New feature: Highest Impact - the all-time most impactful Test cricketers

[chasingthedon]

Yes, expanding the window that way makes sense.

This doesn't resolve the #2 and #1 issues however. How do you handle "time/overs left"? If you don't, your probabilities won't make sense for especially the 3rd/4th innings where the win probability depends a lot on time left.

Because there's not enough info for early games to accurately gauge the time left, all I could do was use stumps scores on the scorecards to calculate which day it was.

 

[chasingthedon]

Another thing I noticed is that you were giving 50% of the dismissal credit to the fielder. I think that's too much - going by the average drop rate % which is around 15%, the fielder should get around that much for a regulation catch with 85% going to the bowler.

I think that would be a little more objective than just giving a flat 50%.

Great job btw - I'm trying to figure out what you've done to get a better underatanding of whether it is actually a viable method.

Again I didn't think we had enough data going back to change it from 50/50 - plus as I mentioned if you give only 15% to the fielder there will be no WKs featured anywhere near the top, not even Gilchrist.

 

[chasingthedon]

Never mind - I thought of starting from the first article, didn't realize you switched methods midway.

I didn't switch midway, I published all of the series points data starting in 2009, then later decided it was not accurate enough insofar as points were allocated on a series basis, not even per match.

So I started on match impact about three years ago.

In answer to your first question, for Series Points what I did to get the batting/bowling split was first to work out the average score and wickets given against opponents of various strength levels (I'd previously gone back and worked out the ICC team ratings going back to 1877). That was used on a series basis to find the initial batting/bowling split.

Then I used those initial numbers to find out the overall split for each era, then went back and offset the original series scores based on the era split values.

 

[chasingthedon]

Yes, expanding the window that way makes sense.

This doesn't resolve the #2 and #1 issues however. How do you handle "time/overs left"? If you don't, your probabilities won't make sense for especially the 3rd/4th innings where the win probability depends a lot on time left.

I forgot to mention that I take the possibility of a draw into account too, which doesn't really have much impact early on, but is a factor when the batting team is trying to see out time and batsmen should get credit for that.

 

[viriya]

Because there's not enough info for early games to accurately gauge the time left, all I could do was use stumps scores on the scorecards to calculate which day it was.

Understandable.. Does that mean you adjust the win probability based on your best assessment of days left (how would you handle my #2 hypothetical case)?

 

[viriya]

Again I didn't think we had enough data going back to change it from 50/50 - plus as I mentioned if you give only 15% to the fielder there will be no WKs featured anywhere near the top, not even Gilchrist.

It was just a suggestion since if the average drop rate % is 12.5-15% (which I've found to be a relatively accurate value from my fielding work), it doesn't make sense to give 50% to a fieilder for doing this job.

A more objective way of looking at your results is that you are probably overrating wicketkeepers and underrating bowler imo.

 

[chasingthedon]

It was just a suggestion since if the average drop rate % is 12.5-15% (which I've found to be a relatively accurate value from my fielding work), it doesn't make sense to give 50% to a fieilder for doing this job.

A more objective way of looking at your results is that you are probably overrating wicketkeepers and underrating bowler imo.

I don't understand why you feel fielders should be rewarded based on their failure rate?

 

[chasingthedon]

Understandable.. Does that mean you adjust the win probability based on your best assessment of days left (how would you handle my #2 hypothetical case)?

As I said, all I can do is work based on the day, as ball by ball isn't available for the first 125 years of Test cricket.

 

[viriya]

I don't understand why you feel fielders should be rewarded based on their failure rate?

The reasoning is that if fielders take regulation catches 100% of the time they should be awarded 0% of the credit for taking a regulation catch because it's expected (kind of like a wicketkeeper catching a ball when a batsman misses it). But since they drop ~15% (meaning they take the catch ~85% of the time), it makes sense to give them that extra ~15% they actualize by taking the catch (~85% -> 100%), with the rest going to the bowler.

imo that makes a lot more sense than just crediting them with 50% of the value - why would 50% be a better way to split the credit? 50-50 isn't automatically objective, it's also making an assumption, and I think using drop rate % is a better assumption because it's backed by numbers.

It's your call though, I think it would improve the ratings.

 

[chasingthedon]

The reasoning is that if fielders take regulation catches 100% of the time they should be awarded 0% of the credit for taking a regulation catch because it's expected (kind of like a wicketkeeper catching a ball when a batsman misses it). But since they drop ~15% (meaning they take the catch ~85% of the time), it makes sense to give them that extra ~15% they actualize by taking the catch (~85% -> 100%), with the rest going to the bowler.

imo that makes a lot more sense than just crediting them with 50% of the value - why would 50% be a better way to split the credit? 50-50 isn't automatically objective, it's also making an assumption, and I think using drop rate % is a better assumption because it's backed by numbers.

It's your call though, I think it would improve the ratings.

haha, you don't reckon much to fielding do you. I did actually re-run the numbers once with bowlers getting all the credit for a parallel project, but that is on bowlers only.

I'd be persuaded by an argument for less than 50/50, but not that fielding basically amounts to nothing.

EDIT: or 15%, come to that.

 

[viriya]

I think it would make sense to go up to 20-25% since you are not distinguishing between regulation and great catches. Anything more would be overrating fielders imo.

 

[chasingthedon]

Also what I haven't done is post the relative levels of batting, fielding and bowling for each player, partly because it goes against what I was trying to do. It might be interesting though, particularly with all-rounders such as Shaun Pollock, to see what the proportions were. Or for a WK like Knott versus Jack Blackham and Adam Gilchrist.

Ran some impact numbers resolving the disciplines:-

Knott - BT: 47.5% FLD: 52.5%
Gilchrist - BT: 53.8% FLD: 46.2%
Flower - BT: 61.1% FLD: 38.9%
Duckworth - BT: 12.6% FLD: 87.4%

So Andy Flower made about 61% of his impact batting, about 39% behind the stumps, while George Duckworth is about 13%/87% respectively.

Is that fair?

Looking at all-rounders:-

Sobers - BT: 55.4% FLD: 10.0% BW: 34.6%
Miller - BT: 44.3% FLD: 8.5% BW" 47.2%
Hadlee - BT: 27.1% FLD: 3.9% BW: 69.0%
Botham - BT: 32.8% FLD: 12.1% BW: 55.1%
Kapil Dev - BT: 26.5% FLD: 5.8% BW: 67.7%
Imran - BT: 29.0% FLD: 2.2% BW: 68.8%
Faulkner - BT: 46.3% FLD: 7.3% BW: 46.4%

Looking at players renowned for their fielding:-

Constantine - BT: 23.4% FLD: 20.8% BW: 55.9%
JN Rhodes - BT: 78.5% FLD: 21.5%

Essentially about twice as much fielding value as e.g. Sobers.

 

[watson]

Ran some impact numbers resolving the disciplines:-

Knott - BT: 47.5% FLD: 52.5%
Gilchrist - BT: 53.8% FLD: 46.2%
Flower - BT: 61.1% FLD: 38.9%
Duckworth - BT: 12.6% FLD: 87.4%

So Andy Flower made about 61% of his impact batting, about 39% behind the stumps, while George Duckworth is about 13%/87% respectively.

Is that fair?

Looking at all-rounders:-

Sobers - BT: 55.4% FLD: 10.0% BW: 34.6%
Miller - BT: 44.3% FLD: 8.5% BW" 47.2%
Hadlee - BT: 27.1% FLD: 3.9% BW: 69.0%
Botham - BT: 32.8% FLD: 12.1% BW: 55.1%
Kapil Dev - BT: 26.5% FLD: 5.8% BW: 67.7%
Imran - BT: 29.0% FLD: 2.2% BW: 68.8%
Faulkner - BT: 46.3% FLD: 7.3% BW: 46.4%

Looking at players renowned for their fielding:-

Constantine - BT: 23.4% FLD: 20.8% BW: 55.9%
JN Rhodes - BT: 78.5% FLD: 21.5%

Essentially about twice as much fielding value as e.g. Sobers.

I guess those results mean that Knott and Gilchrist were great at both batting and keeping, while Flower was a lot better at batting than he was keeping. Duckworth was just a crap batsman.

Out of those allrounders listed Keith Miller and Faulkner seem to be the most balanced with not difference much between their batting and bowling. Would have thought that Botham's batting might have been a bit higher due to his number of innings at No.6.

For non-slips fieldsman those stats for Constantine and Rhodes are pretty awesome.

 

[chasingthedon]

I guess those results mean that Knott and Gilchrist were great at both batting and keeping, while Flower was a lot better at batting than he was keeping. Duckworth was just a crap batsman.

Out of those allrounders listed Keith Miller and Faulkner seem to be the most balanced with not difference much between their batting and bowling. Would have thought that Botham's batting might have been a bit higher due to his number of innings at No.6.

For non-slips fieldsman those stats for Constantine and Rhodes are pretty awesome.

Touched on the balance issue below - Giffen added to Faulkner and Miller:-

The Most All-Round All-Rounder | Cricket Web

Re Constantine and Rhodes, that's one of the benefits of covering all disciplines with one measure.