# T20 - Target score for first innings

Finally I have come around to my first T20 analysis. I had to do some serious T20 ratings analysis work related to another project and as part of that work, I looked at T20 matches from a totally different angle. One aspect of this analysis was to determine a reasonable target score for the first innings (the target score for the second innings is no problem, even when the learned professors, M/s Duckworth and Lewis come in with their umbrellas!). The team’s achievement in terms of exceeding or falling short of the dynamically computed target score is determined to compute one segment of the individual ratings. It also allows me to allocate the credits between bowlers and batsmen.

Let me add that my database, current upto the West Indies - England game, is limited to T20 International matches and as of now I have no intention of building a Database of other club-based T20 matches.

First some facts about T20 matches. Let me say that I have completely ignored team strengths, pitch conditions et al since there is not enough data and in this short version of the game, there is lot more evening out between the two teams.

1. A total of 84 matches have been played and completed. Out of these, 4 have been tied. 2 matches outside these 84 have been washed out.

2. 34 (out of 80) matches have been won by the team batting first. This represents 42.5% of the completed matches. One of these wins has been through D/L method.

3. 46 (out of 80) matches have been won by the team chasing. This represents 57.5% of the completed matches. One of these wins has been through D/L method.

4. Out of these 46 matches, the top 4 run chases have been against scores of 165 and above. These four succesful run chases are detailed below. In other words, any team scoring 165 and above has a 90+ % chance of winning the game. This seems to be true irrespective of the relative team strengths. It is also possible that the weaker team might bat first more often than not.

020 2007 Win 205/ 6 (20.0) Lost to Saf 208/ 2 (17.4) at Wanderer's, Jo'burg 082 2009 Slk 171/ 4 (20.0) Lost to Ind 174/ 7 (19.2) at Premadasa, Colombo 016 2007 Win 169/ 7 (20.0) Lost to Eng 173/ 5 (19.3) at Oval, London 047 2007 Aus 166/ 5 (20.0) Lost to Ind 167/ 3 (18.1) at Brabourne, Mumbai

5. It is a reasonable assumption to make that the team batting first should set themselves a Target score of 165 runs to have a 90+ % chance of winning. Anything more would obviously further increase the chance of winning. However we are not looking at a Target score with 100% chance, which, at the current moment is 206.

6. If we drop the number from 165 to 160, the number of losses is more than doubled since 5 more matches are won by batting teams chasing 164, 164, 164, 162 and 162 successfully. The win % drops to 80% so there is a need to retain the Target score at 165.

It is possible that in the next 5 matches, 170+ scores would have been chased. However that does not make the idea of working on a Target score invalid and as things stand, **165 seems to be a very good number for a captain to write on the team sheet**.

The reason this score is very relevant is because of what happened in the two T20 matches between New Zealand and India. Each time India had an explosive start, looked good to score 200, tried to score 200 and finished with 162 and 149. Both scores were chased down with ease, although New Zealand were too cautious in the middle overs in the scond T20 and almost threw the match away. They should have won more comfortably with the explosive start set by the openers.

The importance of not aiming for too much cannot be over-emphasized especially in T20 matches. In T20 it is paramount for the captains to understand the nuances of the game. It is possible that Dhoni is aware of this. However his batsmen, Gambhir, Sehwag, Yuvraj, Sharma et al tried to attack without a clear understanding of the par score.

In ODIs, nowadays even scores of 300+ are chased quite comfortably. However even there a reasonable target score should be aimed at. The 100% winning score is 435. However the par Target score might very well be 285. But it must be remembered that data is available for 2822 matches for us to make a facts-based determination of a par Target score for a venue.

Just to sum up the first batting wins. Out of the 34 wins, 8 teams have won by putting up a total of 200 and above, 11 by posting wins of between 180 and 200, 10 by posting between 150 and 180 and 4 have been bowling wins with sub-150 totals. One has been an amazing defence by Ireland of a total of 43 for 7 in a D/L match.

It is impossible to infuse the other Test/ODI parameters such as Ground/Pitch conditions, Team strengths, Average scores et al because of the low number of matches, the absence of any meaningful statistics and the very nature of the game.

Out of the 86 T20 matches, a whopping 34 have been played in South Africa, mainly because of the 2007 WC, in addition to one washed out match. 11 matches have been played in Ireland, in addition to one washed out match. 10 matches have been played in New Zealand. 8 matches have been played in Canada, all in one centre.

Wanderer's has staged the maximum of T20 matches, 16 in all. Just to give the readers an idea of the analyst's nightmare of determining a target score at Wanderer's, I have given below the 16 first innings scores. These seem to move like a yo-yo although there seems to be a recent trend for lower scores.

133, 201, 126, 129, 205, 164, 260, 164, 190, 189, 164, 147, 157, 129, 131 and 118.

**Note**: I stayed up to watch the interesting T20 match between Australia and South Africa. Australia scored 166 (just passing the par Target score mentioned) and lost a very close match. Strike 1 against me, I suppose.

**Important footnote**:

This refers to the points raised by Aneesh and Kieran. They have correctly questioned my 90+% figure.

First let me say that the 90+% is based on all instances of chasing team winning, which is 46. Out of these 46, only 4 chases have been of scores of 165 and above. Thus the figure of 90% came in.

However stricly speaking, both Aneesh and Kieran are correct. My sample should be the **teams which crossed 165** and not the successful chases. Let me work out that figure below.

26 teams crossed 165 (barring the last Saf-Aus match, which has been excluded for sake of consistency). Out of these, 4 teams lost and the other 22 won. So the winning % is 84 and not 90.

Hence I am going to change my Target score to 170, which would lead to 24 wins and 2 losses (win % of 92).

My thanks to Aneesh and Kieran.

T20 Batsman Strike Rates (Min 200 runs) - (Gokul)No Batsman Ctry Mat Runs Balls S/R BatAvg

1 Symonds A Aus 13 337 198 170.2 56.17 2 Yuvraj Singh Ind 9 262 159 164.7 32.75 3 Gayle C.H Win 7 261 162 161.1 37.29 4 Oram J.D.P Nzl 13 293 187 156.6 36.62 5 Jayasuriya S.T Slk 11 341 221 154.3 34.10 6 Imran Nazir Pak 10 201 134 150.0 25.12 7 Sehwag V Ind 11 223 154 144.8 20.27 8 Pietersen K.P Eng 15 375 260 144.2 26.79 9 Hayden M.L Aus 9 308 214 143.9 51.33 10 Jayawardene M Slk 11 210 147 142.8 23.33 11 Duminy J.P Saf 9 256 181 141.4 32.00 12 Gilchrist A.C Aus 13 272 192 141.6 22.67 13 Morkel J.A Saf 15 270 193 139.9 24.55 14 Collingwood P.D Eng 15 344 246 139.8 24.57 15 Masakadza H Zim 7 258 190 135.7 36.86 16 Aftab Ahmed Bng 9 215 161 133.5 26.88 17 Ponting R.T Aus 14 375 283 132.5 34.09 18 Gibbs H.H Saf 13 225 171 131.5 18.75 19 Shah O.A Eng 11 241 186 129.5 26.78 20 Taylor R.L Nzl 17 323 253 127.6 21.53 21 Misbah-ul-Haq Pak 14 398 312 127.5 56.86 22 Smith G.C Saf 12 364 286 127.2 36.40 23 Gambhir G Ind 11 328 259 126.6 29.82 24 Kemp J.M Saf 8 203 160 126.8 50.75 25 McCullum B.B Nzl 21 582 464 125.4 34.24 26 Shoaib Malik Pak 16 383 307 124.7 31.92 27 Younis Khan Pak 15 260 223 116.5 18.57 28 Styris S.B Nzl 15 272 240 113.3 19.43 29 Dhoni M.S Ind 12 215 207 103.8 23.89 30 Salman Butt Pak 12 266 288 92.3 26.60

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

This stat is of waste.I personally feel that the average score depends on the pitch conditions and the quality of the players.If there is a bit of moisture on the pitch and if the bowlers get a bit of swing then it is absolutely going to be a bowler's game and on the other side if the bowl gets on to the bat then it is going to be a batsman's game.One day if Shewag plays throughout the score may reach upto 190-200 and on one day if he gets out for a golden duck the score is obviously going to decrease.So i reckon these stats really won't work and the score absolutely depends upon that day's perfomance.

Great comment from Rex.

I get frustrated by the lack of meaningFUL stats available in cricket (as compared to baseball for example) and equally as frustrated by the continual use of largely meaningLESS ones. My biggest bugbear is the use of batting (and bowling) averages as a key stat in T20 (and to a lesser extent ODI) cricket. Take Symonds as an example – his T20 batting average is 56 and yet only once in 10 inns has he actually scored as many as 56. Why then, do we insist on quoting this figure?

The key factors in T20 batting are: How many balls does the batsman face before he’s out (or the inns ends)? How quickly does he score while he’s there?

And given there are only 120 balls per inns, I’d say the 2nd stat is the more important one.

We need a stat which combines these 2 factors in the right way to provide a meaningful way to judge how well a player uses the resources available. We need a “Bill James” for cricket…

And regarding new tools for analysing players how about dot ball % and boundary balls % for bowlers? similarly, balls taken to hit a six/four by the batsman? mccullum has played 464 balls and hit 23 sixes ie a six every 20 deliveries, yuvraj 21 in 159 ie a six in 8 deliveries, morkel 18 in 284 ie 1 in 16...

Gokul,

kindly note that the par strike rate for all 20-20s is not 137.5 as you are assuming on basis of the "safe" 1st innings score of 170.I ran the statsguru and including the 2 matches of sa-aus the average runs-per-over in all 88 t20s is only 7.49 ie nearly 7.5. this amounts to 150 runs in 20 overs and an strike rate of 125. knock off the extras and a strike rate of 120 seems fair enough (par). I agree with krish in that you need both types of batsman- a couple of bats who get 30 odd runs at strike rate for 120 and hold one end up (accumulators) and 3-4 others who get quick 20-25s at strike rate of 140 or more including one batsman with extra ordinary fire power. Example gambhir, dhoni accumulator, sehwag,raina,yuvi,yusuf hitters (the last 2 have huge strike rates and six hitting abilities) morkel and symonds do similar hard-hitting for their teams. continued

An excellent logical analysis. I feel that T20 has made cricket like baseball with respect to statistics, where in there are lots of stats (slugging average, batting average etc) which have no clear meaning. All those stats have to be interpreted in the right manner to arrive at conclusive results, and most of the time the stats don't reveal clear flaws or strengths.

The length of the T20 game means that traditional averages and strike rates and all such stats aren't a reliable judge of team strength or player performance.

Runs saved by the fielding team, dot balls earned by stopping singles and many such factors are vital.

More than forming a list of stats and arranging them in ascending or descending order, we need to logically analyse the situation of each game and come to a conclusion.

I feel this article is a good way of analysing T20. I hope we get more articles that analyze T20, and those that "invent" new tools to measure player or team performance in T20.

would it be appropriate if you also consider the instances (if any) where a team chasing a lesser total won and was in a postive position of chasing a greater total (say winning by chasing 155 in 17 overs with 5 wickets)

Krish,

I disagree. Gambhir averages nearly 30, which means he scores runs in nearly every game that he plays in at a strike rate less than the par-strike-rate. So he is indeed doing some damage by doing that.

Similarly Dhoni averages nearly 24 at a very poor strike rate. This is really surprising to me.

Of course the 137 is a par strike rate when batting first. I believe Ananth's list is the average strike rate for 1st and 2nd innings - so perhaps that makes a difference. I guess Dhoni's 1st innings SR is quite higher than a mere 103.

For anybody scoring more than 10 runs, a strike rate of less than 125 is not good. Overall strike rate which should give a good comfort level for a T20 match should be around 138, i.e. around 168 runs from the bat in 20 overs (assuming 2 overstepings in 20 overs). Total number of extras (around 10) needs to be added to this total. Hence, a total of 178 should be a decently defendable total.

An overview of your batsmen's list raises a pertinent point. Dhoni and Gambhir both vital cogs in India's batting have a less than acceptable strike rate (137.5) but India have gone on to win more than they have lost. This suggests that along with the dashers like Viru and Yuvi, a team does require relatively composed accumulators.

165 in 20 overs means a strike rate of 137.5 (ignoring extras). It'd be interesting to see a list of specialist batsmen who have a strike rate higher than this. Actually a list of batsmen with a strike rate lower than this would also be interesting, specially if their batting average is high. Someone averaging 25 with a strike rate of 110 would be disastrous for a team. Whereas someone averaging 15 with a strike rate of 150 would count as a potential match winner.

[[ Gokul, That is a good point. Since it will be difficult for me to provide you the list here, I have added the same to the main article itself at the end. Ananth: ]]This stat is of waste.I personally feel that the average score depends on the pitch conditions and the quality of the players.If there is a bit of moisture on the pitch and if the bowlers get a bit of swing then it is absolutely going to be a bowler's game and on the other side if the bowl gets on to the bat then it is going to be a batsman's game.One day if Shewag plays throughout the score may reach upto 190-200 and on one day if he gets out for a golden duck the score is obviously going to decrease.So i reckon these stats really won't work and the score absolutely depends upon that day's perfomance.

Great comment from Rex.

I get frustrated by the lack of meaningFUL stats available in cricket (as compared to baseball for example) and equally as frustrated by the continual use of largely meaningLESS ones. My biggest bugbear is the use of batting (and bowling) averages as a key stat in T20 (and to a lesser extent ODI) cricket. Take Symonds as an example – his T20 batting average is 56 and yet only once in 10 inns has he actually scored as many as 56. Why then, do we insist on quoting this figure?

The key factors in T20 batting are: How many balls does the batsman face before he’s out (or the inns ends)? How quickly does he score while he’s there?

And given there are only 120 balls per inns, I’d say the 2nd stat is the more important one.

We need a stat which combines these 2 factors in the right way to provide a meaningful way to judge how well a player uses the resources available. We need a “Bill James” for cricket…

And regarding new tools for analysing players how about dot ball % and boundary balls % for bowlers? similarly, balls taken to hit a six/four by the batsman? mccullum has played 464 balls and hit 23 sixes ie a six every 20 deliveries, yuvraj 21 in 159 ie a six in 8 deliveries, morkel 18 in 284 ie 1 in 16...

Gokul,

kindly note that the par strike rate for all 20-20s is not 137.5 as you are assuming on basis of the "safe" 1st innings score of 170.I ran the statsguru and including the 2 matches of sa-aus the average runs-per-over in all 88 t20s is only 7.49 ie nearly 7.5. this amounts to 150 runs in 20 overs and an strike rate of 125. knock off the extras and a strike rate of 120 seems fair enough (par). I agree with krish in that you need both types of batsman- a couple of bats who get 30 odd runs at strike rate for 120 and hold one end up (accumulators) and 3-4 others who get quick 20-25s at strike rate of 140 or more including one batsman with extra ordinary fire power. Example gambhir, dhoni accumulator, sehwag,raina,yuvi,yusuf hitters (the last 2 have huge strike rates and six hitting abilities) morkel and symonds do similar hard-hitting for their teams. continued

An excellent logical analysis. I feel that T20 has made cricket like baseball with respect to statistics, where in there are lots of stats (slugging average, batting average etc) which have no clear meaning. All those stats have to be interpreted in the right manner to arrive at conclusive results, and most of the time the stats don't reveal clear flaws or strengths.

The length of the T20 game means that traditional averages and strike rates and all such stats aren't a reliable judge of team strength or player performance.

Runs saved by the fielding team, dot balls earned by stopping singles and many such factors are vital.

More than forming a list of stats and arranging them in ascending or descending order, we need to logically analyse the situation of each game and come to a conclusion.

I feel this article is a good way of analysing T20. I hope we get more articles that analyze T20, and those that "invent" new tools to measure player or team performance in T20.

would it be appropriate if you also consider the instances (if any) where a team chasing a lesser total won and was in a postive position of chasing a greater total (say winning by chasing 155 in 17 overs with 5 wickets)

Krish,

I disagree. Gambhir averages nearly 30, which means he scores runs in nearly every game that he plays in at a strike rate less than the par-strike-rate. So he is indeed doing some damage by doing that.

Similarly Dhoni averages nearly 24 at a very poor strike rate. This is really surprising to me.

Of course the 137 is a par strike rate when batting first. I believe Ananth's list is the average strike rate for 1st and 2nd innings - so perhaps that makes a difference. I guess Dhoni's 1st innings SR is quite higher than a mere 103.

For anybody scoring more than 10 runs, a strike rate of less than 125 is not good. Overall strike rate which should give a good comfort level for a T20 match should be around 138, i.e. around 168 runs from the bat in 20 overs (assuming 2 overstepings in 20 overs). Total number of extras (around 10) needs to be added to this total. Hence, a total of 178 should be a decently defendable total.

An overview of your batsmen's list raises a pertinent point. Dhoni and Gambhir both vital cogs in India's batting have a less than acceptable strike rate (137.5) but India have gone on to win more than they have lost. This suggests that along with the dashers like Viru and Yuvi, a team does require relatively composed accumulators.

165 in 20 overs means a strike rate of 137.5 (ignoring extras). It'd be interesting to see a list of specialist batsmen who have a strike rate higher than this. Actually a list of batsmen with a strike rate lower than this would also be interesting, specially if their batting average is high. Someone averaging 25 with a strike rate of 110 would be disastrous for a team. Whereas someone averaging 15 with a strike rate of 150 would count as a potential match winner.

[[ Gokul, That is a good point. Since it will be difficult for me to provide you the list here, I have added the same to the main article itself at the end. Ananth: ]]A few people have mentioned about IPL. Let me state my views on IPL. I do not rate IPL as anywhere near the international matches. It is a "commercial child" of our times, mainly meant for television viewing and revenue generation. The alacrity with which the IPL was moved to South Africa proves this. Can you imagine the EPL moving to Portugal and Manchester United playing Arsenal at "home", which is Lisbon or Bilbao. However I would have major problems even in cricketing and technical terms because of the following problems. 1. No country players only club players. 2. Records could be for multiple clubs. 3. There is no clear home for the club. 4. T20 Intl records have no relevance. 5. A player could play multiple teams within a single tournament (probably not now, but who knows). 6. Building a club-based Database which is completely different to my country/player/ground based database. I would have re-write every program I have. I won't do anything in the near future. Probably much later.

This refers to the points raised by Aneesh and Kieran. They have correctly questioned my 90+% figure.

First let me say that the 90+% is based on all instances of chasing team winning, which is 46. Out of these 46, only 4 chases have been of scores of 165 and above. Thus the figure of 90% came in.

However stricly speaking, both Aneesh and Kieran are correct. My sample should be the

teams which crossed 165and not the successful chases. Let me work out that figure below.26 teams crossed 165 (barring the last Saf-Aus match, which has been excluded for sake of consistency). Out of these, 4 teams lost and the other 22 won. So the winning % is 84 and not 90.

Hence I am going to change my Target score to 170, which would lead to 24 wins and 2 losses (win % of 92).

My thanks to Aneesh and Kieran.

This point, because of its importance, has been incorporated in the article itself.

Your calculations don't work. You say that out of 46 successful chases, 4 were against scores of 165 or more. Then you say that means that if you score 165+ you have a 90% chance of winning. That doesn't follow at all. To actually know the chance of winning, compare the number of winning scores of 165+ with the number of losing scores.

As always good Analysis! wonder if you have considered IPL in your analysis. Agreed IPL cannot be classified as international tournament. But given the fact that it has teams drawn from many countries the performances of the teams is emphatically a good measure to be considered for assessing T20 trends.

But I don't think your observations will vary significantly with the inclusion of IPL matches as the teams batting second might have done well over all in IPL as well, which sits confortably in your final analysis. However, I would be interested to see your observations after the inclusion of IPL matches.

Regards, Sampath N

Dear Ananth:

Could you please do some statistical analysis on which is the best IPL team on paper? This would be most interesting since we could how good is the co-relation after the second round of IPL. You could perhaps use a mixture of ODI and T20 data to come up with general stats. Ofcourse, I predict the Superkings would come out on top.

Very interesting analysis. Gives a very good idea of the way a match might turn, the emphasis here being on might as relative strengths of the sides such as batting, bowling and fielding do matter though they cannot be properly quantified

"Out of these 46 matches, the top 4 run chases have been against scores of 165 and above ... In other words, any team scoring 165 and above has a 90+ % chance of winning the game."

That's not quite the right conclusion, Ananth. You mentioned that 42/46 (~90%) chases are under 165. But some of the 42 teams that chased lower targets could well have been on pace to overhaul a 165+ run target. So a team that posts 165 has less than a 90% chance of winning. A better estimator of a "90% safe" target would be to look at all teams that posted ~165 runs, and see what percent of them lost.

Good analysis and well explained. Most teams aim for too much and end up getting too little. 165-170 is definitely what they should be aiming for at the start of the innings.

Great job as always. Don't worry about the Strike 1 against you. Albie does not play for all teams.

Rather than 165, 170 seems to be a score which significantly increase the chances of the win. By the same argument that you gave for going from 165 to 160, going from 170 to 165 also doubles the number of successful chases. The 2 mentioned above and the recent South African chase of 166. Also, mentally it makes a huge difference when you think about 160s and 170s, even though the actual difference is not huge. I feel 170 is the "300" of T20.

interesting analysis done by u but in sports u can only play wid new plans.The stats of previous can't affect the future matches.we can make only observation.

No featured comments at the moment.

interesting analysis done by u but in sports u can only play wid new plans.The stats of previous can't affect the future matches.we can make only observation.

Rather than 165, 170 seems to be a score which significantly increase the chances of the win. By the same argument that you gave for going from 165 to 160, going from 170 to 165 also doubles the number of successful chases. The 2 mentioned above and the recent South African chase of 166. Also, mentally it makes a huge difference when you think about 160s and 170s, even though the actual difference is not huge. I feel 170 is the "300" of T20.

Great job as always. Don't worry about the Strike 1 against you. Albie does not play for all teams.

Good analysis and well explained. Most teams aim for too much and end up getting too little. 165-170 is definitely what they should be aiming for at the start of the innings.

"Out of these 46 matches, the top 4 run chases have been against scores of 165 and above ... In other words, any team scoring 165 and above has a 90+ % chance of winning the game."

That's not quite the right conclusion, Ananth. You mentioned that 42/46 (~90%) chases are under 165. But some of the 42 teams that chased lower targets could well have been on pace to overhaul a 165+ run target. So a team that posts 165 has less than a 90% chance of winning. A better estimator of a "90% safe" target would be to look at all teams that posted ~165 runs, and see what percent of them lost.

Very interesting analysis. Gives a very good idea of the way a match might turn, the emphasis here being on might as relative strengths of the sides such as batting, bowling and fielding do matter though they cannot be properly quantified

Dear Ananth:

Could you please do some statistical analysis on which is the best IPL team on paper? This would be most interesting since we could how good is the co-relation after the second round of IPL. You could perhaps use a mixture of ODI and T20 data to come up with general stats. Ofcourse, I predict the Superkings would come out on top.

As always good Analysis! wonder if you have considered IPL in your analysis. Agreed IPL cannot be classified as international tournament. But given the fact that it has teams drawn from many countries the performances of the teams is emphatically a good measure to be considered for assessing T20 trends.

But I don't think your observations will vary significantly with the inclusion of IPL matches as the teams batting second might have done well over all in IPL as well, which sits confortably in your final analysis. However, I would be interested to see your observations after the inclusion of IPL matches.

Regards, Sampath N

Your calculations don't work. You say that out of 46 successful chases, 4 were against scores of 165 or more. Then you say that means that if you score 165+ you have a 90% chance of winning. That doesn't follow at all. To actually know the chance of winning, compare the number of winning scores of 165+ with the number of losing scores.

This refers to the points raised by Aneesh and Kieran. They have correctly questioned my 90+% figure.

First let me say that the 90+% is based on all instances of chasing team winning, which is 46. Out of these 46, only 4 chases have been of scores of 165 and above. Thus the figure of 90% came in.

However stricly speaking, both Aneesh and Kieran are correct. My sample should be the

teams which crossed 165and not the successful chases. Let me work out that figure below.26 teams crossed 165 (barring the last Saf-Aus match, which has been excluded for sake of consistency). Out of these, 4 teams lost and the other 22 won. So the winning % is 84 and not 90.

Hence I am going to change my Target score to 170, which would lead to 24 wins and 2 losses (win % of 92).

My thanks to Aneesh and Kieran.

This point, because of its importance, has been incorporated in the article itself.