 71.2-28.8, it is almost certain that the 2011 defeat was far more devastating since India was the #1 team in the world and possessed no less than 8 world class players.
Australia's recent domination is confirmed by the fact that the three series, played during the past 15 years, which figure in this list are all Australia's 5-0 trouncings of West Indies and England, including the most recent Ashes series. But it must be accepted that most of the recent five-Test series are between Australia and England. South Africa, No.1 team in the world, are offered a two-Test series by India, on a take-it-or-leave-it basis.
Finally, an interesting bonus. The most evenly conducted Test series was when West Indies played Pakistan at home in 2000 (Series #428). After a two-innings draw with a slight edge to Pakistan and an exciting but very close draw, West Indies won the third Test by one wicket to just about wipe out the slight advantage Pakistan had. The series ended 39.0-38.9 for West Indies, a wafer-thin gap between the two teams. The third Test was pegged at 50.6-49.4 for West Indies!
However, the closer series was the five-Test encounter, played between West Indies and New Zealand during 1972. All five Tests were drawn. In the first and the last Tests West Indies missed wins. However the middle three Tests had New Zealand at advantageous positions and the series ended 41.6-41.4 in favour of New Zealand.
To download/view the complete table of all qualifying series, please CLICK HERE. My take is that many of the questions can be answered if you download this file, extract the component files and view the contents. Instead of asking me obvious questions for which the answers are already there in the tables, you could download the file and view the tables.
I have just about completed the top levels of the ODI matches and will come out with an article on ODIs. However, the problem is that there are so many tournaments, at last count approaching 200, and I have to think of a way to handle these.
Anantha Narayanan has written for ESPNcricinfo and CastrolCricket and worked with a number of companies on their cricket performance ratings-related systems