Trivia - batting August 13, 2008

Richards the king, Tendulkar his heir

The ODI batting average is a single-dimensional measure incorporating only one part of the total measure needed to measure a batsman, viz, runs scored

In view of the huge number of comments I have to resort to providing a generic response to the comments. This has been shown at the end of the article.

Martin Crowe, who needs no introduction, had sent in a very valuable comment and Kartik had requested whether Martin could comment on the changes in ODI game over the years. Martin kindly responded to this request and his perceptive comments are shown below. Martin, thanks a ton.

The basic fundamental of the change has been change itself - in the rules. The skill level over all generations has always been constant and consistent. But the framework of each era is determined by the rules of the day. E.g in general there were no 15 over restrictions or powerplays in the 80s, and rules for boundary length was determined by size of ground (inside fence) - but now its a standard 65m. A few examples - We started the exploitation of rules with Greatbatch as pinch hitter opening the batting, Patel opening the bowling Sri Lanka took it a step further in 96'. Personally, I used to practice chipping the ball 45-50m over the inner ring and way short of the boundary rider standing at 80-90m, to score 2 runs. I only ever used to attempt hitting a six over a fielder if there was a short boundary like at Eden Pk, 50-60m square of the wicket otherwise I never tried to clear a fielder 80-90m away. i.e. hitting 6's is so much easier now with standard length boundaries of 65m in place. In this aspect alone, there is a major difference between scoring 250 and 300. And yes the bats are bigger and lighter, but not in my opinion necessarily better for Tests. In the 80's I used a bat weighing 2'4"-2'6" to combat the 4 prong pace attack of the Windies.

With each change or addition to the rules brings an evolution of playing strategy, mainly in batting but also captaincy. T20 will only further encourage the evolution. In 5 years ODIs will be 4 x 20 overs each. In summary the rules will continue to evolve to excite the fan. As it should be. Test cricket on the other hand will rightly be left alone.

Martin Crowe

The ODI bowling average is a fantastic measure since it incorporates the two key components needed to measure a bowler's performance, viz, strike-rate and economy-rate, as shown in the following equation. If either of the economy-rate or strike-rate goes up the bowling average goes up and vice versa.

                  Runs scored off
Bowling Average = ----------------       & can be rewritten as
Wickets captured

Runs scored off Balls bowled Bowling Average = --------------- x --------------- Balls bowled Wickets captured Hence

Bowling Average = Economy Rate (R/B) x Bowling Strike Rate (B/W)

Unfortunately the batting average is a single-dimensional measure incorporating only one part of the total measure needed to measure a batsman, viz, runs scored. The batting strike-rate (runs per ball) is another independent measure and the two have to be considered together to determine the quality of a batsman. This article attempts to locate a single measure, somewhat equivalent to the bowling average.

Note the following players' figures.

                            M    Runs  Balls   St Rt  BatAvge
Turner G.M           Nzl   41    1598   2291   69.75   47.00
Greenidge C.G        Win  128    5134   7748   66.26   45.04
Javed Miandad        Pak  233    7381  10979   67.23   41.70
Haynes D.L           Win  238    8648  13566   63.75   41.38
Broad B.C            Eng   34    1361   2425   56.12   40.03
These are a set of high-average, low-strike rate batsmen who once ruled the roost in ODIs. There was a time when an ODI opener had to fall into this mould of a patient 150-ball-century person. Scores of 250-plus were eminently defendable and these batsmen fitted into the scheme of things.

                            M    Runs  Balls   St Rt  BatAvge
Ganguly S            Ind  300   11363  15416   73.71   41.02
Chanderpaul S        Win  230    7407  10473   70.72   40.26
Today the situation has changed. It is uncommon to see such a figure among current players who are playing ODIs. Even the opening batsmen have to have a higher strike-rate. If, in the bargain, their batting average drops, it is acceptable. The current batsmen who fit into the above group are, say, Ganguly and Chanderpaul. Both have reasonably high batting averages and not-too-high strike rates.

Now note the figures for the following batsmen.

                      M    Runs  Balls  St Rt  BatAvge
Shahid Afridi        248   5469   4911 111.36   23.47
Sehwag V             186   5867   5884  99.71   32.78
Powell R.L           100   2085   2157  96.66   24.82
Kapil Dev N          198   3783   4146  91.24   23.79
Jayasuriya S.T       404  12688  13931  91.08   32.87
McCullum B.B         111   2602   2874  90.54   29.24
These batsmen have very high-strike rates, above 90, but sub-35 batting averages. These are acceptable today, whether for opening batsmen or middle-order finishers. While everyone would like a Shahid Afridi or Sehwag to have higher batting averages, their very method of playing, attacking from the first ball, prevents them from having higher batting averages. Their role in today's ODI matches, however, cannot be denied.

However there is a need to look at things in perspective. Batsmen such as Chanderpaul and Ganguly, with averages of 40+ and strike rates of 70+ are still valuable in ODIs. How do they compare with, say, Andrew Flintoff, who only averages 31+ but has a strike rate of 87+. Similarly how do two middle-order batsmen of contrasting styles such as Yuvraj Singh and Mohammad Yousuf compare.

To compare ODI batsmen, I have thought of a simple index, hereinafter called ODI Batting Index (OBI). Since there are two measures both of which have the characteristic of, the higher the better, we multiply the batting average by the strike-rate and arrive at the OBI. Because it is a product, both values have equal weightage. This OBI Index is used to compare the ODI batsmen.

Now let us look at a table of the top 20 ODI batsmen ever based on the OBI. The minimum criteria is 2000 ODI runs which represents between 50 and 60 ODI matches, a fair measure of a settled batsman. A total of 143 batsmen qualify under this criteria. The analysis is correct until match # 2739, Ireland vs Scotland, one of the many inconsequential matches.

SNo LH Batsman             Ctry Inns  NO   Runs    Avge   StRt   Idx1

1 ~ Hussey M.E.K Aus 68 26 2307 54.93 85.6 47.04 2 Dhoni M.S Ind 101 27 3536 47.78 91.9 43.93 3 Richards I.V.A Win 167 24 6721 47.00 88.7 41.67 4 Pietersen K.P Eng 70 13 2687 47.14 86.6 40.85 5 ~ Bevan M.G Aus 196 67 6914 53.60 74.4 39.85 6 Tendulkar S.R Ind 407 38 16361 44.34 85.5 37.90 7 Symonds A Aus 157 33 5006 40.37 92.8 37.46 8 Zaheer Abbas Pak 60 6 2572 47.63 78.6 37.44 9 ~ Klusener L Saf 137 50 3576 41.10 89.9 36.95 10 Clarke M.J Aus 121 28 4037 43.41 80.5 34.93 11 ~ Gilchrist A.C Aus 279 11 9619 35.89 96.9 34.79 12 Ponting R.T Aus 292 35 11112 43.24 80.3 34.72 13 ~ Hayden M.L Aus 155 15 6132 43.80 78.9 34.58 14 ~ Gambhir G Ind 56 7 2018 41.18 82.7 34.06 15 ~ Smith G.C Saf 131 9 5016 41.11 81.8 33.63 16 Sarwan R.R Win 123 25 4251 43.38 76.5 33.17 17 Sehwag V Ind 186 7 5867 32.78 99.7 32.68 18 Jones D.M Aus 161 25 6068 44.62 73.2 32.65 19 Mohammad Yousuf Pak 254 40 9243 43.19 75.4 32.55 20 ~ Lara B.C Win 289 32 10406 40.49 79.7 32.27

To view the full list, please click here.

This is a true measure of the greatness of an ODI batsman. The batsmen with very high batting averages such as Michael Hussey, Mahendra Singh Dhoni, Viv Richards and Kevin Pietersen are at the top because they also possess very good strike-rates, in excess of 85. Bevan drops off a little, because he does not have a very high strike-rate. Then come Sachin Tendulkar and Andrew Symonds, both with very good averages and eminently acceptable strike-rates. Zaheer Abbas has a very high batting average and an acceptable strike-rate. Lance Klusener is a surprise presence in the Top 10, mainly because people tend to think only of his explosive striking ability, forgetting that he has a batting average better than Adam Gilchrist or Brian Lara. There is no doubt his spat with Graeme Smith has deprived South Africa of an outstanding player. No one can deny Michael Clarke a place in the Top 10.

The batsman with the lowest batting average to make it the list of Top 20 is Sehwag and he has made to the Top 20 because of his near-100 strike-rate. Gilchrist is in with similar credentials. Similarly the batsman with the lowest strike-rate in this Top 20 is Dean Jones, again another deserving candidate with a very high average. It's a similar case with Mohammad Yousuf.

One can conclude that an OBI value of above 30 signifies a very good ODI batsman.

Alternative 1

Some people might question that the product is not a true measure to define the combined value of two diverse factors because of skews. A 10% increase in either of the measures, will increase the Index value by 10%. A 10% increase in both the measures will increase the Index value by 21%. This may not be acceptable and readers may be justified in suggesting that a sum, rather than a product, should be done. It can easily be achieved with the following, slightly more complex, method.

OBIdx2 = Batting Average + (50.0 * Strike-Rate).

This is an excellent way of measuring the batting qualities since a batting Average of 50.0 is almost the pinnacle and a strike-rate is 1.00 is almost the same lofty level. I personally prefer this index as the better balanced of the two.

SNo LH Batsman             Ctry Inns  NO   Runs    Avge   StRt   Idx2

1 ~ Hussey M.E.K Aus 68 26 2307 54.93 85.6 97.75 2 Dhoni M.S Ind 101 27 3536 47.78 91.9 93.75 3 Richards I.V.A Win 167 24 6721 47.00 88.7 91.33 4 ~ Bevan M.G Aus 196 67 6914 53.60 74.4 90.77 5 Pietersen K.P Eng 70 13 2687 47.14 86.6 90.47 6 Tendulkar S.R Ind 407 38 16361 44.34 85.5 87.08 7 Zaheer Abbas Pak 60 6 2572 47.63 78.6 86.93 8 Symonds A Aus 157 33 5006 40.37 92.8 86.77 9 ~ Klusener L Saf 137 50 3576 41.10 89.9 86.05 10 ~ Gilchrist A.C Aus 279 11 9619 35.89 96.9 84.36 11 Clarke M.J Aus 121 28 4037 43.41 80.5 83.64 12 Ponting R.T Aus 292 35 11112 43.24 80.3 83.39 13 ~ Hayden M.L Aus 155 15 6132 43.80 78.9 83.27 14 Sehwag V Ind 186 7 5867 32.78 99.7 82.63 15 ~ Gambhir G Ind 56 7 2018 41.18 82.7 82.54 16 ~ Smith G.C Saf 131 9 5016 41.11 81.8 82.01 17 Sarwan R.R Win 123 25 4251 43.38 76.5 81.61 18 Jones D.M Aus 161 25 6068 44.62 73.2 81.21 19 Kallis J.H Saf 260 49 9542 45.22 71.4 80.90 20 Mohammad Yousuf Pak 254 40 9243 43.19 75.4 80.87

To view the full list, please click here .

The reconstituted table is almost the same, barring minor moves indicating either of the measures can be used. An index value of or around 100.0 is the indication of a truly great ODI batsman.

Michael Bevan and Pietersen swap places, Clarke and Gilchrist also swap places. Gilchrist moves into the Top 10 at the expense of Clarke. The batsman who has benefited most is Sehwag, who moves up three places. Jacques Kallis comes into the Top 20 at the expense of Lara. This computation method, in general, will benefit the low average - high strike-rate batsmen since the multiplication method will not help them greatly.

Alternative 2

There is no denying that the impact of not-outs is too pronounced in the first two cases. While not denying Hussey's place in the top echelons of ODI batsmen, he is placed too high with a batting average of 54.93. He has 28 not-outs out of 68, a very high 41%. Similar cases exist with Dhoni, Bevan, Klusener et al.

Hence a third alternative is tried out. This time I have computed the OB Idx as follows.

OBIdx2 = (Runs/Innings) + (40.0 * Strike Rate).

The multiplying factor is 40.0 (as against 50.0) since the highest runs per innings value is 42.87 and only four batsmen exceed a RPI value of 40.0. Similarly only one batsman exceeds a Strike Rate of 100.0.

This takes care of all anomalies. Only the middle-order batsmen might complain. But it is now fair across the board. Now let us look at the table.

SNo LH Batsman             Ctry Mat Inns   Runs    Avge   StRt   Idx3

1 Richards I.V.A Win 187 167 6721 40.25 88.7 75.71 2 Tendulkar S.R Ind 417 407 16361 40.20 85.5 74.39 3 Zaheer Abbas Pak 62 60 2572 42.87 78.6 74.31 4 ~ Gilchrist A.C Aus 287 279 9619 34.48 96.9 73.25 5 Pietersen K.P Eng 76 70 2687 38.39 86.6 73.05 6 Dhoni M.S Ind 115 101 3536 35.01 91.9 71.79 7 Sehwag V Ind 191 186 5867 31.54 99.7 71.43 8 ~ Hayden M.L Aus 161 155 6132 39.56 78.9 71.14 9 ~ Smith G.C Saf 133 131 5016 38.29 81.8 71.00 10 Ponting R.T Aus 301 292 11112 38.05 80.3 70.18 11 ~ Trescothick M.E Eng 123 122 4335 35.53 85.2 69.63 12 ~ Gambhir G Ind 56 56 2018 36.04 82.7 69.12 13 Symonds A Aus 193 157 5006 31.89 92.8 69.00 14 ~ Saeed Anwar Pak 247 244 8823 36.16 80.7 68.45 15 ~ Gayle C.H Win 183 179 6488 36.25 80.3 68.35 16 ~ Hussey M.E.K Aus 90 68 2307 33.93 85.6 68.18 17 ~ Lara B.C Win 298 289 10406 36.01 79.7 67.89 18 ~ Jayasuriya S.T Slk 415 404 12688 31.41 91.1 67.84 19 de Villiers A.B Saf 67 64 2147 33.55 85.3 67.65 20 ~ Ganguly S.C Ind 311 300 11363 37.88 73.7 67.36

Ah, now we see the real quality. Richards at the top, followed by Tendulkar, two ODI masters nonpareil. The only surprise in the top 10 is Zaheer Abbas in third position. The other nine will be in anybody's list of top ODI batsmen of all time. Richards, Tendulkar and Gilchrist gain at the expense of Dhoni, Hussey and Bevan, the later two move way down the field. Now about Zaheer Abbas. One of only four batsmen to average over 40 runs per innings (Richards, Tendulkar and Greenidge the other three) and a strike-rate of nearly 80. His RPI value is 42+, two above Richards and Tendulkar. He was and is one of the most under-rated players of all time.

An Idx3 value of 80 is the pinnacle, 70+ signifies greatness, 60+ signifies an outstanding batsman and 55+, a very good batsman. Shahid Afridi, despite his low average, gets an excellent value of 66.60 and is at no. 26.

To view the full list, please click here

I can anticipate a common reader observation. "How did you determine the career balls played for older players such as Richards, Greenidge, Zaheer Abbas et al". The following method is used.

  1. Even for some early matches the balls played is available straight off in the Cricinfo archive scorecards.
  2. Non-Cricinfo archives have been searched for some important innings.
  3. Bill Frindall's excellent ODI scorecards book contains the balls played for a number of key innings.
  4. Where nothing is available, the individual batsman strike-rate is extrapolated from the team rate. This method is not complete but gives us a fair idea of the balls faced. It might also be unfair to the more attacking players such as Richards, Kapil Dev et al and would benefit players such as Desmond Haynes, Sunil Gavaskar et al. I can assure readers that this method has only been resorted to when all other avenues have been investigated.
The most acceptable index value is the third one (Idx3) since it negates the impact of not-outs very effectively.

This method treats batsman Nos 1-5 in a fair manner and is possibly slightly unfair to positions 6-7, which is quite acceptable. The addition method is also devoid of statistical imperfections. Hence the third index method is the one I would recommend for consideration.

Based on this index, Viv Richards is at the top, closely followed by Tendulkar and Zaheer Abbas.

Unlike the Extended Batting Average [for Test batsmen], which was a substitute for batting average, this is a new measure which encompasses two of the existing measures. It would be a very useful tool for ranking batsmen and determining the strength of the team. I will be using this measure in the ODI Team Strength Analysis, which will be done sometime in the future.

This concept was first used and explained by me during the Wisden 20-20 Television shows broadcast on Doordarshan during 2001-02. These ideas have since been refined considerably.

Response to user comments

1. This is not an overall "Who is the greatest ODI Batsman" analysis. That has been done by me for Wisden a couple of years back and I will do it again in this column later. In that analysis I will take ino account factors such as Runs scored, Wins, % of Team runs, MOM awards, Quality of opposition et al.

This is a list of top ODI batsmen using only the two relevant factors, viz., Runs scored (Avge or RPI) and Scoring Rate. This should only be compared to "Listing the best ODI bowler by Bowling Average".

2. David Barry has correctly mentioned the Cricinfo Statsguru figures. However let me mention that my need is not an overall "Richards scored xxx runs in yyy balls" information. That will only help a Career analysis. I need information by match and I have resorted to the method shown. May not be perfect. But in this particular case, nothing will be perfect. My Database methodology does not permit (correctly) a career sum, different to the sum of individual innings values.

3. Unlike what some people have implied, I have no intention or desire to push one batsman or other up or down. I do the analysis and leave the readers to judge for themselves. If in the bargain the readers' favourite batsman is not shown where he should be (in their perception), my apologies. Only thing is, don't be rude to me or other readers (your comment will go the trash then, even if it is a good one) nor ascribe the variation to deliberate intention. After all these are only analyses.

4. If the limit of 2000 runs is too low, one could increase it to, say 2500. Maybe then Gambhir would not get in. Then people should accept that Hussey, Botham, Greg Chappell et al would also not get in.

5. David (Barry) has since pointed out that the revised Cricinfo scorecards have complete balls faced data. It is possible that I have used the older version with incomplete scorecards and had to resort to extrapolation. It will take me some time to convert to the newer version since it involves downloading, parsing and updating the database. I will immediately start work on this. Many thanks to David.

6. There has been an excellent suggestion to do some equalizing of Strike Rates between the early years and recent years. Let me look into that. That will at least enable olden greats such as Greenidge, who certainly is not a career strike rate of 66.3 into something more acceptable, say 75.0. In this case a case should be made for equalizing averages or RPI also.

7. Abhijeet Dongre has made a good suggestion. To consider the not out innings as not an innings if the runs scored falls below the average is a novel idea and is worth considering. Let me work on it.

8. Jayasuriya's somewhat low position seems to have offended quite a few people, especially from the lovely little emarald isle. My own take on this is that he would certainly in the Top 10 when I do a comprehensive analysis of ODI Batsmen considering all facts mentioned in point 1 above. Until then my Sri Lankan brethren, please bear with me. At least you don't abuse me. You even make your protests in a gentle manner.

9. Jeff Grimshaw's Balls per dismissal suggestion has a very authentic and solid feel about it and I am seriously looking at incorporating the same. Same as with Abhijeet's suggestion.

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