Friday, July 20, 2012

On The Effect Of Their Team's Bowling Quality On Test Batsmen

A few weeks ago, I published Test Match Ratings for all Teams since 1945. The method described there takes into account the extent of victory by producing a score for each team in each Test based on the number of runs they scored, the number of wickets they took, and the "Strength" of the opposition. In this post, I apply that method to partnership level information which I have been developing over the past few months.

Take the current England v South Africa Test as an example. As it stands now the score reads:
England 385 all out and 0/0, South Africa 86/1 and 0/0.
Suppose that this Test ends (as I expect it will, with England winning). Hypothetically, lets say it ends as follows:
England 385 all out and 200 all out, South  Africa 320 all out and 220 all out.
England win by 45 runs
In this instance, using the method outlined in my Test Ratings post, the basic points tally will be as follows:
Runs/Wicket for Match = 28.125
England points = 1147.5
South Africa points = 1102.5
Since England win, for the purpose of the Ratings, a win bonus is awarded to them while SA's score is adjusted by taking into account SA's "Strength" score.

At this time, at the end of Day 2, the points tally is as follows:
England = 385 + 1*28.125 = 413.125
South Africa = 367.25

England's Position in the match right now is 413.125/367.25 = 1.125
South Africa's Position in the match right now is 367.25/413.125 = 0.889

In this way, a team's position can be calculated at any point in the match (this is possible once the match is completed, since the Runs/Wicket score is only available after the match is complete).

As you will have immediately seen, if both teams are at par then each team's "Position" will be exactly 1. The team which is ahead in the game, will have a "Position" score greater than 1, while the team which is behind, will have a "Position" score less than 1.

I have built a database calculating these position scores at the start of each partnership in each Test innings since 1877. So for every single innings played by every player, I can measure the position of his team at the start of his innings and at the end of his innings. Based on this, I can calculate an average team position when a given batsman started his innings during his Test career (Given by "IN" in the charts below) and an average team position when a given batsman started his innings during his Test career (Given by "OUT" in the charts below).

I also calculate an average Player Score. For each innings, a player's average score is given by:
Batsman's Score/Runs/Wkt in the match.
The average Player Score (P-SCR in the charts below) is the average score (calculated as above) over a player's career.

In the first match innings, opening batsmen tend to have negative "IN" scores. The "P-IN" and "P-OUT" scores give a percentage above (or below) Par (1) that the team's position was when the player began his innings and when a player ended his innings.

The chart immediately below shows the first match innings (i.e. 1st or 2nd innings in the match). Compare for example, two number 3 batsmen - Rahul Dravid and Ricky Ponting. the "IN" and "OUT" scores show the difference that their respective team's opening stands and bowling strengths made. The chart is sorted by 1st innings aggregate. You could also compare two number 4 batsmen - Mark Waugh and Mahela Jayawardene. Waugh's began his average innings when his team was 38% above par, while Jayawardene began his innings on average when his team was 16% above par.

The figures get really interesting for the 2nd match innings (i.e. 3rd or 4th innings of the Test). The players who started their average second innings with their team behind in the game and the players who started their average second innings with their team ahead in the game break down quite clearly. The Australians, the South Africans, and a few of the contemporary Englishmen have tended to start their 2nd innings with their team ahead in the game. All these teams have had strong bowling attacks with bowling "Strengths" below 30.

What a player's team position is at the start of his 2nd innings has little to do with the player's first innings average. This is significant as the player's in this list are the top run scorers in 2nd innings in Tests (the chart is organized in descending order of aggregates). Brian Lara and Sachin Tendulkar have both started their average 2nd innings with their team between 5 and 9% behind in the game despite averaging over 60 in the 1st innings.

A team's position makes a big difference on the difficulty of batting, especially in the 2nd innings of a Test. Teams that are behind in the game cannot afford to concede runs, and hence cannot afford to attack the batsman very much. Often, they're usually quickly put in a position where the most they can do is delay the declaration. Batting in the 2nd innings with your team ahead in the game is therefore a huge advantage.

What this chart also shows is that even the greatest batsmen cannot overcome the disadvantage of being behind in the game as a rule. Despite improving their team's position, batsmen like Lara and Tendulkar are still dismissed on average with their teams 3.3 and 8.9% behind in the game respectively. This fact seriously challenges the prevailing wisdom of the "match winning" batsman.

If I was the owner of a hypothetical cricket franchise that took part in a league which played 5 day matches, if I had 10 million dollars to set aside for Sachin Tendulkar, then I would make sure I have 30-35 million to set aside for Dale Steyn or James Anderson or Zaheer Khan. That is, if I wanted my team to win.

Do you think we value any top bowler that much more than any top batsman in Cricket? The great superstars in Cricket (and not just in India) have traditionally been batsmen (from Denis Compton onwards) and all rounders (Keith Miller and co.), but rarely fast bowlers or spinners.

Yet those are the players who make match winning innings possible. Those are the players who are genuine match winners.

Perhaps we ought to value bowlers more than batsmen. As the chart below shows, the impact of batsmen rarely overcomes to limitations posed by the bowlers in their teams.


  1. your collection is great but first of all SACHIN TENDULKAR is not a match wining player. if sachin got century in any match, it is sure that india lose this match.other players are well and great.

  2. A fascinating, painstaking analysis. I think your main conclusion merits wider debate. I don't think your analysis necessarily means that batsmen aren't match winners. Your tables show career average match position changes. Even the best batsmen (Bradman, as usual, excepted) only score 100s every four-five tests and one match winning innings per series could be decisive. If your data allows it I would be interested in:
    1) which batsmen have the highest proportion of innings that swung the course of a match by >x - 'x' being a figure you estimate to be 'decisive' for a game
    2) moving away from batsmen, which are the tests which were won or drawn against the odds? In other words, does your data support the view that celebrated matches such as Headingley '81 were the great turnarounds?
    Thank you for a really engaging piece of work.