Statistical challenge: Estimating a batsman's equivalent of a bowler's performance and vice versa.

[patriot]

If a Shane Warne has 708 test wickets, he would have achieved a batsman's equivalent of how many test runs ? 12,000 ? 15,000 ..Or how much ? For people who argue that such an estimation makes no sense because a bowler can bowl as much as 50 % of the overs in a test - my argument is a batsman can bat even more than that because he can bat at both ends and is usually not under pressure to score at any rate unlike ODIs, therefore I feel a case can be made of this. I honestly think such an estimation is very feasible. But what would be the right multiplier ? Most probably somewhere between 20-25. One estimation method could be multiplying the average og the top 10- 15 bowlers average during a bowler's playing career with his wicket tally. In Shane Warne's career span if that is say 23 is the avg of the bowling avgs of the top 15 bowlers, it would be 23* 708 = 16,284 or an achievement equivalent to scoring 16,284 runs. What got me curious on this, is trying to gauge the utility of great all rounders, ( Kapil, Botham etc viz a viz purely great batsmen, through a single metric like runs) who have won matches with both with bat and ball. Bossbhai - Isme dimaag lagao apna. :--D:--D Would love to here arguments from folks who feel such a conversion/comparison is no way feasible.

[Guest BossBhai]

-- Removed on request of the user --

[yoda]

Interesting. You haven't included the bowler's own bowling average. That needs to factor in somewhere.

[patriot]

Traditionally..statisticians list 5W hall as a celebrated landmark as much as a 100. So a 20 multiplier seems to be a rough equivalent. But for the last decade or so where batting has become so much more favorable with a few more batters likely to breach 10,000 test runs in the next 2-3 years, I think the multiplier should be higher and be based on the avg bowling avg of the top 10 bowlers of an era.

[f.b.m]

30-31 is the average runs/wicket these days. Then again, you will need to subtract the runs conceded to make it a fair comparison. So 5/151 would mean that your team is actually worse off for your efforts and 5/51 would mean that you haven't actually scored a 100. Having said that you should factor in the dismissed batsmens average to sort out the tail end terrorizers. After doing all this you can calculate the batting strike rate based on the bowling strike rate and calculate the overall number of runs that this bowler might have scored. Then you wake up, put on your glasses and find out that you're Ananth Narayanan.

[patriot]

F.B.M - Your attempt is like Ananth Narayanan - makes no sense and offers no explanation but seems to be an effort that has a lot of work behind it.

[patriot]

I tried a little test on what I thought should be the best estimate of a benchmark ( as stated in OP) - and it seems to be a very good estimate indeed. 2000 - Date 1) For the period ( Jan 1, 2000 - Current ) I took the avg of the avgs of the top 15 bowlers ( min 100 wickets ). That avg was 26.89. 2) Now the avg number of wickets by the top 15 bowlers in this period is 265.06 3) Therefore based on 1) and 2) the avg run equivalent of this works out to be 26.89 * 265.06 = 7127.643 4) Now, I take the top 15 run getters for the same period ( Jan 1, 2000 - Current ). 5) The avg runs scored by the top 15 run getters turned out to be 7361.6 An error of 3.28 % . Not perfect but good enough IMO. I will try the same for another decade to see if it works out the same.

[teacup]

Traditionally..statisticians list 5W hall as a celebrated landmark as much as a 100. So a 20 multiplier seems to be a rough equivalent. But for the last decade or so where batting has become so much more favorable with a few more batters likely to breach 10,000 test runs in the next 2-3 years, I think the multiplier should be higher and be based on the avg bowling avg of the top 10 bowlers of an era.

It could be because of a combination of lack of good fast bowling resources and flatter pitches. If that is true, how will you separate the effect of these two factors? Clearly, the idea should be to compensate only for the flatter pitches and not for lack of quality fast bowling, right? Another point to consider, do you really think spinners were also disadvantaged during the last decade? A quick way to verify would be to compare 3rd and 4th inning runs/wicket across various decades.

[patriot]

1990- 2000 1) 1) For the period ( Jan 1, 1990 - Jan 1, 2000 ) I took the avg of the avgs of the top 15 bowlers ( min 100 wickets ). That avg was 24.182 2) Now the avg number of wickets by the top 15 bowlers in this period is 222.867 3) Therefore based on 1) and 2) the avg run equivalent of this works out to be 24.182 * 222.867 = 5389.369 4) Now, I take the top 15 run getters for the same period ( Jan 1, 1990 - Jan 1, 2000). 5) The avg runs scored by the top 15 run getters turned out to be 4928.286 An error of 8.55 %. Higher than previous, but still good enough for something that's not exactly a lunar mission.

[patriot]

1980-1990 1) 1) For the period ( Jan 1, 1980 - Jan 1, 1990 ) I took the avg of the avgs of the top 15 bowlers ( min 100 wickets ). That avg was 25.72 2) Now the avg number of wickets by the top 15 bowlers in this period is 179.42 3) Therefore based on 1) and 2) the avg run equivalent of this works out to be 25.72 * 179.42 = 4616.056 4) Now, I take the top 15 run getters for the same period ( Jan 1, 1980 - Jan 1, 1990). 5) The avg runs scored by the top 15 run getters turned out to be 4575.5333 An error of just 0.877 %. This is working.......

[patriot]

1970-1980 1) 1) For the period ( Jan 1, 1970 - Jan 1, 1980 ) I took the avg of the avgs of the top 15 bowlers ( min 100 wickets ). That avg was 27.16 2) Now the avg number of wickets by the top 15 bowlers in this period is 146.6 3) Therefore based on 1) and 2) the avg run equivalent of this works out to be 27.16 * 146.6 = 3981.753 4) Now, I take the top 15 run getters for the same period ( Jan 1, 1970 - Jan 1, 1980). 5) The avg runs scored by the top 15 run getters turned out to be 3684.666 An error of just 7.46 %. Not bad ..eh ?

[Roshanrocks]

Nice :icflove: carry on bro

[patriot]

So over a huge sample of 4 decades the errors of this experiment have been 3.28 %, 8.55%, 0.877 % and 7.46 % The decade wise errors of the experiment have turned out to have a range of ( 0.877 % - 8.55 % ), with the average error of the experiment about 5%. Conclusion: 5% error IMO means that it's a damn good estimate measure and therefore can be used to estimate a batsman's equivalent of a bowler's performance and more importantly can be used to gauge an all rounder's utility solely by the metric of runs. What this yields is that... 1) Kapils 434 wickets over the period that he has played for are equivalent to runs as follows: 1970 's - 87 wickets or 2362.978 equivalent runs 1980's - 272 wickets or 6997.589 equivalent runs 1990's - 63 wickets or 1523.466 equivalent runs ---------------------------------------------------------------------------------- 434 wickets or 10884.033 equivalent runs Incidently his 434 wickets were a long standing world record like Gavaskar's 10122 runs. 10886.033 and 10122 ? Not too far off eh ? Apna theory solid hai boss, Practically bhi successful hai !B-> Add to that Kapil's 5248 runs off the bat and his net comes out to be ( 10844 + 5248 )16132 equivalent runs. Most will agree that Kapil's net contribution and longevity and utility were about 60 % of Sunny Gavaskar.

[Roshanrocks]

wow.... m guessing Kallis will be Number 1 in this calculation :hmmm: 10843 runs and 261 wickets

[S.Bond]

The day will come, you will read " murli is better batsman than lara" on this forum.... More seriously, interesting stats....

[S.Bond]

Oups, i hope i have not read well and i have not read everything.... But if i understand well, you multiply bowling average with number of wickets and that gives you numbers of runs... The problem i see with this, is the lower the bowling average will be better for the bowler it is... But in the calculation you are doing the higher the bowling averages are the better it is....

[patriot]

Oups, i hope i have not read well and i have not read everything.... But if i understand well, you multiply bowling average with number of wickets and that gives you numbers of runs...

Umm...then read properly.:winky:

The problem i see with this, is the lower the bowling average will be better for the bowler it is... But in the calculation you are doing the higher the bowling averages are the better it is....

The higher the avg Top 15 bowling avg means generally batters have prospered and hence runs are " cheaper " , therefore when you multiply the higher bowling avg with wickets you get a bowlers equivalent runs to the batsman's" cheaper" runs .

[patriot]

Again the conversion in this whole experiment is purely quantitative BUT based on a qualitative metric- the top 15 bowler's avg decade wise. ( with a minimum 100 wickets) And you will see below why I believe we hear the cliche - Bowlers win test matches and therefore contribute more in tests. Anil Kumble - 619 wickets. 1990's - 264 wickets or 6384.048 runs 2000's - 355 wickets or 9545.95 runs -------------------------------------------------------------- Anil Kumble 619 wickets or 15,929.99 equivalent runs. Even adjusting for the 5% historic error - that is atleast 15,000 test runs equivalent. I think that is a very fair retribution to a tireless bowling great who has won many a match for India. And what it means is that when SRT reaches this mark, with the bat , which he surely will, he will quantitatively reach Kumble's utility as a test cricketer for India.

[Holysmoke]

Anil Kumble 619 wickets or 15,929.99 equivalent runs. Even adjusting for the 5% historic error - that is atleast 15,000 test runs equivalent. I think that is a very fair retribution to a tireless bowling great who has won many a match for India. And what it means is that when SRT reaches this mark, with the bat , which he surely will, he will quantitatively reach Kumble's utility as a test cricketer for India .

Bhajji has taken 322 wickets this decade, equivalent to 8658 runs. Dravid has scored 8558 runs this decade. By the same formula, we can safely say that Bhajji's utility is more than, or atleast equal to Dravid's utility as a test cricketer for India this decade. PS - :hysterical:

[patriot]

Bhajji has taken 322 wickets this decade, equivalent to 8658 runs. Dravid has scored 8558 runs this decade. By the same formula, we can safely say that Bhajji's utility is more than, or atleast equal to Dravid's utility as a test cricketer for India this decade.

I said 2 times on the same thread - this is a QUANTITATIVE conversion of total utility of a bowlers wickets in terms of runs . Alec Stewart was the top test run scorer of the 90's but he wasn't better than SRT or Lara or Steve Waugh. That is why when you multiply Bhajji's wickets with a bowling avg - you don't use HIS bowling avg which is on the higher side and would get him higher equivalent runs - you multiply it with the avg of TOP bowling avgs. Also, there is no point belittling Bhajji . You think taking 330 wickets in a decade is a joke ? Bhajji has won us many many matches this decade. If you go back to the numbers,and see how many people have taken more wickets than Bhajji last decade , probably the same number of people have scored more runs than Dravid. And it goes without saying - in tests bowler's utility exceeds batsmen as wickets come at a premium.