ODI Runs and their relationship with Match Results……

This thread has its roots in Kumar Sangakkara’s comment on Cricinfo about the relationship between Sanath Jayasurya making runs and Sri Lanka winning ODI games. For a long time, there has been a perception that Tendulkar hasn’t always had that match winning touch – and on the face of it, this appears to be true. 4 of Tendulkars 5 ODI centuries against Pakistan have resulted in defeats (contrast this with two other innings of 98 and 95 resulting in wins). These instances are anecdotal at best. The following table shows a comparison of the top ODI batsmen in the world – Tendulkar, Jayasurya, Lara, Ponting, Gilchrist, Kallis, Yousuf and Dravid – all having played over 200 ODI games. It seeks to describe the following:

1. How consistent are these players?
2. How does the number of runs they make affect the outcome?
3. Is there a difference in the way an openers runs affect the outcome, when compared to a middle order player?

(Please click on the table to see a bigger view)

The table compares the improvement in win % when each batsman makes more than 30, more than 50, more than 70 and more than 100, with the base win % for that batsman in all the games that he has played in. The Consistency field gives the percentage of total innings played in which a scored in excess of 30, 50, 70 or 100 was achieved.

The results show that opening the batting is not that different from batting in the middle order when it comes to influencing results. Tendulkar as opener has been the most consistent batsman in ODI cricket, for the longest period of time. In terms of a non-failure with the bat (scoring more than 30) affecting the outcome, Gilchrist, Ponting, Inzamam and Kallis have the minimum impact on the result, while Lara has the maximum impact on the result. Kallis’s runs don’t seem to matter very much, because South Africa win less often when he makes atleast 50 than they do when hes in the side. This is characteristic, as South Africa’s great strength has been the depth and quality in the lower middle order. The Australians are a strong outfit, and hence Ponting and Gilchrist’s runs matter less to Australia than Lara’s runs matter to West Indies or Tendulkar’s runs matter to India. Big runs from the openers seem to have a more telling impact on the game than big runs from the middle order – Tendulkar, Gilchrist, Jayasurya all affect the outcome more positively when they make hundreds than when they make 50’s. Lara and Dravid do so as well, but both Lara and Dravid have played in net unsuccessful teams (teams which have lost more than they have won). The batting position probably matters less than the general strength of the side in which the batsman plays.

CricketingView

Cricinfo’s Statsguru….

Statsguru on Cricinfo.com is one of the best, if not the best stats databases available online. One of the stats in this database has always intrigued me. It is the “bowler/fielder summary” in the list of batting stats. This provides a list of bowlers that a batsman has been dismissed by, and a run average for which this happens. I used to think at first, that when it said that Batsman X averaged Y against Z bowler, it is a measure of the number of runs the batsman has made off bowler Z per dismissal by bowler Z.

Not so. It is in fact, the number of runs made by a batsman in innings when he was dismissed by a particular bowler. So it does not provide a measure of a bowler’s mastery of a batsman. Take the example of Glenn McGrath and Sachin Tendulkar. McGrath has dismissed Tendulkar 6 times in Test cricket, and Tendulkar’s average against him is 22.16. This is Tendulkar’s average in all the innings in which he has been dismissed by McGrath. If you consider all the test matches that Tendulkar and McGrath have both featured (which obviously includes all the innings in which McGrath has bowled at Tendulkar in a Test match), Tendulkar’s scores have been

11, 0, 61, 0, 116, 52, 45,
4, 76, 65, 10, 10, 126,
17, 8, 2, 5, 55

So Tendulkar’s average in innings where McGrath bowled at him is 36.8.

The point is, that the 22.16 average appears in Tendulkar’s statsguru list, not McGrath’s. 22.16 is not a measure of Tendulkar’s performance against McGrath, 36.8 is. Similarly, McGrath’s “batsmen dismissed” results should list 36.8 vs Tendulkar. If the statistic is to be a real McGrath v Tendulkar statistic, then it must list the number of runs Tendulkar has made off the bowling of Glenn McGrath (in his innings against Australia) and divide that figure with the number of times McGrath has dismissed Tendulkar. That would be the truest McGrath v Tendulkar statistic. The current statistical measure is inaccurate, assuming that the purpose of the statistic is to provide a measure of a batsman’s performance against a particular bowler. This is important because statistics invariably cause the reader to find conclusions why they do not actually support, but seem to support.

Brian Lara has a 41.40 average in 15 innings in the “bowler’s dismissed by” result against Glenn McGrath’s name. In Test Matches where both Lara and McGrath have featured, Lara’s scores are as follows:

65, 9, 88, 43, 24, 14*,
65, 0, 26, 44, 2, 1, 2,
2, 9, 78, 132, 62, 3,
213, 8, 153*, 100, 7 , 0,
4, 0, 17, 182, 39, 16, 0,
35, 28, 14, 42, 68, 60,
30, 14, 13, 45, 226, 17

Lara has made 2000 runs in 42 complete innings in Tests where McGrath has played, an average of 47.16.

Lara’s average against Australia is 52.12 over 30 Tests, Tendulkar’s is 53.11 over 18 Test matches.

McGrath v Lara should read 47.16 (52.12)
McGrath v Tendulkar should read 36.80 (53.11)

It is clear that McGrath has had more success against Tendulkar than he has had against Lara. It is also clear that his difference is not as much as 22.18 vs 41.40 suggests. What it also reveals is the disparity between the number of Test innings that Tendulkar and Lara have played against McGrath (16 for Tendulkar vs 44 for Lara).

Cricinfo needs to revise this statistic in their statsguru in my view….

CricketingView