Trivia - batting May 23, 2008

Why Australia's 2001 line-up is the best ODI side

How strong is an ODI team and how do the teams compare over the 37 years of ODI cricket


"I urge readers to read and understand the reasoning behind the analysis. It is NOT to determine the best ODI team across years or teams. Rather it is to determine the best team that walked on to the field, as 11 players. Many comments have been made ignoring this fact. So much so, no comment which lists the readers' favourite team will be published. Let me add that over 50 comments have gone unpublished because of this."

For my next post, I wanted to stay away from Test cricket, on which most of the recent It Figures posts have been. At the other extreme we have Twenty20, which has had an all-pervading presence on almost all the channels on television, and the web and print media as well. That leaves the often-ridiculed form of cricket, one-day internationals. I never thought I would say this, but I have already started longing for ODI cricket.

This time I have taken for analysis a topic which I had looked at for Tests, and am now adapting to ODIs: how strong is an ODI team and how do the teams compare over the 37 years of ODI cricket? Where does the 2007 Australian team stand when compared to the West Indian teams of early 1980s, or for that matter the Australian teams of the late 1990s? It has turned out to be a fascinating study.

The one significant advantage we have when comparing ODI teams is that even the 1975 West Indies team had players most of us [barring those below 30, who would anyhow be familiar with them] have seen. It is not very difficult to identify with Viv Richards, Ian Chappell, Clive Lloyd, Michael Holding etc. unlike in Test cricket, where George Lohmann, with a Test bowliing average of 10.76, was born nearly 143 years ago. It isn't easy to relate to either fact.

A team is as strong as its batsmen, bowlers and fielders are. If we consider fielding as part of the bowling, these two main areas have to be given equal weightage. ODI laws might be tilted towards the batsmen, but the role of bowlers can never be underestimated. This happens even in the Twenty20 game.

Hence I have given a weightage of 50 for batting and 50 for bowling (further split as 45 for bowling and 5 for fielding). Because there is no quantified data for fielding per se, the weightage for fielding is in reality for catching/stumping. This also explains the low weightage.

The one thing I want to ensure is that this analysis will comprise only of measurable, objective parameters. The other areas such as captaincy, recent form, home advantage etc. are intangibles and subjective. A captain is only as good as his team is. Recent form has more relevance in Test matches. Home advantage is a mirage. The non-Australian strokemakers would love to play on the bouncy Australian pitches and the non-New Zealand seam/swing bowlers would love to bowl in Auckland or Hamilton.

Readers might be tempted to send the usual comments that these are obvious and why should there be a need to do analysis. Let me remind such readers that their conclusions would be based on error-prone subjective inferences and also not indicate how much a team is better than another. My results are based on objective analysis and indicate the quantitative differentials between teams.

ODI batting consists of two distinctly measurable and independent factors: how many runs are scored and how fast they are scored. In other words, the batting average and the strike-rate. No one can question the decision to treat these two parameters equally.

The average is taken rather than the lesser known and acceptable runs per innings or my own development, the extended batting average. The average is a widely accepted measure and presents the best method of measuring runs scored. Only two batsmen in ODI history, Michael Bevan and Michael Hussey, have averages higher than 50 (among those with a minimum of 20 innings), mainly because of their number of not-outs. However this is partly rectified by limiting the average to 50.0 for these two batsmen.

There is no problem with strike-rate. That is available as a straight computation of runs scored / balls faced. The averages and strike-rates of the top seven batsmen in the team's batting order are summed. The averages and strike-rates for batsmen nos. 8 to 11 are given a 25% weightage each. The arrived total is divided by eight and the Team Batting Average and Team Strike-Rate are arrived at.

The batting average index points are determined by dividing the team batting average by two. The maximum value for this is 25.0.

The strike-rate index points are determined by multiplying the team strike-rate (runs per ball) by 25.0. The maximum value for this is just over 25.0. Only one batsman in ODI history, Shahid Afridi, has a career strike rate of over 1.00.

Care is taken that these full values are applied only for career aggregates of 1000 runs and above. Otherwise Arvind Kandappah of Canada and Alex Obanda of Kenya will single-handedly make their team's batting averages huge. These two have Bradmanesque career batting averages of 97.0, although scoring only 97 and 194 runs respectively.

SNo. Year MtNo I Team   vs     AvIdx  SRIdx  Bat

1. 2005 2257 2 AUS (vs Bng) 19.89 20.99 40.87 (Gilchrist, Hayden, Ponting, Martyn, Clarke M,
Symonds, Hussey). 2. 2005 2261 2 AUS (vs Eng) 19.91 20.90 40.81 3. 2005 2259 1 AUS (vs Eng) 19.67 20.78 40.45

Next 105 teams are Australian, followed by

109. 2005 2282 2 ICC (vs Aus) 18.15 20.47 38.62 (Sangakkara, Sehwag, Kallis, Lara, Dravid, Pietersen, Flintoff, Afridi)

Next 16 teams are Australian, then

126. 2004 2202 1 IND (vs Bng) 18.06 20.43 38.49 (Sehwag, Tendulkar, Ganguly, Dravid, Kaif, Yuvraj Singh, Dhoni).

Then another 5273 teams

5400. 2004 2172 1 USA (vs Aus) 3.27 9.73 13.00 5401. 1979 0067 1 CAN (vs Eng) 5.05 7.85 12.91 5402. 1979 0070 1 CAN (vs Aus) 4.76 6.98 11.74

Note: Out of the 2703 matches considered, two matches were abandoned without even the team information being available.

The first 108 teams in the batting list are Australian. These 108 matches have come over a nine-year period, from 1999 to 2008, a period of total Australian domination, punctuated by three World Cup wins. The three batsmen who have been part of almost all these matches are Adam Gilchrist, Ricky Ponting and Andrew Symonds.

Like batting, bowling also has two components, the bowling strike-rate and accuracy. However, unlike batting, the bowling average is a fantastic measure since it encompasses both these key measures in a single value. Consider the following.

Runs Conceded
Bowling Average = -------------
Wickets Taken

Rewriting this as

Runs Conceded Balls Bowled Bowling Average = ------------- x ------------- Balls Bowled Wickets Taken

This can be written as

Bowling Average = Bowling Accuracy x Bowling Strike-Rate.

There is no need to measure these two factors independently. It is sufficient to take the single composite measure, bowling average and work on it.

Unlike the batting computation, the bowling averages of the best five bowlers is taken and divided by five. This is because it is expected the team would use their best five bowlers. Even if Jacques Kallis bats at No. 3, he is likely to be used as a bowler if he is one of the best five. Whether he bowls in the concerned match or not is outside the scope of this analysis since this study only measures how strong a team potentially is, not how strong the team actually was.

Here also care is taken that bowlers with less than 50 wickets have their figures scaled down suiitably. Otherwise Gary Gilmour, with 16 career wickets at 10.31, will completely tilt the figures of the late-1970s Australian teams.

The bowling index is determined by subtracting the Team Bowling Average from 60.0. Since the best bowling average for qualifying bowlers [minimum 50 wickets] is 18.85 by Garner, the highest value will not exceed the maximum weightage given to bowlers, of 45.

For both batting and bowling, I have also taken the full career figures rather than the career-to-date figures in view of the complexity of calculation and the fact that we are averaging and the minor differences tend to get ironed out.

Only catches and stumpings are considered. The values for all 11 players are added, divided by 11, and multiplied by two to get a team fielding average. The highest value is 1.95 and the maximum index value is 3.90. It is obvious that this figure will be strongly influenced by the wicketkeeper's figures. A per match average rather than catches/stumpings aggregate is taken to be fair to weaker teams.

SNo. Year MtNo I Team   vs    Fld   Bow   Tot

1. 1981 0116 2 WIN (vs Eng) 1.55 38.65 40.20 (Roberts, Holding, Croft, Garner) 2. 1982 0134 2 WIN (vs Pak) 2.25 37.79 40.03 3. 1982 0135 2 WIN (vs Aus) 2.25 37.79 40.03

Next 21 teams are West Indian, then

25. 2001 1670 2 AUS (vs Win) 2.44 35.95 38.38

Then another 5374 teams

5400. 1979 0070 1 CAN (vs Aus) 0.17 10.00 10.17 5401. 1979 0067 1 CAN (vs Eng) 0.17 10.00 10.17 5402. 1979 0064 1 CAN (vs Pak) 0.17 10.00 10.17

The first 24 teams in the batting list are West Indian teams. These 24 matches have come over a six-year period, from 1981 to 1987. The two bowlers who have been part of almost all these matches are Holding and Garner.

Final Team Strength
This arrived by adding the batting, bowling and fielding indices. The maximum is 100, making it easier to see things in perspective.

SNo. Year MtNo I Team   vs     Bat   Bow  Fld  Team

1. 2001 1670 2 AUS (vs Win) 38.95 35.95 2.44 77.34 (Gilchrist, Waugh M, Ponting, Bevan, Lehmann, Symonds, Martyn, Warne, Lee, Bracken, McGrath). 2. 2004 2180 1 AUS (vs Eng) 39.28 35.39 2.56 77.23 (Gilchrist, Hayden, Ponting, Martyn, Lehmann, Clarke M, Symonds, Lee, Gillespie, Kasprowicz, McGrath). 3. 2004 2172 2 AUS (vs Usa) 39.28 35.39 2.56 77.23 (Same as previous team) 4. 2004 2131 2 AUS (vs Zim) 39.10 35.30 2.69 77.09 5. 2003 1951 2 AUS (vs Ind) 39.47 35.02 2.43 76.91

Next 140 teams are Australian, then

146. 1982 0139 2 WIN (vs Aus) 33.82 37.79 2.25 73.86 (Greenidge, Haynes, Richards, Gomes, Lloyd, Bachhus, Dujon, Roberts, Holding, Clarke ST, Garner).

Next 19 teams are Australian/West Indian, then

166. 2005 2282 2 ICC (vs Aus) 38.62 32.75 2.28 73.65 (Sangakkara, Sehwag, Kallis, Lara, Dravid, Pietersen, Flintoff, Shahid Afridi, Pollock S, Vettori, Shoaib Akhtar, Muralitharan).

Then another 5233 teams

5400. 1975 0024 1 EAF (vs Ind) 13.06 10.00 0.52 23.58 5401. 1979 0067 1 CAN (vs Eng) 12.91 10.00 0.17 23.07 5402. 1979 0070 1 CAN (vs Aus) 11.74 10.00 0.17 21.91

The first 145 teams in the list are Australian. These 145 matches have come over a nine-year period, from 1999 to 2008, a period of total Australian domination, punctuated by three World Cup wins. The five players who have been part of almost all these matches are Gilchrist, Ponting, Symonds, Brett Lee and Glenn McGrath.

Finally I have done another "fourth dimension" formation. Australia have had the best batting teams ever and West Indies, the best bowling teams ever. Let us combine the two into one all-time great ODI team. Take the first seven players from the Australian 2005 team [Match no. 2257] and add to it the best four bowlers from the 1981 West Indies side [Match no. 116]. Given below is the final squad.

Just to round up the analysis, this all-time great team has an index value of 77.05, which is lower than the Australia 2005 figure. This has been caused no doubt by the loss of batting and fielding points of the Australian team (Watson/Lee/Gillespie/Kasprowicz are much better batsmen and fielders than the West Indian bowling quartet). However, the team listed below is an outstanding one with a superb bowling attack.

Adam Gilchrist
Matthew Hayden
Ricky Ponting
Damien Martyn
Michael Clarke
Andrew Symonds
Michael Hussey
Andy Roberts
Michael Holding
Colin Croft
Joel Garner.
Readers should not forget that this not necessarily the best ODI Team of all time, It has been formed by merely taking the first 7 players from the best ever Batting line-up and adding the 4 bowlers from the best ever Bowling line-up.

Theoretically this team can be further improved by taking in Tendulkar, Richards, Dhoni, Wasim Akram, Shane Warne et al. That is a different day and different motivation. For the present let us enjoy the combination of two different eras.

If we tamper with this team, the charm would be lost. The Australia-West Indies combination would be missing. After all, these two countries have dominated the ODI scene during these 37 years, West Indies during the first ten years and Australia, the last 20.

ODI Analysis - by decade
Batting AllMats 1970s 1980s 1990s 2000s
Matches played 2703 82 516 933 1172
Runs scored 1119374 30292 202284 386508 499690
balls bowled 1445956 46208 277516 505727 616505
Batsmen innings 46968 1418 8838 16266 20446

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

Comments have now been closed for this article

  • testli5504537 on June 28, 2008, 9:35 GMT

    Dear Ananth,

    More Brilliant work!

    Your analysis always provides a great starting point for discussion. I am sorry there are so many who fail to read the articles properly, understand the purpose or simply resist the temptation to thrust their favourite player forward.

    My only comment is that the higher scores and strike rates in the modern game benefits the batting index more than the bowling index suffers because there are always more batsmen than bowlers in a team. So it is no surprise that a modern Aussie team will win.

    People have said that modern Aussie bowlers have had to contend with higher scoring, but the effect on team scores is less than older Windies batsmen contending with lower scoring environments...

    I do believe there needs to be scaling of team performances against changing averages for scoring rates since the 1970s

    Finally, if you post any composite team, you are inviting a thousand back! i wont bother you with mine...

  • testli5504537 on June 6, 2008, 15:58 GMT

    to anyone even comparing Dhoni to Adam Gilchrist do you even know anything about cricket? Gilly has changed the way that teams select a wicket keeper. No longer is his job to simply stand behind the stumps and catch the ball he now must also be able to score runs at a very quick rate. Dhoni altho being a very good player is no near the same calibre as Adam Gilchrist who has proved over many years against the strongest attacks in the world that he is a master whereas Dhoni has been playing for only couple of years against poor bowling attacks on completely batsman friendly pitches and grounds where a mis hit doesn't only go for 6 but it goes 15 rows back and you would be caught in Australia.Does nobody realise that Australia were the favourites for the 1996 world cup in Australia and New Zealand and had been the dominant team in the year leading up so if my maths are correct then Australia has been the dominant teeam for the last 23 years so well done to Ananth for getting it correct

  • testli5504537 on June 4, 2008, 13:12 GMT

    Great job and a wonderful analysis, good thoughts and well laid out. However, the way i see it is just how statistics can be so misleading. Just the very fact that your final team has all batsmen from post 1990 era and the bowlers from the pre 1990 era shows how the paradigm of cricket has changed into being batsmen-oriented in these two eras. Honestly I love statistics but the way of interpretation is something I have not found appealing especially showing hypothetical teams, the very fact that you have assumed that players from the 1975 can be related to current unlike in test cricket where comparisons go more than a century is in itself a big flaw I would say. [[Please note I do not have a final team. I merely put together the 7 Australian batsmen and 4 West Indian bowlers to do a "what if" situation. These are not to be taken too seriously since Hayden and Holding can only play together in fantasy teams.]]

  • testli5504537 on May 30, 2008, 17:38 GMT

    Good analysis. But this statement may not be correct: No one can question the decision to treat these two parameters(the batting average and the strike-rate) equally.

    There are match-winning performances with low averages but high strike rates and in some case high average with lower strike rates. How can these two parameters can be equal?

  • testli5504537 on May 30, 2008, 6:20 GMT

    You have missed the point. I have clearly demonstrated a case (4-5 highly erratic but extremely attacking bowlers) where a "best" team, picked on the sole "composite" criterion - bowling average, would consistently concede 265+, well above what you would expect on average from the best ODI bowling attack (38.0 on your index vs 38.65 for your uncorrected best side - EPIC FAIL!) Note that you are implicitly using the economy rate (i.e, it's independent effect is masked). By using the composite you are relying on the synergy of the two variables which as I have shown can produce misleading results. You must consider each effect independently with appropriate weights. You could also explain the process a little better; why is the team bowling average subtracted from 60, how did you arrive at 60, did it conveniently satisfy the best case? How will this analysis then stand the test of time if we get a better bowling team in the future? You also seem to have missed my previous follow-up : Response [1. 60-Avge was taken since it allowed me to contain the Bowling Index within the limits. There is no great need for this algorithm to stand the test of time. The purpose is to let people think and participate in useful and friendly discussions. 2. If there are 5 Joel Garners in a team, the Bowling Index will be 41.15 (out of a maximum of 45.0). I cannot even comprehend a bowling attack better than this. 3. I seriously thought of considering the two factors Bowling Strike Rate and Bowling Accuracy separately. However because of the evolution of the game into a more-batting dominant one and also the change in emphasis on bowling requirements, I felt it would be fair to consider the Bowling Average as a single composite entry across years. I could still do separate weighting of the two factors.]

  • testli5504537 on May 29, 2008, 22:34 GMT

    "Keep Gilchrist, and have Dhoni replace Ponting at No. 3"

    Oh my, I'm so glad you're joking.

    I enjoyed this analysis, keep up the good work.

  • testli5504537 on May 29, 2008, 18:57 GMT

    Great analysis. I have a interesting analysis. If you did the same kind of analysis on individuals rather than teams, would it look like this: (Assuming a team needs at least 1 wk, and 2 spinners)

    Adam Gilchrist Niel Johnson Jacques Kallis Chris Cairns Andrew Flintoff Kapil Dev Imran Khan Lance Klusner Shaun Pollock Shane Warne Daniel Vettori

  • testli5504537 on May 29, 2008, 13:21 GMT

    i think everyone who played before the introduction of boundary ropes should have an increase in their average. however, a better statistical analysis would involve changing to a weighting of the "normal" average for the year. ie, take the 100th best player in the world for the year, and set him to 1.00 rating, then make everyone a rating based on that.

    this type of rating will show domination far more easily, and will mean that the "lower bowling" abilities of 1995+ will show up less in the batting averages. this will decrease the domination of the windies attack and improve the batting of the early 1980s against such an attack.

  • testli5504537 on May 29, 2008, 10:12 GMT

    You've really got me thinking here (the sign of a good blog :-)

    If I take your No. 1 team - their AvIdx is 19.89, meaning that their predicted score based on this would be 318 (19.89 x 2 x 8).

    But as their Strike Rate is 0.84 (20.99 / 25) then they would only be able to score 252 in 50 overs - so this is actually their predicted average score.

    So for this team, Strike Rate is clearly more important (in fact their AvIdx could be as low as 15.75 and they would still be predicted to score 252)

    I'm sure there is a calculation that could be made to determine what would be the predicted score based on Ave & SR (with the 300 balls the limiting factor)and the team with the highest score would be the best batting team.

    Maybe something similar could be done with bowling (I haven't had time to look at this) and then the best team would be the one with the biggest difference between predicted batting & bowling scores?

    I assume it will still put those pesky Aussies first...

  • testli5504537 on May 29, 2008, 8:34 GMT

    The problem with only considering the bowling average, is it can mask the economy rate. As the strike rate approaches the bowling average, the economy rate increases. Consider a bowler with a bowling average of 22 (which is right up there with the top averages, if a correction is made accounting for the average team scores across the years, which imho is only fair if you truly want to compare teams across eras.) and a strike rate of 25 (which is also brilliant). This bowler would have an economy rate of 5.3. You can expect his usual bowling figures would be 10-X-53-2. Take four such bowlers, and off 40 overs, they would have conceded 212-8. If the fifth bowler is at least as good, you can expect a normal score of 265 all out, though I would suspect the score might range from 270-280. Would you expect the best bowling team to concede this score on a consistent basis? By masking the economy rate, your analysis has no mechanism to address this, lucky the WI team had high strike rates, eh? Response [The Economy rate is not masked. It is an integral part of the Bowling Average. If the reader re-reads the article carefully, he will understand this. The algorithms are clearly explained there. A Bowling Average of 24 could be achieved by a combination of Strike Rate of 36 and Economy Rate of 0.666(rpo-4.0) (a very attacking but erratic strike bowler). Alternately it could be through a Strike Rate of 48 and Economy Rate of 0.5(rpo-3.0) (an accurate and steady bowler). Thus it can be seen that both Strike Rate and Economy Rate have equal importance and one is not masked.]

  • No featured comments at the moment.