March 12, 2012

Batsman by bowler / pitch quality: The final grouping

Anantha Narayanan
David Gower scored nearly 99% of his runs in the tougher groups  © Getty Images

I was about three-fourths into this article when the news about Rahul Dravid's retirement came through. After wiping a tear or two (ok, a lot more), I was tempted to replace this article with a tribute to Dravid. It took me only a minute to dismiss that idea. That is not the way the incomparable Dravid would have planned his innings and I was going to take time, look at all nuances, derive additional facts and figures and build the article, brick by brick. That is the fitting tribute we can give to one of the greats. All these cliches found their true meaning when applied to Dravid. In many other cases, these are but hyperbole.

I have finally arrived at the concluding article on this theme. I am going to classify runs scored by batsmen in the composite group comparing of the Bowling Quality and Pitch type, referred to in my last article. I believe that almost all problems present with the earlier Pitch Quality Index have been taken care of. I will do the classification with a brief introduction and will let the readers digest the same.

I have shown one major table and provided a number of tables for readers to unload and view. For a change let me keep one of my articles brief and to the point.

1. The major benefit is that I have avoided the wide variations which occurred when a single Test was considered and negated the impact of one's own bowlers had. Of course when I do the Innings Ratings analysis, I will re-visit the single Test theme and do what is required based on inputs given by Unni, Arjun, Ali, Gerry et al.

2. The problem of double counting has disappeared. When I look at a batsman's score I slot it into a group based on how easy or difficult run-scoring was, in the concerned location, during the specified period. The bowlers do not get into this at all.

3. A country might have started in one manner but completely changed its character over the years. Pakistan was one of the toughest countries to visit during the early years and over the past decade has completely changed. Compare England twenty and thirty years back. All these variations have been taken care of. Pakistan, during the 1946-1950 period had a PTI ratio value 1.23 while Pakistan's PTI ratio value during the period 2005-2012 is 0.74. England's figures for the two successive periods of 1980s and 1990s are 1.06 and 0.97 respectively.

4. There is no assumption that scoring at home is/was easy and scoring away is/was tough. This facet of the analysis is covered based on hard numbers. In New Zealand scoring was very difficult, often more so for the home team than the visiting teams. This is taken care of.

5. Within one country, there is a clear separation of home teams and visiting teams. So there is no adding of 80 (home) and 60 (visiting) in Australia during the 2000s to arrive at an unsatisfactory 70.

6. Within the same country group, batsmen runs are further classified based on the bowling quality. Example, Laxman's 167 in 2000 goes into the combined group 8 (Pitch-3 & BowQ-5) while Laxman's own 148, four years later goes into combined group of 7 (Pitch-3 & BowQ-4).

7. There is no grouping based on absolute values. Rather it is based on a true peer comparison basis and is only a dimension-less ratio. Changes over the various eras will be reflected correctly. This is peer comparison at its best. Same era and across countries.

8. One good point is that many teams starting in Test cricket have had a tough time, even while playing at home. The runs scored by the batsmen from these teams get recognition of the tougher conditions faced by them.

9. Readers can argue that I could have taken 15 time periods instead of 9. Possible. Readers can also argue that I could have taken the top-7 scores. Granted. However there is no end to these suggestions. These periods reflect distinct eras and have sufficient Tests played during each period to have a very sound basis.

10. Finally a quick perusal at the tables will indicate that the sharp differences which existed in the earlier analysis have now gone since the base has moved from a single Test to a period/country combination. There are some intriguing changes. It is now clear that the objections put forward by Unni, Ali and couple of other readers were quite valid. Some players, indeed, benefited by their bowlers' quality, very significantly. Look at the revised tables. The West Indian pitches during 1970-80 were good to bat on, at least for the home team. They were averaging 75.0 (all-teams 67.7) and 72.2 (all-teams 63.3) and batsmen like Richards scored a fair bit of runs at home. And they rarely faced a Group 5 bowling attack. Many thanks to all these readers.

First, the grouping methodology for the Pitch Type Index. The Pitch Type Index is the ratio between the Home/Visiting Top-7 Partnership average for the period/country and the Home/Visiting Top-7 Partnership average for the period/all-countries. A ratio of greater than 1.0 indicates tough situations and ratio below 1.00 indicates easier batting situations. I will not cover this in any greater detail. Details are available in the previous article, link provided here.

PTI-Home groupings: Total - 2022 innings (The 12 neutral Tests have no home teams)
PTI value of 1.15 - 1.50 : PTI Group 5    238 (11.7%)
PTI value of 1.04 - 1.15 : PTI Group 4    432 (21.2%)
PTI value of 0.93 - 1.04 : PTI Group 3    745 (36.6%)
PTI value of 0.88 - 0.93 : PTI Group 2    406 (20.0%)
PTI value of 0.70 - 0.88 : PTI Group 1    201 ( 9.9%)

PTI-Away groupings: Total - 2034 innings
PTI value of 1.09 - 1.50 : PTI Group 5    198 ( 9.7%)
PTI value of 1.06 - 1.09 : PTI Group 4    441 (21.7%)
PTI value of 0.94 - 1.06 : PTI Group 3    731 (35.9%)
PTI value of 0.89 - 0.94 : PTI Group 2    456 (22.4%)
PTI value of 0.70 - 0.89 : PTI Group 1    208 (10.2%)

The working out of these groups has not been rocket science. I have done this based on my pet theory of normal distribution. Approximately 10% at either end, approximately 20% at either next-to-end group and 30% in the middle group. I know this adds to 90, but readers will get the drift. And this varies between Home and Visiting sets in order to get the required distribution.

The Bowling Quality Index runs from 5 (real tough bowling attack) to 1 (very weak bowling attack). The Pitch Type Index runs from 5 (very difficult to score period) to 1 (real batting feast period. The appropriate home/visiting numbers are used. Thus the composite group runs from 10 (runs are to be treasured like Platinum) to 2 (free buffet table, full of runs).

I have decided to present this in only two broad groups. The B group which comprises of the composite groups 2, 3, 4 and 5. Possible combinations of BQI & PTI are 3+2 4+1 2+2 or 3+1 or 1+2 or 1+1. These combinations clearly prove that either of the indices almost never exceed 3 and a 1 was a distinct possibility. These were really the conditions in which it was quite easy to score runs. Compare with the other group called A group which incorporated composite groups 10 to 6. At 6, it was either 5+1 or 4+2 or 3+3. If there was a 1 there was a 5 to compensate for that. Hence these runs were relatively tougher to score.

I have added below, the updated table which will have the A group having the composite groups 10-6 and the B group having the composite groups 5-2. This will ensure that the lowest composite group in the tougher A group is 6 (3+3 or 4+2 or 5+1). No one can now complain that the 6 does not represent relatively difficult conditions for the batsman to play on.

In summary, the alternate table incorporates a switch of composite group 5 from the tougher A to easier B group.

The A (10-6) and B (5-2) broad groups summary for top 40 batsmen

BatsmanCountryInnsNOsRunsAvge (10-6) A Group Summary ====> (5-2) B Group Summary ====>
Inns NOs Runs Avge %C-Runs Inns NOs Runs Avge %C-Runs
Tendulkar S.R Ind311321547055.4515312 7247 51.4046.8%15820 8223 59.5953.2%
Dravid R Ind286321328852.3114312 5698 43.5042.9%14320 7590 61.7157.1%
Ponting R.T Aus276291319653.43125 8 5558 47.5042.1%15121 7638 58.7557.9%
Kallis J.H Saf254391226057.0216419 6998 48.2657.1% 9020 5262 75.1742.9%
Lara B.C Win232 61195352.89178 5 8930 51.6274.7% 54 1 3023 57.0425.3%
Border A.R Aus265441117450.5619334 7529 47.3567.4% 7210 3645 58.7932.6%
Waugh S.R Aus260461092751.0615225 5868 46.2053.7%10821 5059 58.1546.3%
Gavaskar S.M Ind214161012251.12128 6 5518 45.2354.5% 8610 4604 60.5845.5%
Jayawardene M Slk213131008950.44116 7 5376 49.3253.3% 97 6 4713 51.7946.7%
Chanderpaul S Win23437 970949.2817927 7259 47.7674.8% 5510 2450 54.4425.2%
Sangakkara K.C Slk17912 934755.97 95 6 4619 51.9049.4% 84 6 4728 60.6250.6%
Gooch G.A Eng215 6 890042.58183 3 7253 40.2981.5% 32 3 1647 56.7918.5%
Javed Miandad Pak18921 883252.57 95 5 3827 42.5243.3% 9416 5005 64.1756.7%
Inzamam-ul-Haq Pak20022 883049.61123 8 4706 40.9253.3% 7714 4124 65.4646.7%
Laxman V.V.S Ind22534 878145.9711410 4222 40.6048.1%11124 4559 52.4051.9%
Hayden M.L Aus18414 862650.74 80 0 3097 38.7135.9%10414 5529 61.4364.1%
Richards I.V.A Win18212 854050.24101 6 5111 53.8059.8% 81 6 3429 45.7240.2%
Stewart A.J Eng23521 846539.5620217 6909 37.3581.6% 33 4 1556 53.6618.4%
Gower D.I Eng20418 823144.2517614 6962 42.9884.6% 28 4 1269 52.8815.4%
Sehwag V Ind167 6 817850.80 69 1 3183 46.8138.9% 98 5 4995 53.7161.1%
Boycott G Eng19323 811447.7311311 4780 46.8658.9% 8012 3334 49.0341.1%
Sobers G.St.A Win16021 803257.78 69 8 3458 56.6943.1% 9113 4574 58.6456.9%
Waugh M.E Aus20917 802941.8213612 5188 41.8464.6% 73 5 2841 41.7835.4%
Smith G.C Saf16811 776149.43107 6 4197 41.5554.1% 61 5 3564 63.6445.9%
Atherton M.A Eng212 7 772837.70195 6 6830 36.1488.4% 17 1 898 56.1211.6%
Langer J.L Aus18212 769645.27 83 4 3020 38.2339.2% 99 8 4676 51.3860.8%
Cowdrey M.C Eng18815 762444.07115 5 4433 40.3058.1% 7310 3191 50.6541.9%
Greenidge C.G Win18516 755844.72107 4 4222 40.9955.9% 7812 3336 50.5544.1%
Mohammad Yousuf Pak15612 753052.29110 6 4188 40.2755.6% 46 6 3342 83.5544.4%
Taylor M.A Aus18613 752543.50117 9 4465 41.3459.3% 69 4 3060 47.0840.7%
Lloyd C.H Win17514 751546.68102 9 4334 46.6057.7% 73 5 3181 46.7842.3%
Haynes D.L Win20225 748742.30107 4 3925 38.1152.4% 9521 3562 48.1447.6%
Boon D.C Aus19020 742243.6612210 4186 37.3856.4% 6810 3236 55.7943.6%
Kirsten G Saf17615 728945.27109 9 4201 42.0157.6% 67 6 3088 50.6242.4%
Hammond W.R Eng14016 724958.46 27 1 887 34.1212.2%11315 6362 64.9287.8%
Ganguly S.C Ind18817 721242.18106 8 3264 33.3145.3% 82 9 3948 54.0854.7%
Fleming S.P Nzl18910 717240.07146 8 5334 38.6574.4% 43 2 1838 44.8325.6%
Chappell G.S Aus15119 711053.8611814 5177 49.7872.8% 33 5 1933 69.0427.2%
Bradman D.G Aus 8010 699699.94 38 4 3085 90.7444.1% 42 6 3911108.6455.9%
Jayasuriya S.T Slk18814 697340.07110 4 3867 36.4855.5% 7810 3106 45.6844.5%
Flower A Zim11219 479451.55 7813 3191 49.0966.6% 34 6 1603 57.2533.4%

The batsmen to note are the ones who have averaged over 50% in the A group. Not many in this selected list of 40. Tendulkar., Lara, Sangakkara, Richards, Sobers, Bradman, Chappell and Andy Flower (nearly there). Hutton also gets in.

The other factor to look at is the % of runs scored in the A group. Lara, Border, Chanderpaul and a host of English batsmen led by Gooch belong to the list of batsmen who have over 60%.

I started this odyssey about 9 months back. I have had nearly ten articles on the same. The Readers' response has been fantastic. We have kept on improving and I am quite happy with what we have achieved finally. There will obviously be differing views. That is fine. But I am certain that the journey covering this complex area over the past 9 months has been truly worthwhile. I myself have learnt a lot. I am sure the same would apply to most readers. I think I can confidently say that we, as a team, have done justice to the batsmen who faced the twin formidable adversaries, bowlers and pitch conditions, very effectively. Now I will switch to analysing the bowlers.

To download/view the document containing the Innings values of BPI, PTI and BQI, please click/right-click here.

To download/view the revised table containing the A & B group values for all 263 batsmen please click/right-click here.

Arjum Hemnani's request: To download/view the revised table containing the 9 group values for all 263 batsmen please click/right-click here.


Anantha Narayanan has written for ESPNcricinfo and CastrolCricket and worked with a number of companies on their cricket performance ratings-related systems

RSS Feeds: Anantha Narayanan

Keywords: Stats,

© ESPN Sports Media Ltd.

Posted by Waspsting on (April 1, 2012, 12:01 GMT)

re: Ambrose and Mcgrath - I think Ambrose was a more gifted bowler, but McGrath was much smarter.

Ambrose bowled the same way to everyone - back off a lenght, off stump, seam up. He did away with boucners and yorkers far too early.

Mcgrath bowled to the batsman, depending on the batsman. Consistently wide to batsmen who kept playing at them or were likely to have a frustrated wild swipe, closer to the patient player who would let them all go, setting up many players with as series of wide balls, and then the odd one closer. Back off a lenght to the front foot player, with the odd one thrown up, and the opposite for the back foot player. He used yorkers and bouncers though he didn't have the pace for it to be highly effective - its effective in the context of pinpoint accuracy regardless of McGrath's lack of pace.

Little to choose between them, but Ambrose made for more interesting watching since he made the batsmen play all the time.

Posted by shrikanthk on (March 21, 2012, 17:12 GMT)

Possibly many English bowling attacks were quite weak during that time.

The only weak English attacks that Bradman faced were in '46-47 and '48. Though in retrospect I wouldn't class them as too weak either. Bedser and Laker went on to become all-time greats!

Posted by Waspsting on (March 21, 2012, 11:45 GMT)

Seem to have missed a good discussion here. Quick thoughts on the data.

Little surprised to see Jayawardena so highly placed in tough group. My impression of him is he scores exceptionally heavily on the flat tracks of Sri Lanka (whatever the attack). Partticularly surprised to see his record against high and low are so similar. Sangakarra too, to a lesser extent.

Bravo Gary Sobers, in all ways.

Kallis and Hammond's record against the weaker sides - no surprise.

Thought Bradman would average even higher against low group. Average close to 200 against SA and India and West Indies weren't a "weak" attack. What pulled his figure down to only 108? [[ Possibly many English bowling attacks were quite weak during that time. Ananth: ]] Will look through the comments soon and try to offer some feedback.

Posted by Pankaj Joshi on (March 21, 2012, 6:32 GMT)

Dear Ananth, Logically clean as always. What would conclusions be when these same metrics without change are put in two different contexts - runs scored as % of team score and utility of runs where win is given 2 points, draw 1 and loss 0? I admit in advance that the entire onus for win can't be on batsmen, there is also the formality of taking 20 wickets. Pl give it a thought in any case. [[ Would not be worthwhile, Pankaj. In my Innings Ratings work I have a parameter called "Contribution to a team's result". I allot, say, 10 points for a win and 5 points for a draw (suitably adjusted for home/away: += 2 or 1). Then I allocate these points between the batsmen and bowlers equally. Then I allocate the allocated points between batsmen and bowlers based on their contribution, not in terms of runs or wickets but rating points. Then only would it make sense. For instance, for the Perth Test, Warner would get the lion's share of the batting points and Hilfenhaus and Siddle the major share of the bowling points.. Ananth: ]]

Posted by shrikanthk on (March 19, 2012, 17:03 GMT)

I also have to be convinced that this is really needed to solve the problem, if there is one.

Okay. One at a time:

Is there a problem : Yes.

What is the problem: BQIs may be overly influenced by the bowling of part-timers / lesser bowlers in long innings.

Should the batsman deserve credit in such cases : Only if the part-timer was forced to bowl by the batsman. Not if the part-timer bowled because of a strike bowler's injury/resting.

Possible Solution : Consider only the bowlers who bowled the most number of overs. Why 80%? Why not? One way of rationalising it is - In a typical innings of 100 overs, you'd expect your specialist bowlers to bowl 80 of those overs. Hence 80 is a good number.

You talked about a scenario where M Waugh may be Aus' 4th bowler and Warne may be the 5th bowler. Suppose you need only 4 bowlers to cover 80% of overs, don't consider Warne. That's it.

Posted by shrikanthk on (March 19, 2012, 11:53 GMT)

I do not want to implement Shri's 80% concept only to find that it does not work when 4 top bowlers shared the bowling amongst themselves

That's not an issue. My point : Consider the top X bowlers where X is the least number necessary to account for atleast 80% of all overs. So if 4 bowlers have bowled all the overs, while the top 3 have bowled 75%, then X = 4. If 8 bowlers have bowled all the overs, while the top 5 have bowled 81%, then X = 5 If 4 bowlers have bowled all the overs, while the top 3 have bowled 84%, then X = 3

Is this a coding challenge? [[ Merely stating a fact, I can say that there is NOTHING related to Cricket analysis, using my database that I cannot do. However what I have to ensure that anything I do would work in ALL the situations, when Wasim and Waqar ran through a side themselves, when McGrath/Gillespie/Kaspro/Warne bowl all overs amongst themselves in varying shares and when a few other bowlers bowl extra overs and one of them more than Warne et al. I also have to be convinced that this is really needed to solve the problem,, if there is one. And that the 80% has a basis. Ananth: ]]

Posted by arijit on (March 19, 2012, 11:39 GMT)

No, no, no, shrikanthk, I am not saying Ambrose was better than McGrath. Both were great and my personal favourites (along with Marshall) and I do rate them in this order: Marshall, McGrath, Ambrose. I was only saying that if we make a comparison of their performances in Australia, then we should note for fairness' sake that one of them bowled only against Australian batsmen and the other against all the rest. Well, I guess you will not be reading this clarification since Anantha's next post (on Dravid) is already out. [[ From what I know of Shri, he WILL certainly read your comment and respond. Ananth: ]]

Posted by Harwinder Singh on (March 19, 2012, 9:18 GMT)

Hi Anantha, A very good analysis indeed... This analysis is excellent when we see the Averages of players in difficult and easy conditions. But i think it would be unfair if we compare % of runs scored in the A by each batsman, for example: Atherton M.A he has played 195 innings in difficult conditions and only 17 in easy. So, obviously, his % would much higher (89%) as compared to other batsmen(Like SR Tendulkar) who have not played that much number of matches in difficult conditions (158 and 153) as compared to easy conditions. Thanks.

Posted by Pratik on (March 19, 2012, 9:16 GMT)

This is a great attempt, however for me the percentage of runs scored against tougher conditions vs easier conditions does not make much sense. What is a player's fault if he faced more weaker opportunities and exploited it to the hilt. A fairer way to look into things would probably be sum up the runs scored for all the above players in tougher and weaker conditions and come to a ratio A = ( sum of runs in tougher conditions/ total sum of runs scored in all conditions) , B = ( sum of runs in easier conditions/ total sum of runs scored in all conditions) and multiply the runs / average scored in tougher conditions by a batsman by B and easier conditions by A and display and compare that.

This attempts to provide a fair weightage (defined by past players performace itself) to the tougher conditions relative to the easier conditions and hence a composite score can be achieved which gives more benefit to runs scored in tougher conditions compared to easier ones. But recognizes both.

Posted by Gerry_the_Merry on (March 19, 2012, 7:11 GMT)

Ananth, just to correct myself - i meant that (as you have said, unless there are injuries etc.) the bowler weightages should be identical in both innings. The BQI itself should be a function of Team Ist / Team IInd inn bowling averages, where mostly, almost universally, IInd inn averages will be lower than first inn.

The only exception I can think of off the top-of-my-head is the great West Indies teams of the Richards era (1976-92). They would have had a higher second innings batting average than the first inn, since they usually were brilliant at turning the matches around, and at home, lost very few second inning wickets. In fact, in 1983-84 against Australia in WI, they did not lose any second inn wickets at all for the whole series.

Barring that, the team I, team II differential will be anywhere between 10%-30%. I would say a simple across-the-board smear of -5% / +5% on pre-final BQI should do the job, instead of CTD etc. [[ Gerry Since the Dravid article has been posted the comments have started comin in and that will occupy my time. The bottomline is that each of these suggestions make sense, in many cases at micro level. But I have to make sure that these are workable at all levels in all Tests. I do not want to implement some upper level of balls bowled % only to find that it does not work when Waqar and Wasinm ran through a team themselves. I do not want to implement Shri's 80% concept only to find that it does not work when 4 top bowlers shared the bowling amongst themselves. I do not want to implement some past tests bowling indicator to find that when Imran bowled a few overs in a Test he is treated as the equivalent of Balwinder Singh Sandhu. I do not want to take the match as the basis to find that the two innings are chalk and cheese a la Calcutta 2001. So you have to trust me to take care of these. I will however ensure that the real casual bowlers do not have any influence in lowering the quality of bowling and the the lesser bowlers have a limited influence. Ananth: ]]

Comments have now been closed for this article


Anantha Narayanan
Anantha spent the first half of his four-decade working career with corporates like IBM, Shaw Wallace, NCR, Sime Darby and the Spinneys group in IT-related positions. In the second half, he has worked on cricket simulation, ratings, data mining, analysis and writing, amongst other things. He was the creator of the Wisden 100 lists, released in 2001. He has written for ESPNcricinfo and CastrolCricket, and worked extensively with Maruti Motors, Idea Cellular and Castrol on their performance ratings-related systems. He is an armchair connoisseur of most sports. His other passion is tennis, and he thinks Roger Federer is the greatest sportsman to have walked on earth.

All articles by this writer