Making sense of Cricinfo’s partnership statistics….

Cricinfo’s superb stats writers have published statistics about partnerships in Test and ODI cricket. They provide an array of partnership records for individual batsmen as well as pairs. The three Australians – Ponting, Hayden and Langer, not surprisingly dominate the Test match lists, along side all time greats like Jacks Hobbs and Herbert Sutcliffe. The statistics do not make a distinction between opening partnerships and other partnerships.

I tried to make some sense of the data, mainly because the partnership list by itself while it is impressive, does not provide any usefully complete information. I considered three variables –

• Partnerships/Innings
• % of partnerships over 50 runs
• Batting average

The premise is, that the most valuable batsman to a side is one who is most difficult to dismiss, has the highest percentage of 50 run partnerships and has the best batting average. For example, a batsman might have a higher partnerships/innings ratio than another batsman, but that might simply be a function of the other batsman in his team being weaker. In such an event, if the batting averages of the two batsmen in question are equal, then it can be said that the batsman with the better partnerships/innings ratio is more valuable. The basic table provided by Cricinfo provides a core list of top class batsmen who may be measured relative to each other by measuring these three variables. The three variables are weighted equally while calculating the cumulative rank. A batsman’s weighted rank for each variable is calculated using ((best value – value for batsman)/Range of values)

The ranks for each of the three variables are relative position on a scale from 0 – 1. Therefore, Ponting’s batting average, which is the best is 8.5% better than Rahul Dravid’s batting average, while it is 77.4% better than Graham Gooch’s batting average. These percentage are relative to the 28 batsmen who make up this list. If Cricinfo can publish data for all batsmen who played Test cricket and featured in 50 run partnerships atleast once, and had a finite batting average, it would be possible very easily to build a realistic rank of batsmen.

Rediff’s "Pressure Index" and Match Track…..

Rediff have published a Pressure Index, which attempts to describe the pressure level during a run chase. It is an interesting methodology, as it involves setting the “par” based on “expert opinion”. I did a Match Track of the Rajkot ODI between India and Sri Lanka (the same game that was used for the pressure index). Match Track uses the runs/wicket ratio in order to establish par. It also takes advantage of the fact that an ODI contest is a 600 ball contest from the outset. The result of the match track is shown below:

Match track suggests that India lost this game after the dismissal of Dinesh Karthik. It seems to track the progress of the game faithfully, without the introduction of random “expert” opinion. There was a revival in the Indian run chase after the mid innings slump (where the two worms briefly meet upon Sehwag’ dismissal). If my understanding is correct, the Rediff Pressure index seems to suggest that India were always ahead of the game until the fall of the 7th wicket.

Match Track is also quite easy to read – in that the closeness of a contest can be determined by how close the two worms are, and by how many times the two worms cross each other. The progress of the match is determined in terms of the portion of the 100 overs that have been completed, and in terms of the number of wickets that have fallen. This allows the reader to identify the impact of each individual performance on the contest as a whole. The issue of determining “par’ is further illustrated in the next example:

This is the 4th ODI of that same series, where Yuvraj Singh, Robin Uthappa and Virender Sehwag all scored at better than a run a ball in India’s chase of almost the same score. In this example, if you see the the graphs during the Sri Lanka batting (before 50.00 % progress), then you will find that in the worms show Sri Lanka to be below par, even with their recovery. Thus, the notion of par being established after the game is completed (since this is aimed at a post game analysis any ways) using runs/wicket is illustrated. The best and most complete way of factoring in quality of opposition, playing conditions etc, is the runs/wicket ratio for a match, and that of each side in comparison to it.

I suppose this discussion might seem boring to some readers – but i would appreciate some feedback and i would be happy to clarify any queries about Match Track.

CricketingView