There is a reasonably strong correlation between a Test player's batting average in a particular batting position, and the frequency with which the player is dismissed for a score under 15 runs in a Test innings. In the chart below, I have plotted the batting averages of Test players in distinct batting positions (minimum 20 innings in a batting position) and their tendency to fail (i.e. score less than 15 runs in an innings). I consider positions - Opener, 3, 4, 5, 6 and 7.
A similar correlation is not found when one considers the frequency with which Test batsmen are dismissed for scores between 15 and 50. Some reasons for this are obvious. Some players who tend to get out early a lot, tend not to miss out often once they are set, but still have low averages. Others tend to find it easy to start, but then tend to get out between 15 and 50. David Gower would be a fine example of this type of player.
A specialist batsman getting out between 15 and 50, can often mean the difference between a team scoring 250 or 400 in an innings. At Chepauk last week, Cheteshwar Pujara was dismissed for 44 when he completed missed the line of a short of length ball which kept low. This left India at 3/105. Had Tendulkar, Kohli and Dhoni not followed with substantial innings, we might have looked back at Pujara's lapse in concentration as being the turning point of the game.
In this post, I have considered 2 types of failure among Test players. The first, represented by dismissal under a score of 15, could be considered being dismissed early in one's innings. The second, represented by dismissal between 15 and 50, could be considered being dismissed when one is set. Some batsmen, as the chart shows, tend to do this more than others. While the first, as I've shown with two scatter plots above, is correlated well with the batting average, the second type of batting failure is not. "Ratio" (short for "Failure Ratio") is a statistic which I propose should accompany the batting average to reflect this.
Ratio is given by (% failures between 15-50 / % failures under 15).
The same VVS, when he batted at number 5, converted over 70% of his starts. He also tended to be dismissed early more often, especially at 5. VVS's Failure Ratio at number 3 is 2.4, while at 5 and 6 it is very close to 1.
The overall comparison between Virender Sehwag and Sunil Gavaskar is instructive. While the two were about equally likely to be dismissed early, Gavaskar was about 25% more likely than Sehwag to convert a start.
Generally, the most successful batsmen in any given batting position are those who combine a high average with a low failure ratio. All batsmen are dismissed early a significant portion of the time. The ones who hurt their team the least are the ones who fail the least to convert starts if they survive. I have seen little evidence in my data to show that aggressive batsmen convert starts less often than the more cautious ones. Viv Richards converted 65, 79 and 75% of his starts at 3, 4 and 5 respectively. Brian Lara converted 65 and 75% of his starts at 3 and 4.
I have a hunch that of the two types of failure described, the first kind attracts the attention of the selectors far more than the second kind, especially once a player is established.