Recently @CricVestigate on Twitter posted about their all-rounder index, that seeks to rank Test match all-rounders on the basis of not only their averages, but their contributions to their teams. Batting average minus bowling average has been conventionally used to assess all-rounders, but as CricVestigate say on their
Great work. First time I’m seeing Bayesian statistics in cricket - probably the Cricviz folks use it as well. One way you could make the model more era-adjusted is by using batting and bowling averages in their eras - I believe you will get 30 in each - else you can just take the top 30 bowlers and batters of the era. Kallis and to some extent Sobers played in an extremely batting friendly era. And time to add Jadeja…..
Wonderful analysis and methodology. Imran and Sobers were truly freakish allrounders.
I was wondering if there could be more of an absolute workload analysis for these all rounders in terms of balls faced as a batsman and overs bowled. Would love to see the amount of effort they put into the match and consequently the effort they extracted out of their opponents. For instance the difference between a run a ball 50 ( likely hit in pursuit of quick runs) vs 100 ball 30 would be that the latter requires more focus/energy and extracts more out of the opponent team.
Similar reasoning for bowling. Regarding Imran- going by the timeline of Imran’s tests, his bowling peak shown in your chart coincides with the 1982 /83 series vs India where he destroyed the Indian batting. A cursory glance at those scorecards and the matches preceding did not seem to indicate him bowling 40% of Pak's overs ( majority of those tests were played at spin friendly wickets in Pakistan/Asia) . Or perhaps I misinterpreted this :-)
In any case I do feel that metric- percentage of team overs bowled does not give a full picture of actual bowling workload. For instance low batting totals coming in dominant bowling performances ( like Imran vs India 82/83) mean that the bowler in question would have bowled a bigger share of a low number of overs. Also bowling in unfavorable conditions means the bowler does not get to bowl enough.
Kapi Dev's workload stats could be interesting. I think he probably took on a higher bowling workload but did not stick around to play long enough innings.
Great Work Arjun. Balanced Article. However, you didn't mention that you considered Longevity. Agreed, most of these players had long careers. I think Being able to play 100 and 150 test matches should have positive effect. Also did you penalize an allrounder for not bowling enough overs. Did you consider Strike rate or Economy Rate. Can you apply this analysis for ODIs too.
Great work. First time I’m seeing Bayesian statistics in cricket - probably the Cricviz folks use it as well. One way you could make the model more era-adjusted is by using batting and bowling averages in their eras - I believe you will get 30 in each - else you can just take the top 30 bowlers and batters of the era. Kallis and to some extent Sobers played in an extremely batting friendly era. And time to add Jadeja…..
Wonderful analysis and methodology. Imran and Sobers were truly freakish allrounders.
I was wondering if there could be more of an absolute workload analysis for these all rounders in terms of balls faced as a batsman and overs bowled. Would love to see the amount of effort they put into the match and consequently the effort they extracted out of their opponents. For instance the difference between a run a ball 50 ( likely hit in pursuit of quick runs) vs 100 ball 30 would be that the latter requires more focus/energy and extracts more out of the opponent team.
Similar reasoning for bowling. Regarding Imran- going by the timeline of Imran’s tests, his bowling peak shown in your chart coincides with the 1982 /83 series vs India where he destroyed the Indian batting. A cursory glance at those scorecards and the matches preceding did not seem to indicate him bowling 40% of Pak's overs ( majority of those tests were played at spin friendly wickets in Pakistan/Asia) . Or perhaps I misinterpreted this :-)
In any case I do feel that metric- percentage of team overs bowled does not give a full picture of actual bowling workload. For instance low batting totals coming in dominant bowling performances ( like Imran vs India 82/83) mean that the bowler in question would have bowled a bigger share of a low number of overs. Also bowling in unfavorable conditions means the bowler does not get to bowl enough.
Kapi Dev's workload stats could be interesting. I think he probably took on a higher bowling workload but did not stick around to play long enough innings.
Great blog
Great Work Arjun. Balanced Article. However, you didn't mention that you considered Longevity. Agreed, most of these players had long careers. I think Being able to play 100 and 150 test matches should have positive effect. Also did you penalize an allrounder for not bowling enough overs. Did you consider Strike rate or Economy Rate. Can you apply this analysis for ODIs too.
This is really really great. Thank you! Would you happen to have z-scores for Jadeja?