Thursday, May 31, 2012

Stats across eras 5 : A detailed look at decade by decade scoring

The eight part Statistics series by the author was published in Cricketcountry in April to May 2012 

In the first part of this series we had gone through the overall average of all Test cricketers decade by decade from 1877 to 2012.

From the figures, it seemed that after the first few decades of dodgy wickets, drastically different conditions and the domination of bowlers, from the 1920s run making became more or less. It was argued that although the popular perception is that with the gradual use of covered wickets, protective gear, shorter boundaries and the decrease in number of fearsome pace bowlers, run making has become easier down the decades, in reality the degree of difficulty has remained constant ever since 1920.

Since only the overall average of all players was considered decade by decade in our analysis, the article did give rise to a number of justified questions.

How can we say that this overall consistency is not a result of balancing out of extreme outliers? What if some decades witnessed increased easy scoring by some sides while it was neutralised by ordinary performances by the weaker teams? Does it not seem too simplistic to look just at the overall average?

Very valid questions, indeed.

In our defence, we have looked at the average of all the Test cricketers who played the game. It seems more reasonable to assume that the degree of difficulty of scoring runs has remained constant, new technical adaptations have developed dealing with the new challenges thrown up by time, with the overall talent pool of batsmen and bowlers over the years remaining more or less consistent. Additionally, it is somewhat difficult to believe that with every decade, each time things got easier, some teams started performing with measured ineptness to keep the overall average unvarying.

At the same time, the doubts continue linger. The cognitive biases of Recency Effect and Rosy Retrospection are always at war, and the debate rages on – as thousands of words are exchanged for and against the present and past generations. The good news is that the results are scientifically verifiable.

At this juncture, let us take the help of cricket-agnostic Chinese wisdom and try to equate ten thousand words to a picture.

Produced above is the graph of the average score of each international cricket side in every decade from the 1920s to 1990s. The different teams are distinguished by colour codes and their movement from decade to decade denoted by broken lines. In the middle of the graph is a solid black line representing the corresponding average score of all Test cricket.

To visually gauge whether there has been significant deviation, we draw a carpet across the picture covering the middle range 25 – 35. We have seen that the average scores in the decades hover around 30, and hence create this rule of thumb to consider an average of over 35 or below 25 as outlier. (The actual global averages across decades are 32.41, 30.51, 33.00, 27.66, 30.91, 30.32, 30.69, 29.21)

To translate this into an intuitive cricket match, a typical completed team innings across eras would total around 300. We raise the outlier flag if the team score is less than 250 or greater than 350. It does seem to make a lot of sense.

Looking at the chart, we find 53 data points across the decades based on the teams that have been active at the time. (Australia, England, West Indies and New Zealand in all eight decades, India in seven, Pakistan and South Africa in six and Sri Lanka in two)

  • Of these 10 turn out to be outliers. That is 81% are within the carpet of normalcy.

  • Four of the outliers are above the chart (exceptionally high average) and six are below.

  • Of the ten, eight (four high and four low) outliers occur during the first three decades (1920s to 1940s).

  • The four high outliers appear in the 1920s, 1930s and 1940s. Twice they are the Bradman enhanced Australians, once the Len Hutton driven 903 for seven amassing Englishmen on marl wickets and once the three Ws powered West Indians. From 1950s to 1990s, none of the teams have crossed the upper level of the carpet.

  • Of the four low outliers, India and West Indies, in their early days, perform poorly in 1930s. New Zealand manages it three times (1930s, 1950s, 1960s) and Sri Lanka once when they appear on the scene in 1980s. The low scores are caused more by the lack of experience or quality of the team rather than an uniform change of bowling standards, conditions or wickets.

  • All the teams go through fluctuations, but remain in one concentrated zone from 1950s to 1990s.
The scientific verification about whether the decades produce different figures or not is done through a statistical test called Kruskal-Wallis. The result (a p-value of 0.667) loosely translates into only 32.3% confidence that the figures from different eras are different, whereas for conclusive evidence, the norm is 95% certainty.

Incidentally even the 2010 data, although there is increase in batting averages, also do not show statistically significant deviation.

And finally, apart from the argumentative points – the chart is an excellent indicator of the improvement and deterioration of the standard of each side across decades with respect to the global standard.

If the picture is anything to go by, it was perhaps the great Don Bradman who benefitted from easier conditions. However, the chasm between him and his peers is so great that there is little doubt that he would have been head and shoulders above the rest in any era.

Gary Sobers, Viv Richards, Sunil Gavaskar and Sachin Tendulkar had battle against similar quality and conditions. How they performed during their respective eras is documented in the second part of the series.

No comments:

Post a Comment