Fielding Statistics: A New Approach

[NUFAN]

What does Steve Smith score for that catch just then? (Buttler dismissal).

 

[Athlai]

Not that there's any convincing you but I'll play along. I checked all of his drops and they were legitimate. 20% drop rate is poor for a point fielder based on others like AB/Ponting/Guptill/Hussey/Collingwood who are in the 2-10% range.

And I already pointed out that Philander had an average 2014 in Tests compared to Ishant who had one of his best (42 average vs 31) , so while their careers aren't comparable, current form-wise it's not a stretch to consider them close to each other.

A drop isn't a pass or fail mark though. There are levels of difficulty. This is a complete waste of time tbh. You can't measure fielding on match scorecards and text commentary.

 

[viriya]

What does Steve Smith score for that catch just then? (Buttler dismissal).

He'll get a 604 fielding match rating (assuming he doesn't catch anything else or affect a direct hit) - Buttler's current batting rating. He gets the full rating cos it's a great catch - the term "stunner" is in the commentary which will get parsed. If it was a regulation catch he'd get 20% of that number.

 

[viriya]

A drop isn't a pass or fail mark though. There are levels of difficulty. This is a complete waste of time tbh. You can't measure fielding on match scorecards and text commentary.

I look to ignore drops on tough chances. It's not meant to be a perfect science but I look for specific expressions like "tough chance"/"would've been a stunner" to avoid those.

You'd be surprised how much the drops that are parsed are actually legit drops. I can dig deeper with examples (with the corresponding text commentary) if you want. Also, considering the usual suspects that are considered great fielders are generally rated high, I don't think the setup is that off the mark. I'm sure I'm missing some cases and erroneously marking some as drops when they shouldn't be, but from my testing that doesn't seem to be the norm, more the exception.

In general the text commentary will have something that explains the nature of the drop - whether it "should've been taken" or not. It's not that hard to parse that.

 

[NUFAN]

He'll get a 604 fielding match rating (assuming he doesn't catch anything else or affect a direct hit) - Buttler's current batting rating. He gets the full rating cos it's a great catch - the term "stunner" is in the commentary which will get parsed. If it was a regulation catch he'd get 20% of that number.

What about number 5 on your methodology of runs saved and number 4 great catches?

 

[viriya]

What about number 5 on your methodology of runs saved and number 4 great catches?

Right now runs saved only accounts for misfields or great stops - not hypothetical runs saved based on how many runs Buttler was expected to get. That's something I plan to do next (probably as a separate factor because it's more of a guess compared to actual fielding stops). #4 is where he gets the +604.

On the other end of the spectrum, Sangakkara dropped Kane early on in the NZ vs SL match. He'll get a -906 rating for that since KW is rated #3 in the current ODI batting ratings.

 

[Athlai]

I look to ignore drops on tough chances. It's not meant to be a perfect science but I look for specific expressions like "tough chance"/"would've been a stunner" to avoid those.

You'd be surprised how much the drops that are parsed are actually legit drops. I can dig deeper with examples (with the corresponding text commentary) if you want. Also, considering the usual suspects that are considered great fielders are generally rated high, I don't think the setup is that off the mark. I'm sure I'm missing some cases and erroneously marking some as drops when they shouldn't be, but from my testing that doesn't seem to be the norm, more the exception.

In general the text commentary will have something that explains the nature of the drop - whether it "should've been taken" or not. It's not that hard to parse that.

And how do you measure Guptill saving 3 runs? Or the batsman not running because they know McCullum's fast enough to stop this single and threaten the stumps.

How do you measure catches that occur that wouldn't have been possible with 95% of other fielders like a few of Boult's over the summer?

Taking a little bit of informtion to paint an entire picture not covered by statistics or commentary is pointless.

Until cricket measures these things more accurately, it's a futile exercise.

 

[RossTaylorsBox]

It's a terrible idea to associate the catches with the batsmen rating.

 

[Athlai]

You know what was a worse drop than the Sangakkara one? The Jeevan Mendis drop. Where he displayed poor technique to miss an easy catch.

 

[viriya]

And how do you measure Guptill saving 3 runs? Or the batsman not running because they know McCullum's fast enough to stop this single and threaten the stumps.

How do you measure catches that occur that wouldn't have been possible with 95% of other fielders like a few of Boult's over the summer?

Taking a little bit of informtion to paint an entire picture not covered by statistics or commentary is pointless.

Until cricket measures these things more accurately, it's a futile exercise.

So you're saying measuring drop catches are pointless because there are more complicated stats that would be needed for a more accurate picture?

Of course. But that applies to things like batting average too. That ignores the quality of the bowling, the support the batsmen had, whether he had a platform to work with etc etc.. When we look at the batting average we know all these complicated factors are ignored, but we still put value in a player's batting average because it still gives you some idea about how good the player is in doing his job of making runs.

Guptill saving 3 runs would be parsed if it has certain keywords like "great stop" or "saved a boundary" (among other similar words).
If Boult got a catch that's considered a great one, he will get the credit for it if the commentary has keywords like "fantastic catch"/"stunner" etc.
If he had got to a catch that most wouldn't have and drops it, assuming the commentary states that "would've been a stunner" etc, I ignore that from his drops - this one is tricky but in general all dropped catches have details on how regulation it was.

 

[viriya]

You know what was a worse drop than the Sangakkara one? The Jeevan Mendis drop. Where he displayed poor technique to miss an easy catch.

Yes, in terms of technique. But in terms of match significance Sanga's drop is more important. It's like if you're considering runs made by two batsmen. One of them look like he's going to get out anytime but makes 50, the other is in more control but gets out at 35. The 50 is more valuable even though the 35 was more reliable (assuming SR is not a factor).

 

[Athlai]

So you're saying measuring drop catches are pointless because there are more complicated stats that would be needed for a more accurate picture?

Of course. But that applies to things like batting average too. That ignores the quality of the bowling, the support the batsmen had, whether he had a platform to work with etc etc.. When we look at the batting average we know all these complicated factors are ignored, but we still put value in a player's batting average because it still gives you some idea about how good the player is in doing his job of making runs.

Guptill saving 3 runs would be parsed if it has certain keywords like "great stop" or "saved a boundary" (among other similar words).
If Boult got a catch that's considered a great one, he will get the credit for it if the commentary has keywords like "fantastic catch"/"stunner" etc.
If he had got to a catch that most wouldn't have and drops it, assuming the commentary states that "would've been a stunner" etc, I ignore that from his drops - this one is tricky but in general all dropped catches have details on how regulation it was.

Batting is a game that is literally measured by numbers. Yes there are other aspects to batting but runs scored is the chief and solitary goal of a batsman, the more runs a batsman scores, the better he is.

Fielding is a game that is only in part measured by numbers. The amount of catches, boundaries stopped, run outs made are not reflective of a fielder actually does. You can certainly measure a fielder's impact on a game, but you're going to need a hell of a lot more than what you're going off.

 

[RossTaylorsBox]

So you're saying measuring drop catches are pointless because there are more complicated stats that would be needed for a more accurate picture?

Of course. But that applies to things like batting average too. That ignores the quality of the bowling, the support the batsmen had, whether he had a platform to work with etc etc.. When we look at the batting average we know all these complicated factors are ignored, but we still put value in a player's batting average because it still gives you some idea about how good the player is in doing his job of making runs.

Guptill saving 3 runs would be parsed if it has certain keywords like "great stop" or "saved a boundary" (among other similar words).
If Boult got a catch that's considered a great one, he will get the credit for it if the commentary has keywords like "fantastic catch"/"stunner" etc.
If he had got to a catch that most wouldn't have and drops it, assuming the commentary states that "would've been a stunner" etc, I ignore that from his drops - this one is tricky but in general all dropped catches have details on how regulation it was.

No, it's pointless to try and accommodate a dozen variables because there are some that will never show up through scorecards alone. You're being way too ambitious in thinking you can claim a "Best Fielder" with massively incomplete and noisy information.

 

[viriya]

It's a terrible idea to associate the catches with the batsmen rating.

I was thinking about this and decided that matters. In terms of significance all drops aren't created equal. When the fielder gets under a catch, he knows if it's a significant one or not already so there's pressure involved. So it makes sense to give him more credit for a great catch to get an in-form batsman out, or give negative points for a drop. It's an indirect measure of the runs saved by the fielder by getting the catch. Right now AB has a 1200+ batting rating which is in a way equivalent to being expected to make something like 75 runs at a 90 SR. Getting him out by a great catch or dropping him is a significant event that should have a massive fielding rating (positive or negative). It's the equivalent of saving 75 runs.

 

[Athlai]

Yes, in terms of technique. But in terms of match significance Sanga's drop is more important. It's like if you're considering runs made by two batsmen. One of them look like he's going to get out anytime but makes 50, the other is in more control but gets out at 35. The 50 is more valuable even though the 35 was more reliable (assuming SR is not a factor).

Match significance is entirely meaningless to the quality of a fielder. A bad fielder is as likely to drop Bradman as he is Chris Martin. A good fielder can always drop a difficult chance. Going by arbitrary views of a biased commentator that is either "oh thats a dolly" or "that catch was practically impossible" is pretty ridiculous. There are a lot of grey areas.

You frequently see a cricinfo commentator get it completely wrong missing dropped catches because they didn't think it was close live. Then they can't go back and edit it when they see a reply BECAUSE THEYRE BUSY REPORTING ON THE NEXT BALL.

It's a broken system and entirely unsuitable for any kind of meaningful analysis. Which is why you're numbers are coming up looking like a bingo board.

 

[Athlai]

I was thinking about this and decided that matters. In terms of significance all drops aren't created equal. When the fielder gets under a catch, he knows if it's a significant one or not already so there's pressure involved. So it makes sense to give him more credit for a great catch to get an in-form batsman out, or give negative points for a drop. It's an indirect measure of the runs saved by the fielder by getting the catch. Right now AB has a 1200+ batting rating which is in a way equivalent to being expected to make something like 75 runs at a 90 SR. Getting him out by a great catch or dropping him is a significant event that should have a massive fielding rating (positive or negative). It's the equivalent of saving 75 runs.

Yeah I'm dead certain it occured to Sangakkara how significant the Williamson catch was when he put it down with the .05s he had before it hit him.

 

[RossTaylorsBox]

Yes, in terms of technique. But in terms of match significance Sanga's drop is more important. It's like if you're considering runs made by two batsmen. One of them look like he's going to get out anytime but makes 50, the other is in more control but gets out at 35. The 50 is more valuable even though the 35 was more reliable (assuming SR is not a factor).

But being the "best fielder" is all about technique. The best fielder takes simple chances regardless of the batsmen or match situation.

 

[viriya]

No, it's pointless to try and accommodate a dozen variables because there are some that will never show up through scorecards alone. You're being way too ambitious in thinking you can claim a "Best Fielder" with massively incomplete and noisy information.

How is it massively incomplete when I know all the catches he took, whether those drops were legitimate (based on the commentary stating whether it was expected to be taken) and whether any catches he took were great? It doesn't have all the information but as a first foray to say it has no value just makes no sense. Your argument is that fielding is more complicated than I'm modeling it to be, which is obviously the case, but that argument applies to simple batting stats like average. I'm attempting to create those simple stats first. You gotta walk before you run.

If what you're claiming was accepted, there would be no way to even rate a batsman - because "there are variables that will never show up in scorecards".

 

[RossTaylorsBox]

I was thinking about this and decided that matters. In terms of significance all drops aren't created equal. When the fielder gets under a catch, he knows if it's a significant one or not already so there's pressure involved. So it makes sense to give him more credit for a great catch to get an in-form batsman out, or give negative points for a drop. It's an indirect measure of the runs saved by the fielder by getting the catch. Right now AB has a 1200+ batting rating which is in a way equivalent to being expected to make something like 75 runs at a 90 SR. Getting him out by a great catch or dropping him is a significant event that should have a massive fielding rating (positive or negative). It's the equivalent of saving 75 runs.

Nah, pretty sure fielders don't cognitively think about the match situation when lining up a catch, let alone the ones that are cut straight to gully or whatever.

 

[viriya]

But being the "best fielder" is all about technique. The best fielder takes simple chances regardless of the batsmen or match situation.

The best batsman is not the one with the best technique. It's the guy who makes the most runs (in simplistic terms). Bradman didn't have the "best" technique - he was just so much better than everyone else in getting the job done.

In fielding terms, the "best" fielders would have a low drop rate and make great catches while affecting direct hits and saving runs. Having a good "technique" is irrelevant - results are what matters.