An Ordered Response Model of Test Cricket Performance - summary

Executive Summary

This research paper puts an economic model to use in analysing the results of test match cricket. Three aspects of test cricket are considered: firstly predicting the outcomes of test match cricket, secondly identifying what style of cricket each nation played, and thirdly determining who had a legitimate claim to the title of world champion. All data used for this analysis was taken from the `CricInfo' web-site, and incorporated test matches between the nine test playing nations over the period 1994 to 1999.

The application of the model used could be considered successful as it correctly predicted 71 percent of test match outcomes. Separating out this aggregate figure reveals that 81 percent of test match losses were correctly predicted, compared with 75 percent of wins, and 57 percent of the draws.

By analysing the results on a country-by-country basis the model was most efficient in correctly predicted the outcomes for Zimbabwe (84 percent of the time), and Australia (82 percent of the time) India had the lowest level of correct predictions with the model only predicting their result in 55 percent of matches.

Another interesting issue is that Pakistan had the highest tendency to be involved in `unpredictable' matches or matches not correctly predicted by the model. Sri Lanka on the other hand had the highest tendency to be involved in predictable matches.

Four input measures were used in the analysis; batting performance (runs per wicket), bowling performance (opposition runs per wicket), batting strike rate (runs per wicket), and bowling strike rate (opposition runs per wicket). By analysing the test match results with respect to these inputs it is possible to identify five cricketing `styles' that the nine test playing nations could be classified into:

1. Batting and Bowling Performance- Pakistan, West Indies, and Zimbabwe;

2. Batting Performance- England, New Zealand, and India;

3. Bowling Performance- Australia;

4. Bowling Performance/ Batting Strike Rate- South Africa;

5. Bowling Performance/ Bowling Strike Rate- Sri Lanka.

The final aspect the paper looks at is determining who has a legitimate claim to the title of world champions. Two methods are used; firstly using the probability of winning and losing, and secondly using a `heavyweight clash' type analysis. If the team with the highest probability of winning, or the lowest probability of losing over the test period were deemed the champion then South Africa would have the title, followed by Australia second. South Africa has a 49 percent probability of winning and a 9 percent probability of losing compared to Australia's 41 percent and 11 percent respectively.

Using a `heavyweight clash' situation between the top two cricket nations, South Africa and Australia is somewhat different. During the test period the two nations played two series, Australia winning 2-1 in South Africa in 1996/97, and Australia winning 1-0 in Australia in 1997/98. In these six tests Australia's probability of winning was 74 percent, while South Africa only had a 5 percent probability of winning. By using this method the above result is reversed and Australia would be crowned as the champions.

This paper uncovered some interesting results and is an encouraging first paper. Outcomes suggest there is scope for implementing more complete model specifications and improving on the results that this simple model achieved. The paper highlights the benefit of applying detailed economic models in the analysis of cricket results at test, and potentially one-day level.