Fielding Statistics: A New Approach

[viriya]

Yeah unfortunately which makes it a bit difficult for you. Interested in how this would be rated by the way:

3.6
Shapoor Zadran to Anamul Haque, 1 run, fuller now, and edgy cricket all around. Bijoy hits it to cover where despite a dive from Sami Shenwari there should have been a single. Tamim doesn't want it, though. And the throw is wide but the first slip has not moved behind the keeper to back the throw up. And they get the single

No keywords like "diving stop" or "saved a couple" etc.. This event would be ignored.

 

[marc71178]

I already do this. As long as I can identify the fielder (which is not always possible).

If you can't identify every event for every game then there's no way you can count that game at all.

 

[viriya]

If you can't identify every event for every game then there's no way you can count that game at all.

If you can't get DRS 100% perfect you shouldn't use it at all.

 

[cnerd123]

The reason the batsman's rating is involved is because catches/drops that impact the match are of the opposition's best batsmen.. Even then, great catches are scaled to be 20 times as important.

What would be rated higher than a blinder off Murali is a blinder off Sanga since that catch actually matters in the match while Murali will probably not last another over.

There is an argument for less weight on the batsman's rating, but at least check what I'm doing before going off people.

No that literally makes no sense.

A fielder isn't more likely to drop a catch because it comes off the bat of Sanga, and isn't more likely to catch it if it comes off the bat of Murali.

I don't see why batsman's rating needs to figure in the calculation at all. A player's fielding ability is his ability to stop, catch, chase after and throw the ball. None of these abilities are affected by the quality of the batsman who hits the ball.

I can see the argument for introducing a 'Pressure' variable to take into the account the match situation of when the catch/piece of fielding took place, to see if fielders are better or worse under pressure scenarios...but literally no reason why batsman's rating needs to be considered.

 

[tosser]

considering the amount of effort required to be put into this both financially and time-wise, and considering the extent of objectivity that can ever be reached, pursuing advanced fielding statistics is a fruitless endeavour

 

[tosser]

considering the amount of effort required to be put into this both financially and time-wise, and considering the extent of objectivity that can ever be reached, pursuing advanced fielding statistics is a fruitless endeavour

i'll cap this off with an IMO

 

[Burgey]

Brad McNamara was a good fielder.

 

[viriya]

No that literally makes no sense.

A fielder isn't more likely to drop a catch because it comes off the bat of Sanga, and isn't more likely to catch it if it comes off the bat of Murali.

I don't see why batsman's rating needs to figure in the calculation at all. A player's fielding ability is his ability to stop, catch, chase after and throw the ball. None of these abilities are affected by the quality of the batsman who hits the ball.

I can see the argument for introducing a 'Pressure' variable to take into the account the match situation of when the catch/piece of fielding took place, to see if fielders are better or worse under pressure scenarios...but literally no reason why batsman's rating needs to be considered.

The batsman's rating matters because getting Sanga out with a blinder makes a huge difference in the game compared to getting Murali with a blinder. It's not about catching skill, it's about the match result. Kinda like edging a boundary still counts as 4.. or how a batsman that has no technique can still slog and get a 6. It's about the result at the end of the day.

 

[viriya]

Also I should add even if a fielder gets blinders vs tail-enders exclusively, if he doesn't drop much and gets great catches, he will be rated highly. The quality of the batsman starts mattering for fielders with similar records as differentiation.

 

[viriya]

The Panesar > Symonds one is a good call. Notice that Monty's drop rate is 40% compared to Symonds' 6.25% so it's not obvious why he's rated higher.

Checked this and found that Panesar's high rating was based on a misattributed direct hit. Fixed the bug and reran everything, some changes in the list (Panesar disappears from the top career fielders list for one). The issue was I was not parsing fielder names when "'s" was in the name (Anderson's).

This is still a work-in-progress as this shows, good thing is it just gets better and issues like this should decrease with time.

 

[marc71178]

If you can't get DRS 100% perfect you shouldn't use it at all.

Relevance?

 

[NUFAN]

Viriya, it seems like the commentator while New Zealand was fielding yesterday was trying to help you out as much as possible.

 

[viriya]

Viriya, it seems like the commentator while New Zealand was fielding yesterday was trying to help you out as much as possible.

Yea I noticed lol. But he didn't really because a lot of the keywords were missing. I can't really look for "a flying Baz". I caught the Milne catch though.

 

[NUFAN]

Yea I noticed lol. But he didn't really because a lot of the keywords were missing. I can't really look for "a flying Baz". I caught the Milne catch though.

The innings didn't last that long, you could just manually look it up!

Can you tell us all your keywords?

 

[viriya]

The innings didn't last that long, you could just manually look it up!

Can you tell us all your keywords?

I could, but then it defeats the purpose of the whole automated setup. It would be unfair to single out this match when all previous matches were not looked into the same way.

So many keywords - most are related to dropped catches, and also whether the drop was a tough one or not. Some for ground fielding:
fumble
misfield
poor fielding
poor effort
bad fielding
makes a mess of it
not great fielding
fielded!
great fielding
saved a boundary
saved four runs
saves four runs
saves three runs
saved three runs
brilliant work
diving stop
saved a boundary
saves three runs
saved three runs
saved four runs
saves four runs
saved two runs
saves two runs
saves a couple
saves a couple
saved a single
saves a single
saves a run
saved a run

I usually add ones as I see them if they are generic enough to be used in any match.

 

[NUFAN]

Ok cool, you seem to have most things covered.

I suggest 'overthrow' and 'weak throw' or variants of them too.

 

[Athlai]

saves some

is probably relevant too.

 

[Daemon]

You have saves a couple twice

'Good bit of fielding' maybe?

 

[viriya]

Ok cool, you seem to have most things covered.

I suggest 'overthrow' and 'weak throw' or variants of them too.

saves some

is probably relevant too.

You have saves a couple twice

'Good bit of fielding' maybe?

All good suggestions, will check if useful and incorporate.

 

[viriya]

I redid the way a batsman's current rating is used to rate fielding events. I agree that it shouldn't be the main factor - a great catch is a great catch regardless of who the batsman is, but the impact is greater if it's the opposition's better batsman. So for example for a dropped catch I used to do:
-(current batsman rating)

now I do:
-constant - (current batsman rating)/5

It still is a factor, but much less significant than before. In practical terms, if a fielder were to drop Kane Williamson instead of Kyle Mills before, he would've been assigned -900 instead of -400. Now it would be -416 instead of -316 - a lot smaller difference.

Redid both Test and ODI fielding with this, also added some of the above suggestions and fixed some more bugs.