THE CORDON HOME

BLOGS ARCHIVES
SELECT BLOG
January 27, 2012

Batsman analysis by bowler-pitch quality - part 2

Anantha Narayanan
Ian Botham scored majority of his runs against top-quality bowling attacks  © Getty Images
Enlarge

This is the follow-up article to the one analysing the batsmen performance in conditions related to bowler quality and pitch types. There were a number of very useful suggestions and after a careful study some of these have been implemented. There have been very sound arguments also that there is an element of double-counting and this method, in general, favours batsmen with very good bowling attacks backing them. This point is accepted. However it would be impossible for me to implement these radical suggestions without a lot of work, including quite a bit of validation. Hence I have gone ahead with the current method, modified suitably. The elimination of the double-counting and the development of a single evaluation factor will be done later.

Meanwhile let us look at the current method, which, double-counting notwithstanding, offers many insights. The modifications are summarized below.

1. Take the top-7(or fewer) partnerships rather than individual scores. This was suggested by Arjun again. The partnerships basis might very well end with similar numbers. However it seems to smoothen the outlier/out performer situation. If 320 for 2 was reached through 200, 50, 40 and 30, the 200 seems clearly to be a way-out performance and brings some attendant concerns. However in whichever way the partnerships have been formed, 100/100/120 or 150/20/150, the partnerships clearly convey the comfort feelings for distinct PAIRS of batsmen, rather than single batsman. This will obviate, to a great extent, the need to take off outliers.

3. However I do not want to miss out the low-3 scores, suggested by Anshu, since that normalizes the values. However this time I will take the individual scores since these represent clear failures. But I will be tougher and limit these failures to top-6 rather than top-7 since the top-6, barring stray no.7 guys like Gilchrist, (pre-WC-Win) Dhoni, Vettori et al, represent the real batsmen.

4. I will consider varying numbers for both these measures. Otherwise the impact will go out of proportion. The table is given below.

- Upto 5 innings in match: 3 + 2.
-  6 to 10 innings:        4 + 2.
- 11 to 20 innings:        5 + 3.
- 21 to 29 innings:        6 + 3.
- 30 to 44 innings:        7 + 3.

5. The final Bowler-Pitch index will be derived as a Geometric Mean (GM) of the BQI and RSI values. Both are basically runs. The GM has many benefits. It is ALWAYS a number between the smaller and arithmetic mean values. As the difference between the two numbers increases, it moves closer to the smaller number. And we never get out-of-the-range values.

6. Probably one mistake I made was to combine the first three groups as tough super groups. This set the cat amongst the pigeons. Richards' 82% was way out and readers spent quite some time on that. The first two groups are fine: they are really tough conditions. However the third one is the middle group in which many many runs are scored and should not have been combined with the first two groups. Hence I now have three super groups. The first one is the really tough one (5-4), the second one, the middle group (3) and the third one the easiest to bat against (2-1).

7. I have compiled the values for innings played and the batting averages for each group and have shown these important values, as asked for by many readers.

8. I am not going to give the individual group values. There is too much data. I will give only the summaries by the three super groups.

9. I have done the run-weighted BPI value for each batsman. However I will only show two tables of extreme values of this measure. The similarity of the numbers will warrant some obvious comments.

10. Look at the tough groups numbers carefully. The % of career score only indicates that the batsman faced tough conditions and made some runs on these more often in his career than a more fortunate peer. However an average of above 40 in the revised tough groups is something to sit up and take notice.

11.I have also given a table of selected innings for the top-3 groups. These are not presented as the best innings ever played. However these were made in very tough conditions, bowler-pitch wise. Some of these might be in this table because the bowlers behind the concerned batsmen were outstanding. But they certainly were special innings.

12.Finally I am going to present these tables with minimal comments. My hands are protesting.

1. Player wise distribution table by super groups

BatsmanCtyCareerCareerBatting ToughGrps(5-4) MiddleGrp(3) EasyGrps(2-1)
  InnsRunsAvge InnsRunsAvge% InnsRunsAvge% InnsRunsAvge%
                    
Tendulkar S.R Ind3091543255.71 58168429.5410.9%117552254.1435.8%1348226 69.7153.3%
Dravid R Ind2841326252.63 52155431.7111.7%106392439.6429.6%1267784 74.8558.7%
Ponting R.T Aus2741291552.50 47126628.13 9.8%104461947.1335.8%1237030 68.2554.4%
Kallis J.H Saf2541226057.02 70233737.1019.1% 94391048.8831.9% 906013 83.5149.0%
Lara B.C Win2321195352.89 91346238.9029.0% 76305241.8125.5% 655439 84.9845.5%
Border A.R Aus2651117450.56 90285838.6225.6%107438647.1639.3% 683930 72.7835.2%
Waugh S.R Aus2601092751.06 74220334.9720.2% 94407150.2637.3% 924653 66.4742.6%
Gavaskar S.M Ind2141012251.12 61217738.1921.5% 81345943.7834.2% 724486 72.3544.3%
Jayawardene M Slk2131008950.44 52170734.1416.9% 68262641.0326.0% 935756 66.9357.1%
Chanderpaul S Win234 970949.28 69214735.2022.1% 91350145.4736.1% 744061 68.8341.8%
Sangakkara K.C Slk179 934755.97 36110732.5611.8% 62300551.8132.1% 815235 69.8056.0%
Gooch G.A Eng215 890042.58 97325034.5736.5% 71297143.0633.4% 472679 58.2430.1%
Javed Miandad Pak189 883252.57 63217636.2724.6% 67277848.7431.5% 593878 76.0443.9%
Inzamam-ul-Haq Pak200 883049.61 48141232.0916.0% 82364247.3041.2% 703776 66.2542.8%
Laxman V.V.S Ind223 872846.18 40105827.8412.1% 82289440.1933.2%1014776 60.4654.7%
Hayden M.L Aus184 862650.74 30106435.4712.3% 72297844.4534.5% 824584 62.7953.1%
Richards I.V.A Win182 854050.24 57237043.0927.8% 78354747.9341.5% 472623 63.9830.7%
Stewart A.J Eng235 846539.56 97271630.5232.1% 86326241.2938.5% 522487 54.0729.4%
Gower D.I Eng204 823144.25 85272034.0033.0% 75313245.3938.1% 442379 64.3028.9%
Boycott G Eng193 811447.73 62192533.7723.7% 84402352.9349.6% 472166 58.5426.7%
Sehwag V Ind165 809850.93 30 74024.67 9.1% 49185940.4123.0% 865499 66.2567.9%
Sobers G.St.A Win160 803257.78 43136232.4317.0% 58275754.0634.3% 593913 85.0748.7%
Waugh M.E Aus209 802941.82 62226140.3828.2% 84287936.9135.9% 632889 49.8136.0%
Smith G.C Saf168 776149.43 32101532.7413.1% 60238741.8830.8% 764359 63.1756.2%
Atherton M.A Eng212 772837.70 96255526.8933.1% 79320242.6941.4% 371971 56.3125.5%
Langer J.L Aus182 769645.27 44 95321.6612.4% 63317254.6941.2% 753571 52.5146.4%
Cowdrey M.C Eng188 762444.07 73230633.4230.2% 79339248.4644.5% 361926 56.6525.3%
Greenidge C.G Win185 755844.72 58182432.0024.1% 77323946.2742.9% 502495 59.4033.0%
Mohammad Yousuf Pak156 753052.29 42117530.1315.6% 52194138.0625.8% 624414 81.7458.6%
Taylor M.A Aus186 752543.50 53150930.8020.1% 73294143.2539.1% 603075 54.9140.9%
Lloyd C.H Win175 751546.68 53196339.2626.1% 68277246.9836.9% 542780 53.4637.0%
Haynes D.L Win202 748742.30 66224235.5929.9% 85286039.7238.2% 512385 56.7931.9%
Boon D.C Aus190 742243.66 54174734.9423.5% 69222235.8429.9% 673453 59.5346.5%
Kirsten G Saf176 728945.27 62151226.0720.7% 65280246.7038.4% 492975 69.1940.8%
Hammond W.R Eng140 724958.46 14 31926.58 4.4% 33121439.1616.7% 935716 70.5778.9%
Ganguly S.C Ind188 721242.18 39 93727.5613.0% 73258538.0135.8% 763690 53.4851.2%
Fleming S.P Nzl189 717240.07 64147123.7320.5% 83375748.1752.4% 421944 49.8527.1%
Chappell G.S Aus151 711053.86 59203738.4328.6% 54246152.3634.6% 382612 81.6236.7%
Bradman D.G Aus 80 699699.94 11 53753.70 7.7% 23227598.9132.5% 464184113.0859.8%
Flower A Zim112 479451.55 47126330.0726.3% 35131345.2827.4% 302218100.8246.3%


I am not going to do too much elaboration but will allow the readers to do their own interpretations. Now that the innings and batting averages are shown some points will be obvious.

1. It can be seen that batsmen like Botham, Wood, May et al have the tough super group % as 40+. However it can also be seen that they have all played around 50% of their innings in these groups.
2. There was a comment that batsmen from same teams had similar numbers. This is effectively disproved now. Ponting is 9.8%, Mark Waugh is 28.2%, Martyn 23.4%, Clarke 16.5% and Gilchrist 16.9%. Haynes 29.9%, Greenidge 24.1% and Lloyd 26.1%. Jayawardene is 16.9% and Sangakkara is 11.8%, with similar averages. And so on.
3. Clarke and Gilchrist have tough group averages exceeding 40. The best is McDonald/Bradman with 53+ and R.Mclean with 50+.
4. Kumble's tough group % is 22+ and Dravid's 11.7%. This does not mean anything. However the averages are 13.4 and 31.7. So read and interpret these numbers with care.
5. Slice and dice in whichever way you want to, Hammond props up the tables.

2. Top batsmen by run-weighted BQI values

BatsmanCtyCareerBattingWeighted
  RunsAvgeBPI
McLean R.A Saf 212030.2941.2
Ramprakash M.R Eng 235027.3341.5
Howarth G.P Nzl 253132.4542.0
McDonald C.C Aus 310739.3343.0
Wood G.M Aus 337431.8343.0
Waite J.H.B Saf 240530.4443.2
Botham I.T Eng 520033.5543.4
Bailey T.E Eng 229029.7443.5
Hughes K.J Aus 441537.4243.6
Greig A.W Eng 359940.4443.6
Hudson A.C Saf 200733.4543.6
Benaud R Aus 220124.4643.7
Goddard T.L Saf 251634.4743.8
Wasim Raja Pak 282136.1743.8
Coney J.V Nzl 266837.5843.8
Gregory S.E Aus 228224.5443.9
Randall D.W Eng 247033.3844.0
May P.B.H Eng 453746.7744.1
Cronje W.J Saf 371436.4144.1
Rhodes J.N Saf 253235.6644.1


These are the top batsmen based on the run-weighted BPI values. McLean and Ramprakash lead the field. Most batsmen are in the 1950s-80s period.No modern batsman figures in the top-20. No sub-continental batsman is in the top-20. They all have 45+ values. A few unfancied batsmen like Ramprakash, Wood, Hughes, Howarth have made into this list.

3. Bottom batsmen by run-weighted BQI values

BatsmanCtyCareerBattingWeighted
  RunsAvgeBPI
     
Cook A.N Eng 587648.9752.9
Trott I.J.L Eng 203156.4253.4
Ponsford W.H Aus 212248.2353.6
Jones A.H Nzl 292244.2753.8
Washbrook C Eng 256942.8254.1
Headley G.A Win 219060.8354.2
Hendren E.H Eng 352547.6454.2
Edrich W.J Eng 244040.0055.2
Hammond W.R Eng 724958.4655.4
Ames L.E.G Eng 243440.5757.2

This is the other end. Many modern batsmen and batsmen from the 1920s figure here. Somehow Ames has managed to push Hammond off the last place. Cook and Trott are the leading batsmen of today who have found their place here. Most of today's top batsmen are around the 50 mark.

4. Selected innings which crossed the BPI zone of excellence

MtNoYearBatsmanForVsBatPosRunsBPI GrpBPIResult
          
7881976Amiss D.L EngIndOP179 534.9Won
231886Shrewsbury A EngAus3164 535.0Won
3611952Endean W.R SafAus3162 535.0Won
3311951Simpson R.T EngAus3156 533.6Won
11711991Gooch G.A EngWinOP154 532.4Won
451895Graham H AusEng5105 529.1Won
9151981Hughes K.J AusWin5100 528.9Won
8271978Sadiq Mohammad PakEng2 97 526.6Drawn
451895Trott A.E AusEng9 85 529.1Won
8901980Wasim Raja PakWin6 77 529.1Drawn
10561986Greenidge C.G WinPak1 75 526.1Won
17732005Lara B.C WinAus4226 440.6Lost
14511999Lara B.C WinAus4213 441.4Won
10701987Greenidge C.G WinNzlOP213 440.8Won
2581937Bradman D.G AusEng3212 441.6Won
421894Gregory S.E AusEng6201 438.8Lost
5931965Edrich J.H EngNzlOP310 349.2Won
2361934Bradman D.G AusEng5304 348.2Drawn
16412003Fleming S.P NzlSlk3274 347.3Drawn
6711970Pollock R.G SafAus4274 347.5Won
16972004Dravid R IndPak3270 348.7Won
2571937Bradman D.G AusEng7270 344.2Won
17432005Younis Khan PakInd3267 348.3Won
13581997Young B.A NzlSlkOP267 349.2Won
12711994Houghton D.LZimSlk4266 347.8Drawn
18002006Fleming S.P NzlSaf3262 349.0Drawn
17162004Jayasuriya S.T SlkPakOP253 343.8Won
8451979Bacchus S.F.A.FWinInd2250 349.6Drawn


I am sure there would be adverse comments on this table. Individual innings will be commented upon saying they do not belong here. Possibly they do not. However these were very good innings played, whose inclusion here is based on the parameters set. Any list which includes Gooch's 154, Lara's 223/213, Hughes' 100, Bradman's 212/270/304, Greenidge's 213, Graeme Pollock's 274, Dravid's 270, Jayasuriya's 253 et al cannot really be a bad table. These innings would fill almost anyone's list of top-25 or so innings. The selection criteria is a composite one involving Runs, BPI group and BPI value.

To download/view the document containing the Bowler-Pitch-Index values for 7340 innings, please click/right-click here.

To download/view the document containing the Player tables for selected 261 batsmen tables please click/right-click here.

RELATED LINKS

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 Gerry_the_Merry on (February 4, 2012, 7:39 GMT)

Dear Ananth, So as I understand from this, the ha_wt is influencing BPI to get to run-weighted BPI, but is not influencing the BPI for the purpose of the tough group classification. BPI is nice to know, but the real deal is the tough group runs definition.

I will not press this point further now. I have weighted for 8 months to see the tough group calculations. This edition has many many improvements over the May 2011 edition, but seems not to have team Ist / IInd. Perhaps your next edition will include this.

If I am wrong, and your tough group definition already includes this, a thousand apologies, and I am very happy.

Posted by Gerry_the_Merry on (February 4, 2012, 6:20 GMT)

Dear Ananth, You had mentioned that the run-weighted BPI incorporates Team Ist / Tem IInd delta. I wish to clarify from you if the BQI incorporates the team Ist / team IInd differentials, which as we found in one of your previous articles, was as high as 14% for top 7 batsmen. My apologies if i am raising it too often, but as of now I am unclear if this is incorporated in BQI, and hence if the tough group figures reflect this factor. Would be nice if you dropped a couple of lines explaining if you have done this, and if yes, how and if no, why not. [[ The code is given below. Pl make what you can of it.

if (inns ==1 || inns==2) ha_wt=0.95;
else                     ha_wt=1.05;
bpi=mat[k]->m.bpi[bowinns];
jj=mat[k]->pd[inns][j].player;
psum[jj]->pb_wted_runs+=ha_wt*bpi*mat[k]->pd[inns][j].batruns;
Ananth: ]]

Posted by Vikram on (February 1, 2012, 12:43 GMT)

Hi Ananth, I have a couple of long flights coming up and wanted to see if I could get some data to play around with. While you have provided the BowlerPitchIndex, it doesn't have the names of the players who played in it. What I am asking for is cross table with players on Y axis and matchId on the X axis. That way I can do a more detailed analysis of each player against teams or on a time scale. For example, performance of batsmen against West Indies from 1976 to 1988, or Tendulkar across different time ranges. Let me know if that's possible. [[ Vikram I am sorry I copuld not reply earlier. I was caught up on my next article. I have something in the pipeline on what you asked for. Ananth: ]]

Posted by Ravi on (February 1, 2012, 10:18 GMT)

Ananth, a very good article. Kim Hughes, Botham, G Vishwanath, May et al indeed deserve their high places. I have 2 questions. We saw in one of your recent articles how well MPVaughan has done against top attacks. Why does he suffer in this article? Where do Umrigar, Ranji, Duleepji, Trumper, Hobbs, sutcliffe stand? I think we should look at the batsman’s peer comparison (runs & avg compared to exact contemporaries). This will help rationalise batsmen with fewer aggregate but majority tough group runs (K Hughes etc- no disrespect for Hughes). This will also give credit to long careers (esp those straddling hard and easy bowling eras) and therefore high aggregate runs, Bradman, SRT, Sunny, Ponting, Waugh, Javed, Gooch etc. A big ask. What do you think? [[ Ravi, Vaughan's figures have not been as good, say, Atherton's, Buth then they belong to different generations. However his figures are better than KP and Strauss. The peer comparison has been done in an earlier article. http://blogs.espncricinfo.com/itfigures/archives/2009/08/test_batsmen_peer_players_comp.php http://blogs.espncricinfo.com/itfigures/archives/2009/08/following_up_on_the_test_batsm.php However these are over 2 years old and I may re-do these sometime now. Ananth: ]]

Posted by Ananth on (January 31, 2012, 11:57 GMT)

I have completed some nice weighted batting averages. These have all been included in the PlayerGroup Excel sheet and the same has been uploaded. Please download and check these. The following two calculations have been done. 1. The first is a batting average weighted by (dismissed) innings for the three groups with the middle group as the base. All values are batting averages. The final value will be slightly lower than the career batting average since the first (tough) group is always be going to be less productive than the easy group. This is very indicative of the way the batsmen performed in difficult conditions. 2. This alternative is very interesting, since this takes away the concept of groups, not liked by some because of the seemingly arbitrary fixing of the same, altogether. I have used the Run-weighted BPI and the mean BPI to adjust the batting average. This adjustment is minimal and is probably the best adjustment of the lot. There are no abrupt changes nor is there any arbitrary work. My thanks to Anshu who has done bulk of the work on these.

Posted by Rahul Patidar on (January 31, 2012, 9:49 GMT)

That would be great Anantha!!! Do you mind sharing any single source of match scorecards in text format?

Posted by AD on (January 31, 2012, 9:38 GMT)

Anantha, As it stands: 1) If we change the goal-post (arbitrary marker) between the "Tough" and "middle" groups we can have any of 3 results: 1) No change in figures of the batsmen 2)Slight change 3)Considerable change. It should be relatively easy to move this goal-post 10/20/30% to either side and check which of the above 3 holds. [[ AD Pl see my recent comment and the uploaded file. First, the group split is not as arbitrary as you make it to be. I have looked for normal distribution. Anyhow the effect is minimized when the weighting takes place. Finally the alternate method completely ignores the group aplit. Ananth: ]] 2) Even at a casual glance at the above table there are over a dozen batsmen with better figures than the modern day Indian batsmen in the tough group. To pick any one particular West Indian batsman and compare to one particular Indian batsman to confirm some pre-existing hypothesis would strike most people as odd.This is also what I pointed out to Regi. [[ I repeat again, THERE IS NO PRE-EXISTING HYPOTHESIS. I do the analysis and post these results. That is all. Anyhow who is to say that the 1980s West indian batsmen are inferior to the current Indian batsmen. For that matter who is to say the the 1980s Indian batsmen were inferior to the current Indian batsmen. The top two places in the Career runs table are occupied by two current indian batsmen. However these two occupy the 14th and 22nd positions in the Batting average table. Finally let me say one more thing. I AM NOT DOING ANY COMPARING. I have done some useless work and posted the tables. The readers can and should interpret these in their own ways. Ananth: ]]

Posted by Aditya Nath Jha on (January 31, 2012, 5:45 GMT)

i checked on the entire list of 261 batsmen - the weightages don't change!

Posted by Gerry_the_Merry on (January 31, 2012, 5:29 GMT)

I think reasons why Richards and Lloyd feature in all tables under composite average is that 1) the WI '80s team was not a high scoring team, with usually 2-3 batsmen generally happy to play bailout / anchors rather than the whole team firing a la Aus in the 1999-2004 period and 2) the bowling was murderous.

A simpler method would be to take the ratio of simple batting average / runs weighted BPI and sort on this index. Am not allowed to download files at work, must try at home.

Posted by Aditya Nath Jha on (January 31, 2012, 4:40 GMT)

yes, anantha - i worked on table 1 (39 highest run getters plus flower). let me work it out for the entire 261 batsmen and see if the weightages change significantly (i don't think they will).

Comments have now been closed for this article

ABOUT THE AUTHOR

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