Elite All Rounders...

[Nirvanam]

Folks, I just did some queries on statsguru to figure out some real elite all rounders in Test cricket. I also computed the ARIndex = Batting Avg / Bowling Average. An ARI of 1.00 is a very good all-rounder...the higher the number the better the AR. I also computed a Normalized ARI, the normalization is done based on a factor called "Relative Frequency Factor". The RFF is a value between 0.8 to 1.2 and it addresses consistency over long periods. The logic to include RFF is that, an ARI score of 1.00 achieved over 100+ Tests should have slightly more value than an identical ARI achieved over 25 Tests. The range of 0.8 and 1.2 is decided based on certain assumptions (we can change the range) based on the minimum and maximum number of Tests played by the shortlisted players. the minimum number of Tests was 25, and maximum was 131. So if both of these players had an ARI of 1.00, then the Normalized ARI would be 0.8 and 1.2 respectively. The RFF for all other players would be between 0.8 and 1.2 with the player who played 88 Tests (25 + 53 or 131 - 53) having an RFF of 1.00. Finally, Normalized ARI = RFF * ARI Now the conditional queries on statsguru...I qualified the results to give a shortlist of players who attested to the following conditions - a. Batting Average > 30 b. Bowling Average < 30 c. Minimum Runs scored in career >= 1000 d. Minimum Wickets taken in career >= 50 e. Minimum 1 Wkt/Match (to avoid situations where say a player who has taken 60 wickets from 100 Tests makes the cut) The list of elite ARs (sorted based on rank) - Player Normalized ARI Imran Khan (Pak) 1.71 SM Pollock (SA) 1.56 KR Miller (Aus) 1.47 IT Botham (Eng) 1.29 N Kapil Dev (India) 1.26 Mushtaq Mohammad (Pak) 1.23 GA Faulkner (SA) 1.23 TL Goddard (SA) 1.13 CL Cairns (NZ) 1.07 MA Noble (Aus) 1.05 W Rhodes (Eng) 1.04 GE Gomez (WI) 0.90 I am not in the know on how to format a table in a post...otherwise I could've posted the entire table here. Pls help with the steps to follow on how to make a table inside a post and I'll edit it to include the whole table. Alternately, you can view the whole table here - http://tinyurl.com/22rhyyb ps: Does not include keeper-ARs

[Sachinism]

Using tables http://indiancricketfans.com/faq.php?faq=aboutus#faq_bbtables

[hari504504]

Shaun Pollock ahead of Botham and Kapil ???..wow..didn't see that one coming.. and Hadlee isn't even in the top 12? ...seems dodgy to me..

[ZodiaC]

Shaun pollock is actually one of the most underrated cricketers. Still looks messy :headscratch:

[achilles]

Good work, yet again. But yeah, I wouldn't agree that Shaun was better than Kaps or Hadlee.

[Ranjha]

Shaun Pollock used to choke...otherwise he was a good cricketer.

[saneindian]

Interesting, very good analysis. Good bowlers like Imran and Pollock obviously have an advantage here because the denominator is the bowling average. Perhaps that explains why someone like Pollock is ahead of Kapil and Hadlee.

[putrevus]

Shaun Pollock is reverse of FTB batsmen, his great record is due the fact he bowled mostly in SA, he was solid bowler who took advantage of home conditions and also bowling along with Donald, Nitini

[Nirvanam]

Good work' date=' yet again. But yeah, I wouldn't agree that Shaun was better than Kaps or Hadlee.[/quote']yeah, me too!

[Nirvanam]

If we remove the batting and bowling average filters it will throw up guys like Sachin ahead of Wasim coz Sachin would be 56/52 and Wasim would be 22/23. But on the other hand we will have Sobers, Kallis, Hadlee, Flintoff, etc come in. I think the 30 Average cut-offs are a way to see those very special ARs who could make it to the team purely as either a batsman or a bowler. Apart from Pollock and to a certain extent Mushtaq Mohd, all the other elite chaps seem like guys who could've at various stages in their careers made it to their teams' starting XI as a specialist of either skill. But, if we want to recognize the one-dept-heavy ARs, we can make the cut-offs conditional such that, a. If Bat-Avg >= 45 then Bowl Avg can be <= 35 b. If Bowl Avg <= 22.5/23 then Bat Avg can be >= 25 That will bring in Hadlee, and Kallis and both of them will make it ahead of Kapil and Botham, and Kallis will become the topper ahead of the Niazi....which somehow doesn't sound right. But still we can do that.

[Nirvanam]

Shaun Pollock ahead of Botham and Kapil ???..wow..didn't see that one coming.. and Hadlee isn't even in the top 12? ...seems dodgy to me..

These are the top 12 but also consider them more like the elite 12 who attest to the filtering conditions. There are only 12 cricketers in the History of Test cricket who attest to those filters...that's a more appropriate way of saying it I guess.

[thongale]

There is a problem with bat avg / bowl avg ratio. Batting averages tend to change in a larger span (for top 7 batsmen) than for bowling averages. We know that to average 50 with bat is twice (or bit more) difficult than averaging 25. But a bowling average of 25 and 50 shows a much larger gulf than a batting average. My suggestion is to get an exponential curve for batting and bowling points and get the product of it. That will surely put guys like Imran and Miller at the top over Sobers or Kallis

[Nirvanam]

There is a problem with bat avg / bowl avg ratio. Batting averages tend to change in a larger span (for top 7 batsmen) than for bowling averages. We know that to average 50 with bat is twice (or bit more) difficult than averaging 25. But a bowling average of 25 and 50 shows a much larger gulf than a batting average. My suggestion is to get an exponential curve for batting and bowling points and get the product of it. That will surely put guys like Imran and Miller at the top over Sobers or Kallis

I thought on those lines too...something similar (but don't agree with your example of batting average of 25 and 50 comparison to bowling average of 25 and 50. Maybe the comparison would be better this way - Batting improving from 30 to 45, Bowling improving from 30 to 22.5). So here is what I thought we could do to include the one-department-heavy ARs...oh looks like I have already mentioned it in the previous posts. Anyway copy-paste But, if we want to recognize the one-dept-heavy ARs, we can make the cut-offs conditional such that, a. If Bat-Avg >= 45 then Bowl Avg can be <= 35 b. If Bowl Avg <= 22.5/23 then Bat Avg can be >= 25 OK, pls tweak this a lil more to make it more appropriate and lemme know...I'll pull out the data and present them. Edit - I think a factor of 2 should be the difference for improvement in Batting over Bowling...lemme explain... If Bat of 50+ is considered ATG then Bowl of 25- is considered ATG. If 30 is the cut-off point for G-ness then a. on Bat scale moving from 30 to 40 is equivalent to moving from 30 to 25 on Bowl scale b. on Bat scale moving from 40 to 50+ is equivalent to moving from 25 to 22.5/23- on Bowl scale c. on Bat scale moving from 50 to 60+ is equivalent to moving from 22.5 to 20- on Bowl Scale So maybe roughly, a factor of between 2 and 4. However, I think we should take it step-wise (a, b, c) and make it factor of 2 for a, 3 for b, and 4 for c. How does the above sound?

[The Outsider]

You really need to lay off the sh!t you smoke! Just using random, scientific sounding words does not graduate your 'analysis' from hogwash. Maybe if you do a crap all rounder analysis, Sobers and Kallis will figure in there. Along with Agarkar of course.

[Raghav_12]

Your qualifying criteria for allrounder is flawed. Basically by keeping bowling average criteria as maximum of 30, you are trying to prepare a table for best batsmen among the all time great bowlers. Some great bowlers themselves average around 30 Kapil, Lee, Kumble etc. How many bowlers are there as of now having average less than 30? and compare that number with the batsmen who have average more than 30. Having average of 30 as a bowler you can play 100+ test while having an average of 30 you won't last more than 20 tests, ofcourse unless you are Ashraful. There are two options. If you want to stick with bowling average of 30, then increase qualifying batting average to 40 or move 30 to 35 for both batsmen and bowlers. However, I have more scientific approach for that (if you have enough time to spend :winky:). Basically you have to find a common average which is fair for both bowlers and batsmen. At you selected number of 30, you'll have some 7-8% from the total list of all bowlers making the cut while you'll have some 20-25% batsmen from total list of batsmen making the cut. Now % of bowlers making the cut would be increasing fn of the number that you select while % of the batmen would be decreasing fn of that selected number. So when you start increasing this number of 30, % of bowlers who make the cut will increase and % of batsmen who make the cut will decrease. Eventually you'll reach to a number where % of bowlers making the cut would be exactly equal to % of batsmen making the cut. (There has to be such number, because one of them is increasing fn while other is decreasing fn). And this is the magic number which you should keep as the qualification mark for both batsmen and bowlers. However, there is even better way - Just enjoy third test match btw Ind and NZ as it unfolds and forget about history :cantstop::cantstop:

[Nirvanam]

Your qualifying criteria for allrounder is flawed. Basically by keeping bowling average criteria as maximum of 30, you are trying to prepare a table for best batsmen among the all time great bowlers. Some great bowlers themselves average around 30 Kapil, Lee, Kumble etc. How many bowlers are there as of now having average less than 30? and compare that number with the batsmen who have average more than 30. Having average of 30 as a bowler you can play 100+ test while having an average of 30 you won't last more than 20 tests, ofcourse unless you are Ashraful. There are two options. If you want to stick with bowling average of 30, then increase qualifying batting average to 40 or move 30 to 35 for both batsmen and bowlers. However, I have more scientific approach for that (if you have enough time to spend :winky:). Basically you have to find a common average which is fair for both bowlers and batsmen. At you selected number of 30, you'll have some 7-8% from the total list of all bowlers making the cut while you'll have some 20-25% batsmen from total list of batsmen making the cut. Now % of bowlers making the cut would be increasing fn of the number that you select while % of the batmen would be decreasing fn of that selected number. So when you start increasing this number of 30, % of bowlers who make the cut will increase and % of batsmen who make the cut will decrease. Eventually you'll reach to a number where % of bowlers making the cut would be exactly equal to % of batsmen making the cut. (There has to be such number, because one of them is increasing fn while other is decreasing fn). And this is the magic number which you should keep as the qualification mark for both batsmen and bowlers. However, there is even better way - Just enjoy third test match btw Ind and NZ as it unfolds and forget about history :cantstop::cantstop:

LOL LOL LoL! Fact of the matter is that I am berozgar these days so instead of sitting in front of the Idiot Box I choose to st in from of the Trash Board. But you know what, I've been thinking about the AR thing considerably since the evening and after reading your post I am getting a lot of ideas...I think I'll attempt a Virtuoso Analysis for ARs

[thongale]

I thought on those lines too...something similar (but don't agree with your example of batting average of 25 and 50 comparison to bowling average of 25 and 50. Maybe the comparison would be better this way - Batting improving from 30 to 45, Bowling improving from 30 to 22.5). So here is what I thought we could do to include the one-department-heavy ARs...oh looks like I have already mentioned it in the previous posts. Anyway copy-paste But, if we want to recognize the one-dept-heavy ARs, we can make the cut-offs conditional such that, a. If Bat-Avg >= 45 then Bowl Avg can be <= 35 b. If Bowl Avg <= 22.5/23 then Bat Avg can be >= 25 OK, pls tweak this a lil more to make it more appropriate and lemme know...I'll pull out the data and present them. Edit - I think a factor of 2 should be the difference for improvement in Batting over Bowling...lemme explain... If Bat of 50+ is considered ATG then Bowl of 25- is considered ATG. If 30 is the cut-off point for G-ness then a. on Bat scale moving from 30 to 40 is equivalent to moving from 30 to 25 on Bowl scale b. on Bat scale moving from 40 to 50+ is equivalent to moving from 25 to 22.5/23- on Bowl scale c. on Bat scale moving from 50 to 60+ is equivalent to moving from 22.5 to 20- on Bowl Scale So maybe roughly, a factor of between 2 and 4. However, I think we should take it step-wise (a, b, c) and make it factor of 2 for a, 3 for b, and 4 for c. How does the above sound?

Great. The same thing can be done by introducing a exponential equation like y = a * exp(x). (Exp - exponential, or anti-natural log). That will avoid your grouping bias (the top candidate of the group and the bottom one will be treated similarly when groupings are done) and make it uniform. It's like improving the number of groups to a very high amount.