One may love or hate IPL but certainly cannot ignore it. I might complain and crib about many IPL-related things, but the players, their commitment, performances and their give-all attitude makes you forget the shortcomings and watch the same. It is also an analyst's dream. There is a lot of data waiting to be mined.
Cricinfo provides a comprehensive statistical section on IPL. In addition Rajesh has come out with an excellent statistical overview of IPL2. I have tried to make sure that I do not overlap any of the tables available in these two segments. What is given below is also a look into the future of T20 analysis. This has to be different to traditional Test/ODI analyses.
This is possibly the most significant of all analysis I have done for the shortest format. The spark came from a suggestion made by Nicholas who suggested that I look at each over in isolation and analyse bowler performances. I have worked on that concept and looked at each over from 1 to 20. For each of these overs I have determined the total runs conceded (all including extras), average runs conceded, standard deviation and maximum and minimum runs conceded. I have also looked at the total wickets captured (all including Run outs since we are not doing a Bowler analysis), average wickets and standard deviation. The results are fascinating.
- The drop in number of overs for the 5th over is caused by the D/L finish in 4.5 overs by Delhi. Similarly early finishes cause drops in later overs, after 9.
- The first over has been the most economical of all. This is understandable since the batsmen are yet to find their feet.
- Not surprisingly the seventh over is also as economical as the first. Immediately after the fielding restrictions are removed there is a lull in the scoring rate.
- Note also the sharp drop in scoring rate in the 10th over possibly indicating that the batsmen are playing carefully in anticipation of the strategy break. So it looks as if the strategy break, while filling the IPL/Sony coffers, is proving to be a dampener in scoring.
- Afterwards the rate builds up to a high 10+ runs per over in the 20th overs.
- Can anyone throw a light on the significant dip in the 19th over. It is not because quite a few matches finished in the 19th over since the calculation is based on exact number of balls rather than no of overs. Possible reason might be that very few runs are needed for a win and teams do this carefully.
- The Standard Deviation/Average ratio seems to be quite high during the first 6 overs indicating significant variations. Then it seems to settle to similar levels. However it should be noted that the later overs have high averages.
- The first over has seen a few more wickets fall than the next 14. Again possibly the bowlers are getting the batsmen before they find their feet.
- The 8th over wickets per over value is the lowest at 0.15. Possibly because the teams are still in a Power Play mood in the 7th over and take an over to settle down.
- The 16-19 overs are approximately similar until there is an explosion of wickets in the 20th over, all particularly understandable.
2. IPL2: 20+ Runs overs bowled
No MtId Player Name For Vs Over
1.0048 Singh R.P/H'meet DC KKR 19 28 runs
2.0048 Mashrafe Mortaza KKR DC 20 26 runs
3.0050 Patel M.M RR DD 19 25 runs
4.0057 Sehwag V DD DC 6 25 runs
5.0009 Salvi A.M DD CSK 7 24 runs
6.0022 Kamran Khan RR CSK 18 24 runs
7.0034 Singh V.R.V KXP CSK 15 24 runs
8.0043 Shoaib S.M DC DD 6 24 runs
9.0030 Sreesanth S KXP RR 19 23 runs
10.0056 Singh R.P DC RCB 20 23 runs
11.0043 Vaas WPUJC DC DD 20 23 runs
12.0001 Flintoff A CSK MI 18 22 runs
13.0010 Patel M.M RR KKR 16 22 runs
14.0034 Sreesanth S KXP CSK 17 22 runs
15.0051 Tyagi S CSK KKR 6 22 runs
16.0053 Hodge B.J KKR RR 1 22 runs
17.0009 Flintoff A CSK DD 19 22 runs
18.0056 Kumar P RCB DC 5 22 runs
19.0027 Dinda A.B KKR KXP 4 22 runs
20.0027 Pathan I.K KXP KKR 20 21 runs
21.0042 Sreesanth S KXP MI 5 21 runs
22.0031 Nannes D.P DD KKR 3 21 runs
23.0057 Nannes D.P DD DC 1 21 runs
24.0024 Abdulla Y.A KXP RCB 18 21 runs
25.0008 Kumble A RCB DC 13 20 runs
26.0005 Kallis J.H RCB CSK 5 20 runs
27.0016 Shoaib S.M DC CSK 18 20 runs
28.0036 Sreesanth S KXP DC 17 20 runs
29.0036 Symonds A DC KXP 9 20 runs
30.0037 Harwood S.M RR CSK 16 20 runs
31.0041 Akhil B RCB KKR 17 20 runs
32.0041 Kallis J.H RCB KKR 20 20 runs
33.0016 Gony M.S CSK DC 2 20 runs
34.0021 Mishra A DD DC 13 20 runs
35.0059 Kumar P RCB DC 15 20 runs
First thing to be noted is that these are "score advance" runs, including all extras. RP Singh and Harmeet Singh share the worst over in IPL2, although RPS was indirectly responsible with his beamer-ban from the attack, it was Harmeett who conceded most of the runs. It is another story that Harmeet had a wonderful final. Then Mashrafe's ill-fated last over to Rohit Sharma, followed by Sehwag's desperate attempt to stifle Gilchrist and Patel's forgettable attempt against Deccan.
Sreesanth conceded 20 runs in an over in each of the first four matches he played after returning. Instead of concentrating on bowling he engaged in verbal jousts with great batsmen and paid the price. Once he settled down and let the ball instead of his mouth do the talking, he improved in the next three matches.
There have been 11 20+ over occurrences in the first 6 overs, only 3 occurrences in the overs 7-14 and 21 in the 15-20 overs. All perfectly logical figures.
Surprisingly Kolkata is the best team in this regard having conceded 20 runs only thrice. Chennai and Rajasthan follow with 4 each. Punjab is the worst offender in this, having done this no fewer than 7 times.
At the other end, Deccan batsmen have done this to the bowler 9 times, followed by Chennai and surprisingly Kolkata with 7 occurrences. Four teams, Mumbai, Rajasthan, Punjab and Rajasthan have achieved only 2 times each.
3. IPL2: Analysis of 10-Over scores - Runs scored in 10 overs
SNo. I MtId Team Vs @ 10 overs Final Result % Runs
1. 1 0017 MI KKR 111 for 0 187 for 6 Won 1.68 76
2. 2 0009 CSK DD 106 for 2 180 for 9 Lost 1.70 74
3. 1 0005 CSK RCB 106 for 0 179 for 5 Won 1.69 73
4. 2 0057 DC DD 102 for 3 154 for 4 Won 1.51 52
5. 2 0055 DD MI 93 for 1 166 for 6 Won 1.78 73
6. 1 0030 RR KXP 93 for 0 211 for 4 Won 2.27 118
7. 1 0008 DC RCB 91 for 2 184 for 6 Won 2.02 93
8. 1 0009 DD CSK 90 for 3 189 for 5 Won 2.10 99
9. 2 0034 KXP CSK 90 for 3 174 for 3 Lost 1.93 84
10. 1 0016 CSK DC 88 for 2 165 for 6 Lost 1.88 77
...
...
...
109. 2 0053 KKR RR 38 for 5 102 for 6 Won 2.68 64
110. 2 0002 RR RCB 32 for 5 58 for 10 Lost 1.81 26
111. 1 0004 KKR DC 31 for 3 101 for 10 Lost 3.26 70
Four teams, Chennai twice included, exceeded 100 runs in the first 10 overs. The first three teams lost the way in the next 10 overs. Gilchrist's blitzkrieg was the last instance. A few teams scored below 40 runs in the first 10 overs. One of the teams, Kolkata, recovered to win the match chasing a low total.
4. IPL2: Analysis of 10-Over scores - Ratio of final score to 10-over score
SNo. I MtId Team Vs @ 10 overs Final Result %
1. 1 0004 KKR DC 31 for 3 101 for 10 Lost 3.26
2. 1 0048 KKR DC 50 for 1 160 for 5 Lost 3.20
3. 1 0027 KKR KXP 52 for 1 153 for 3 Lost 2.94
4. 1 0028 MI RCB 53 for 3 149 for 4 Lost 2.81
5. 2 0008 RCB DC 57 for 3 160 for 8 Lost 2.81
6. 2 0047 CSK MI 56 for 2 151 for 3 Won 2.70
7. 1 0021 DC DD 55 for 4 148 for 9 Lost 2.69
8. 1 0035 MI DD 43 for 4 116 for 10 Lost 2.70
9. 2 0053 KKR RR 38 for 5 102 for 6 Won 2.68
10. 1 0056 RCB DC 64 for 2 170 for 4 Won 2.66
...
...
...
109. 2 0004 DC KKR 69 for 2 104 for 2 Won 1.51 35
110. 2 0017 KKR MI 70 for 2 95 for 10 Lost 1.36 25
111. 2 0029 DC CSK 79 for 5 100 for 10 Lost 1.27 21
Kolkata leads in the second half recoveries, almost tripling the score three times.
However it can be seen that most of the matches in which there have been very good recoveries after 10 overs have been lost. Similarly most of the matches in which teams started very well have been won, despite not maintaining the momentum. This indicates that it is very important to start well in the 10 overs. Second half recoveries seem to be like locking the doors after the horses have disappeared.
5. IPL2: Analysis of 10-Over scores - Runs added in second 10 overs
SNo. I MtId Team Vs @ 10 overs Final Result Runs
1. 1 0030 RR KXP 93 for 0 211 for 4 Won 118
2. 1 0048 KKR DC 50 for 1 160 for 5 Lost 110
3. 1 0056 RCB DC 64 for 2 170 for 4 Won 106
4. 2 0041 RCB KKR 70 for 2 176 for 4 Won 106
5. 2 0051 KKR CSK 83 for 1 189 for 3 Won 106
6. 1 0051 CSK KKR 83 for 2 188 for 3 Lost 105
7. 1 0034 CSK KXP 80 for 2 185 for 3 Won 105
8. 1 0041 KKR RCB 69 for 2 173 for 4 Lost 104
9. 2 0008 RCB DC 57 for 3 160 for 8 Lost 103
10. 1 0001 MI CSK 64 for 1 165 for 7 Won 101
11. 1 0027 KKR KXP 52 for 1 153 for 3 Lost 101
...
...
...
107. 1 0053 RR KKR 61 for 5 101 for 9 Lost 40
108. 2 0005 RCB CSK 56 for 5 87 for 10 Lost 31
109. 2 0002 RR RCB 32 for 5 58 for 10 Lost 26
110. 2 0017 KKR MI 70 for 2 95 for 10 Lost 25
111. 2 0029 DC CSK 79 for 5 100 for 10 Lost 21
Rajasthan, led by Smith, added the maximum runs, 118 to be precise, during the second half. Kolkata added 110 runs and three teams added 106 runs.
6. IPL2: Analysis of maidens bowled
No Bowler Team Mat Overs Mdns % Mdns
1 Harris R.J DC 8 31.0 2 6.45
2 Malinga S.L MI 13 49.3 2 4.04
3 du Preez D RCB 2 7.0 1 14.29
4 Mascarenhas A.D RR 6 19.2 1 5.17
5 Lee B KXP 5 20.0 1 5.00
6 Edwards F.H DC 6 23.2 1 4.29
7 Abdulla Y.A KXP 9 28.0 1 3.57
8 Patel M.M RR 12 34.5 1 2.87
9 Morkel J.A CSK 12 40.0 1 2.50
10 Mishra A DD 11 42.0 1 2.38
11 Chawla P.P KXP 14 44.5 1 2.23
12 Kallis J.H RCB 15 46.0 1 2.17
13 Sangwan P DD 13 46.4 1 2.14
14 Kumar P RCB 13 49.4 1 2.01
15 Muralitharan M CSK 13 50.0 1 2.00
16 Pathan I.K KXP 14 50.2 1 1.99
17 Nehra A DD 13 51.0 1 1.96
18 Kumble A RCB 16 59.1 1 1.69
19 Singh R.P DC 16 59.4 1 1.68
A total of 21 maidens were bowled by 19 bowlers. Only Malinga and Harris bowled two each. Can any one forget Malinga's famous last-over double-wicket maiden against Deccan. Similarly no one can ever forget Harris' first over maiden in which he captured 2 wickets which paved the way for the semi final win over Delhi. Also his other maiden was again the first over against Bangalore in the final. Note also du Preez's bowling a maiden in the 7 overs he bowled.
7. IPL2: Highest Boundary % share in batsman innings (>66.7% in 50+ score)
This table analyzes the % of runs scored in boundaries in an innings. Only innings exceeding 50 are considered. Gilchrist's semi-final masterpiece leads the pack with an astounding 82.4% in boundaries. Dilshan's 80% in his knock of 50 separates another Gilchrist blitz. Yusuf Pathan and McCullum follow next with over 75%.
8. IPL2: Defending wins
No. MtId FBt Score Vs Score
1. 0030 RR 211 for 4 KXP 133 for 8
2. 0009 DD 189 for 5 CSK 180 for 9
3. 0017 MI 187 for 6 KKR 95 for 10
4. 0034 CSK 185 for 3 KXP 174 for 3
5. 0008 DC 184 for 6 RCB 160 for 8
6. 0005 CSK 179 for 5 RCB 87 for 10
7. 0029 CSK 178 for 3 DC 100 for 10
8. 0043 DD 173 for 7 DC 161 for 10
9. 0056 RCB 170 for 4 DC 158 for 6
10. 0012 DC 168 for 9 MI 156 for 7
11. 0040 DC 166 for 7 RR 113 for 10
12. 0001 MI 165 for 7 CSK 146 for 7
13. 0022 CSK 164 for 5 RR 126 for 10
14. 0026 CSK 163 for 10 DD 145 for 8
15. 0038 MI 157 for 2 RCB 141 for 7
16. 0050 DD 150 for 3 RR 136 for 9
17. 0010 RR 150 for 6 KKR 150 for 8
18. 0023 MI 148 for 6 KKR 139 for 6
19. 0045 RR 145 for 7 MI 143 for 10
20. 0024 RCB 145 for 9 KXP 137 for 7
21. 0032 DC 145 for 6 MI 126 for 8
22. 0059 DC 143 for 6 RCB 137 for 9
23. 0015 KXP 139 for 6 RR 112 for 7
24. 0049 KXP 134 for 7 DC 133 for 8
25. 0002 RCB 133 for 8 RR 58 for 10
26. 0020 KXP 119 for 8 MI 116 for 7
27. 0054 CSK 116 for 9 KXP 92 for 8
9. IPL2: Chasing wins
No. MtId FBt Score Vs Score
1. 0051 CSK 188 for 3 KKR 189 for 3
2. 0041 KKR 173 for 4 RCB 176 for 4
3. 0011 RCB 168 for 9 KXP 173 for 3
4. 0036 DC 168 for 5 KXP 169 for 7
5. 0016 CSK 165 for 6 DC 169 for 4
6. 0055 MI 165 for 8 DD 166 for 6
7. 0048 KKR 160 for 5 DC 166 for 4
8. 0006 KXP 158 for 6 KKR 79 for 1
9. 0031 KKR 154 for 3 DD 157 for 1
10. 0027 KKR 153 for 3 KXP 154 for 4
11. 0057 DD 153 for 8 DC 154 for 4
12. 0028 MI 149 for 4 RCB 150 for 1
13. 0014 RCB 149 for 7 DD 150 for 4
14. 0021 DC 148 for 9 DD 150 for 4
15. 0047 MI 147 for 5 CSK 151 for 3
16. 0058 CSK 146 for 5 RCB 149 for 4
17. 0018 DD 143 for 7 RR 147 for 5
18. 0025 DC 141 for 5 RR 142 for 7
19. 0037 RR 140 for 7 CSK 141 for 3
20. 0019 KKR 139 for 6 RCB 143 for 5
21. 0052 DD 134 for 7 RCB 135 for 3
22. 0044 CSK 129 for 10 RCB 132 for 8
23. 0039 KKR 123 for 8 DD 125 for 3
24. 0046 DD 120 for 9 KXP 123 for 4
25. 0042 KXP 119 for 9 MI 122 for 2
26. 0035 MI 116 for 10 DD 118 for 3
27. 0033 RCB 105 for 10 RR 107 for 3
28. 0003 KXP 104 for 7 DD 58 for 0
29. 0053 RR 101 for 9 KKR 102 for 6
30. 0004 KKR 101 for 10 DC 104 for 2
The above two tables are to be seen together.
21 teams exceeded 160 runs and two thirds of these totals were defended. On 7 occasions these totals were chased. This indicates that a good par score for the tournament was 160 which gave a 66.7% chance of winning. If we look at 150+ totals reached by 28 teams, 17 defended the totals while 11 chased successfully. The success % drops to 60%. So I will stick to 160, which is likely to be the par score at England during the T20 World Cup also.
8 teams which won defending their scores have dismissed the other teams, mostly for low scores. Only one team which was all out defended its total. On the other hand all 4 teams which were all out have lost their matches. Thus out of 13 all-out situations, 12 lost their matches. Most of the dismissals have been for sub-140 scores. Again perfectly understandable.
Brilliant analysis. I am just fascinated by the diff ways in which the data was pulled out. I have a suggestion though. The first table analysis for each over could be seperated by the innings and it would lead to some more insight as to how teams performed. [[
Shripad
What you suggest would give an insight into the way the overs have been bowled in the first innings (the setting target innings) and the second innings (the chasing one). Will look at it certainly for the T20 analysis which will precede the World Cup. There also 89 matches for T20, which gives a good sample for the individual innings.
Ananth:
]]
Posted by: David at May 27, 2009 4:53 PM
You mention that T20 opens up different kinds of statistical analyses. One that is possible given the small no. of overs involved is to compare individual overs to the total innings. Eg, in IPL2 match 53 (KKR v RR)over #1 for RR netted 22 runs out of an innings total of 9/101. That is 21.8% of the innings score achieved in 1 over! Was that the greatest % contribution of a single over in IPL2 (excluding D/L shortened games)? [[
David
Good point. When I saw snatches of the match (I would go mad if I saw all matches completely), I missed the 22 run over otherwise it might have looked out-of-the-way to me. But your comment raises the point that it has to be an automated process. Will look at it.
Thanks.
Ananth:
]]
Posted by: ramnath at May 27, 2009 5:24 PM
u are once again engaging urself in a futile exercise. The statistical analysis for T20s doesnt make sense because the game is more random and determined heavily by chance
Posted by: Ashik Uzzaman at May 27, 2009 5:34 PM
I got a summarized overview at one shot from you. Thanks.
Posted by: Sid at May 27, 2009 5:52 PM
Godd Analysis. But its Harmeet Singh not Hrpreet Singh
Posted by: rukrimo at May 27, 2009 6:24 PM
Nice work. I am more interested in individual performances. Qualitative comparison has been made between foreign and Indian players and there is always a dissmissive suggestion that Indian players struggled. S. Asnodkar, K. Goel, A. Rane, Badrinath among others are often cited. My feeling was that a lot of them especially those in the middle and lower middle orders performed decently, contributed handily and perished trying to salvage impossible situations. Is there a metric that assigns contextual significance to an individual's performance?
Posted by: murtaza at May 27, 2009 6:58 PM
Table 2: You mistakenly mention KKR as the best team but it is actually MI as none of their bowlers conceded a 20+ over. Pity their batsmen couldn't live up to the good work done by their bowlers.
Posted by: as at May 27, 2009 8:16 PM
If you don't watch matches, you should not do analysis. [[
Thanks for your insight. I don't have to watch the scantily clad cheerleaders dancing, endless ads pushing mobiles and listen to inane commentaries to do my analysis. For your information I watch Test matches completely.
Ananth:
]]
Posted by: Ajit at May 27, 2009 9:21 PM
Great analysis.
Contrary to popular belief, over 11 is neither a very slow scoring one, nor does it yield more wickets than the other overs.
Incidentally, over 8 has the fewest # of wickets among all while not being a low scoring one. Could being smack in the middle of the end of the powerplays and the break have anything to do? Interesting.
Posted by: Sohail at May 27, 2009 9:28 PM
Brilliant Analysis! It figures is my favourite section in cricinfo!
If we can find out the % of boundaries scored for each over, both with respect to the boundaries and with respect to number of runs then I think it will provide some more insight into scoring pattern!
Posted by: James at May 27, 2009 11:14 PM
Interesting stuff. My theory on the 19th over scoring reduction is: that over is probably going to be bowled by your top fast bowler. It would be interesting to see how many times the bowler who bowled overs 1 and probably 3, also bowled #19. It would be at the same end as over #1 to keep the bowler comfortable, and while you may assume you want your best bowler bowling the last over, he may need to bowl the 19th over if the run chase is tight for example. Otherwise the match could be over by the 20th. And if the team is defending first it wouldn't matter which of the last 2 overs he bowled, so it's even more likely your top bowler will bowl the 19th. [[
James
That is as good an explanation as we are likely to get. It also depends on how many overs the top bowler has left. For instance in the final, RP Singh had two overs left, so he bowled the 18th and 20th.
Ananth:
]]
Posted by: ravikant at May 28, 2009 2:23 AM
Ananth, Your analysis is superb.Congratulations!
You have mentioned that "the first over has been the most economical of all"(and so is the seventh)and given the reasons.In that case,does it make a difference if a spinner bowls the first over or a pace bowler bowls the first over??The convention usually followed is that-a pace bowler opens the innings.If the first over bowled by a spinner(in contrast to a pace bowler)yields the least runs,then captains may reserve their pace bowlers for the last overs and use spinners for the first and seventh overs.In T20, as each bowler can bowl only 4 overs,this tactic might give an additional edge for the bowling side.Your thoughts on this Ananth? [[
I have not done the Spin/Pace classification, that has been done by Rajesh. My own take is that the best bowling idea is what you have suggested. Bowl a spinner when the batsman least expects it (e-g, Kumble's first over in the final). Also it is a great idea to deny the batsmen the pace, in other words bowl 3 of the first six with spin. One captain changed the bowler every over upto 7 or so. Again a great idea, if your bowler(s) can take the rotation.
Ananth:
]]
Posted by: David at May 28, 2009 2:34 AM
I disagree entirely with Ramnath. I actually think T20 is more predictable in certain ways than ODIs. Because it's so condensed, one outstanding contribution with either bat or ball (but especially bat) can shut the opposition out of the game. Compare Gilly's 85 in the semi-final, where the game was decided after the 4th over when the RRR was about 6 and there were 9 wickets in hand. On the other hand, WI began the 2006 Champions trophy final outrageously quickly (9.3 ov, 80/2) and then lost 8 wickets for not many over the next 21.1 overs, handing Aus the match. If WI had batted similarly in a T20 context, they would have probably ended up with a slightly below-par, but very defendable total (it's too complicated to give the sums here, but a reasonable estimate would be 122). Aus absorbed the initial onslaught and came back to win the match. T20 doesn't leave time for such a recovery, so the game ends up being much more predictable.
Posted by: uglyhunK at May 28, 2009 9:15 AM
David, you got it totally wrong. When it is said that T20 is unpredictable, it means you can't surely pick a winner before the match. Any team can win the match. This is not the case with ODIs and Test matches. You can with high percentage of success, predict who can win the game.
Posted by: Arvind at May 28, 2009 10:13 AM
Hi Ananth
Very nice. I took the liberty to do some further analysis. I didnt get beyond the first table you provided though ;).
I estimated the correlation between 8 different variables from the first table.
There is high correlation between Avg. Runs scored and Avg. Wickets fallen in an over. Somewhat paradoxical in first instance. I have put the figure on by blog page
right arm over
Arvind
Arvind
I am pssing your commeny without perusing your conclusion. I am sure your correlation findings are correct. Let me see it later.
Ananth
Posted by: Chawlaaa at May 28, 2009 10:42 AM
guys they dont publish my comments at all...because i am a journalist myself and i criticise them acccordingly! they site is bias I am sorry to say!!! [[
Many responses containing adverse comments have been published. Kindly remember that your comments have to conform to acceptable norms and avoid personal attacks. Also you must avoid unparliamentary words. I will gladly publish if you conform to these guidelines.
Ananth:
]]
Posted by: David at May 28, 2009 11:31 AM
Uglyhunk, I think you're not paying close enough attention. First, even in IPL2 there were consistent trends ... KKR and DD for starters, but also CSK and MI. There's no difference between their consistency and that of any international ODI team. Second, T20 results may appear unpredictable (eg, DC 2008 v DC 2009), but that's partly a function of the newness of the format rather than its inherent unpredictability. As teams learn how best to play it, so the good and bad teams will be seen for what they are. Third, if there is any unpredictability in T20, it is due to the franchise nature of the IPL (you never know how players from all over the world will perform) rather than the format itself. When we revert to the international form, it will be as predictable as ever ... just look at the top four in the last T20 world cup: Ind, Pak, NZ, Aus ... all consistently good limited overs countries, with SA and SL close behind & Eng and WI well out of the picture. (Cont'd...)
Posted by: David at May 28, 2009 11:32 AM
… Cont’d.
In fact, if you compare the last ODI world cup with the T20 edition, the results of the ODI world cup were far less predictable: who would have imagined that both Ind and Pak would be knocked out in the 1st round. But after that, who was left? The same top 4 teams as the T20 WC: Aus, SL, NZ, SA.
Posted by: Mike at May 28, 2009 1:39 PM
"Can anyone throw a light on the significant dip in the 19th over" - best way to consider this is to do two distinct tables, one for teams batting first, the second for teams chasing. This will show if the slow-down is due to chasing teams taking more care as they approach their target. [[
Mike
Shripad had already suggested this and you are right. I will do this for the T20 analysis since some additional work has to be done. Separating by innings will also let us derive quite a few other conclusions.
Ananth:
]]
Posted by: Ravi Shankar at May 28, 2009 4:54 PM
Your first table also demolishes the frequent (always if I may say so) refrains by the commentary team that since it is the 11th over a wicket should fall. The 11th seems to be no different to many others. The only (theoretical and not statistical) argument could be that after the break batsmen could have been asked by their teams to get a move on, which can be seen if we can compare first 10 over scores of various innings and then see wickets fall in the 11th if the total is well below average. This may also explain the steep increase in fall of wickets in the 16th as compared to 15th. Somehow last 5 overs makes batsmen think that they need to get a move on.
Posted by: ramesh at May 28, 2009 5:53 PM
great work!! how about a comparison with IPL 1, especially table 1,3 and 5 .. while IPL 1 was generally higher scoring, it would be interesting to see the trend in avg. score and wickets and the impact of the 'strategy' break would be clearer .. [[
Ramesh
As of now I do not have IPL1 data and am not sure whether I will do that work. However what you say is possible with future tournaments such as T20 WC, T20 Intls itself, Champion's League, IPL3 etc.
Ananth:
]]
Posted by: Nifty at May 28, 2009 7:04 PM
A small mistake:
Sreesanth conceded 20 runs in an over in each of the first four matches he played after returning.
Surprisingly Kolkatta is the best team in this regard having conceded 20 runs only thrice.
If someone has already pointed this out, my apologies - please disregard. [[
No it is not a mistake. KKR, the best team in this regard, conceded only 3 20+ run overs while others conceded MORE number of times and KXP falls into this group.
Ananth:
]]
Posted by: Pelham Barton at May 28, 2009 8:29 PM
The simplest explanation for the dip in the run rate in the 19th over is random variation. I do not know in what sense you are using the word "significant", but the difference between 18th and 19th over run-rates is not statistically significant at the 5 percent level, which is usually taken as the least demanding criterion for significance. The interesting comparison would be not to separate the two innings, but to see if the same thing happened last year. Similarly the difference between 9th and 10th over scoring rates is not statistically significant. It may be that the dip there is genuine, and you have a plausible reason for it, but this year's figures are not strong enough to settle the issue with any certainty. Overall your figures show a strong pattern of varying run rate through an innings, but you should not try to read too much into every variation before asking how much things could be expected to vary as a result of chance alone. [[
Pelham
I still think 9.60 (18), 8.98 (19), 10.01 (20) presents an unexpected dip in an over when teams would be expected to go hammer and tongs. The dip is of the order of 7% and then the increase is of the order of 13%. It is possible it falls into the random variations you have suggested. It is also possible that captains tend to bowl their best bowler in the 19th bowler instead of holding them back to the 20th, as a few readers have pointed out.
However I take your point on taking one year's figures. Let us see what happens when I do the T20 Intl analysis. Adter all there are 89 matches there spread over 5 years and a few continents.
Ananth:
]]
Posted by: R. Simha at May 28, 2009 10:06 PM
Any stats to examine how teams changed their game as the tournament progressed? Who got more comforable, who got cautious, who got desperate... and of course, what does a steady dose of T20 do to a player?
Posted by: Hiren at May 29, 2009 4:56 AM
Great Analysis!
While its great that you have shared the stats directly with us and given how spoilt we are with analysis output, would it not be greater if you could chart out some of the trends? E.g. the dip in 19th over, numbers may indicate different insights, but when I put the RR data on a line graph, I had no doubt that the dip is "significant" enough.
Secondly, while there could be various cuts you can take of the data, maybe one could also look at Share of 4's & 6's in the aggregate score (or the entire split from dots to 6's) of the leading 20-30 batsmen (by score)
Thanx!
Posted by: MP at May 29, 2009 6:25 AM
The drop in scoring in the 19th over could be because in some instances the team has lost many wickets and so the remaining batsmen are bowlers who are not swinging wildly as the opportunity cost of getting all out is higher (as you miss the opportunity to score off the remaining 6-11 balls) than if you were in the final over (when the opportunity cost falls).
Posted by: Arjun N at May 29, 2009 7:28 AM
a)
[[
No it is not a mistake. KKR, the best team in this regard, conceded only 3 20+ run overs while others conceded MORE number of times and KXP falls into this group.
Ananth:
]]
Did you forget about MI ?
b) Is it possible to have some data on bowlers and batsmen who played in the particular over and to know if the lived up to their strike rate/average ?
It may be that tailenders faced most of the 19th over and saved their wicket for 20th over or also one top order player saved his wicket for the 20th over there by a considerable dip in 19th over score.
Posted by: ramlal at May 29, 2009 1:20 PM
Hi Ananth,
This is good analysis as usual.
I think it would also be interesting to know how many 'dot'balls did a batsman play ( total / per match), something similar to seeing how many maidens did a bowler bowl.
Posted by: Aneesh at May 29, 2009 6:49 PM
Hi Ananth,
You posted the standard deviation for runs scored in a single over. It might be helpful to also compute the standard deviation for the average runs per over (standard error of the mean). This will be much smaller (about 0.4), and more useful for comparison.
Here is another possible factor that might explain lower 19th over numbers. The very strong teams who score very fast will complete the chase before the 19th over itself. Your data show significantly less balls faced in the 19th over. So only the weaker teams were still batting in the 19th over, and hence the rate is lower. In the 20th over, the teams just start swinging (nearly twice as many wickets fell in the 20th vs the 19th over), so the scoring rate rises.
I also think it's interesting to look at the 1st and 2nd innings separately. I examined this data on my blog, and saw some differences in the run-scoring patterns between the two innings.
[[
Aneesh
For the T20 analysis I have already separated the data between the two innings and have seen a different run-scoring pattern emerging. This article is expected to come out by mid-next week just before the T20 World Cup.
Re SD I have mailed you separately eliciting some clarification.
Ananth:
]]
Posted by: Jonathan at May 31, 2009 8:09 PM
Other interesting stats:
Average RPO was just 7.20 at Cape Town (lowest), compared with 7.97 at Centurion (highest). Average RPW was 18.01 at Cape Town compared with 26.56 at Centurion. This equates to an average 20-over score of 159-6 at Centurion compared to 144-8 at Cape Town. (Not enough games played at Bloemfontein and East London to draw firm conclusions.)
Surprisingly Johannesburg, anticipated to be a heavy-scoring ground, averaged 7.32 runs an over and no totals over 166.
In daytime games average RPO was 7.68, average RPW 25.59, equating to an average 20-over score of 154-6.
In evening games RPO is 7.39, RPW 21.12, equating to an average 20-over score of 148-7.
Of the 13 occasions a team was dismissed, 11 were in night games.
Teams chasing under lights averaged just 7.15 runs per over.
Teams batting first scored 0.35 more runs an over than teams batting second, 7.65 to 7.30, and averaged just under one run extra per wicket.
Y Anantha Narayanan has over 35 years of IT background. Over the past 15 years, he has been concentrating on Cricket analysis and software development. He has been involved with StumpVision, Wisden, Hallmark Software and his own site www.thirdslip.com during this period.
David Barry was cricket-starved when teaching English in France, and
study of cricket stats was his only way to stay sane. He is now back
in Brisbane, Australia, and working towards a PhD in Physics. He once
played for the worst team in the G-division of Muscat's cricket
league.
After doing an MBA in marketing and working in an advertising agency, S Rajesh decided that his skills might be put to better use by number-crunching on cricket. He hasn’t regretted that decision in the last six years, and edits the Numbers Game column on cricinfo.com every Friday.
Andrew Samson had his moments with bat and ball, once scoring 43 and taking 3 for 14 with his legbreaks, but he was much better at arithmetic, which explains why he is where he is today. Andrew has been keeping cricket stats since the days when it used to be done with pen and paper, and has been involved in scoring/stats for Radio and TV since 1987. He has been Cricket South Africa's official statistician since1994.
A former scientist and occasional TV quiz champion, Charles Davis now works full time at sports statistics in Melbourne.
His only real contribution to the Test record books came at age 4, when he formed part of the record 90,800 crowd
who saw West Indies at the MCG in 1961. He has two books to his credit, and claims to be the only cricket statistician
ever who has been quoted in the New York Times and in Australian Federal Parliament on the same day. Not to be
confused with the West Indian batsman Charlie Davis, especially in terms of ability.
Having just taken early retirement as a Mathematics teacher in Hobart, Ric
Finlay now fully devotes his time to recording cricket, both past and
present, for the popular CSW cricket database, along with his colleague
David Fitzgerald (www.tastats.com.au). His interest in the game is
inversely proportional to his ability as a player, but he did once score a
century after being dropped at 3 and running out three of his team-mates.
His first memory of international cricket is the 1962-63 MCC tour of
Australia, described as one of the most boring ever. Totally fascinated, he
was instantly hooked, and has never looked back. Author of three books on
cricket of a historical nature, he has provided statistics and scored for
radio and television cricket coverage since 1983.
RSS Feed
