Fielding Statistics: A New Approach

[viriya]

It has nothing to do with being a good fielder though. Mendis catching a sitter from Anderson shouldn't contribute to him being a good fielder, but dropping it should affect his rating negatively because it's an unexpected event (sort of like information theory I guess). At best you can say that the catch of a good batsmen is influential in the context of the game, but you shouldn't extrapolate this to an individual's general fielding ability.


You can't objectively know the difficulty of the chance without watching the matches. How do you know if a tough chance was made tough only because the fielder was good (or bad) enough to make it one in the first place? What is the average ratio of "tough" to "easy" chances, and is this ratio related to the ranking? One of the problems you have from a statistical perspective is trying to wedge in every variable you have. Simplicity and understandability is far more important.

If Mendis had caught that sitter he would've gotten 20% of the value of a great catch since it's regulation. He dropped it though so he gets -100% instead.

You can't completely objectively know anything in fielding if you wanna go that route. I'm relying on someone else who has enough cricket knowledge to evaluate the catch, and from my experience gets it right most of the time based on my years of following the commentaries and watching the game live at the same time.

 

[cnerd123]

Sorry man. I think it's a fine effort, but giving points for catches based on the quality of batsman is absolutely absurd. The quality of the batsman has 0 connection to the difficulty of the catch.

 

[viriya]

Sorry man. I think it's a fine effort, but giving points for catches based on the quality of batsman is absolutely absurd. The quality of the batsman has 0 connection to the difficulty of the catch.

Yes it has no bearing on the difficulty, but it does on the importance and impact of the catch, which is why it's rated higher. In general, the rating of the batsman is not as important as the fact that a fielder made a great catch or dropped one - it just weights it higher. No one remembers a drop of a tail-ender that will get out soon anyway, but we remember the significant drops that cost the team the match.

To be more clear, if a fielder consistently takes great catches and almost never drops, the quality of the batsmen involved won't be significant when considering his overall fielding career - only specific performances.

I realize that rating catches this way is questionable, so I'd suggest those who disagree to focus on more vanilla stats like drop rate % or matches per direct hit stats - or you're welcome to suggest other possible ways to rate fielders.

 

[viriya]

Hey everyone, Matthew Wade is the best ODI fielder for Australia.

Dilshan is apparently one of the worst fielders in the world.

A few bug fixes (issue with handling fielders with multiple events in a game resolved) and changes (more keywords for great catches) and updates through after the first 2 WC ODIs.

Dilshan benefits in ODIs and Tests. In ODIs some great catches are assigned to him correctly (he drops out of the worst fielders list - his drop rate % is still correct though) and in Tests some direct hits are assigned correctly that makes him jump into the Best Test fielders career list. Also Clarke is the top rated current Australian ODI fielder instead of Wade.

Sanga loses his #1 current ODI fielder spot because of the Kane drop.

 

[RossTaylorsBox]

If Mendis had caught that sitter he would've gotten 20% of the value of a great catch since it's regulation. He dropped it though so he gets -100% instead.

You can't completely objectively know anything in fielding if you wanna go that route. I'm relying on someone else who has enough cricket knowledge to evaluate the catch, and from my experience gets it right most of the time based on my years of following the commentaries and watching the game live at the same time.

OK, so why 20%? Why not 10% or 50%? I'm pretty sure changing this value is going to have a huge impact on the ratings.

 

[viriya]

OK, so why 20%? Why not 10% or 50%? I'm pretty sure changing this value is going to have a huge impact on the ratings.

Picked 20% because generally a fielder's max catches per game is around 5 (maybe 6).. so equalizing the performance of 5 regulation catches with 1 great catch. It's arbitrary yes, 50% is definitely too big, there might be an argument for 10%, but I think it should be 10-25%.

I haven't tested with different rates but I expect it not to matter as much you think (unless I go from 20% to 50% which doesn't make sense).

 

[Athlai]

 

[viriya]

Picked 20% because generally a fielder's max catches per game is around 5 (maybe 6).. so equalizing the performance of 5 regulation catches with 1 great catch. It's arbitrary yes, 50% is definitely too big, there might be an argument for 10%, but I think it should be 10-25%.

I haven't tested with different rates but I expect it not to matter as much you think (unless I go from 20% to 50% which doesn't make sense).

This positively affects wicketkeepers - so it makes a lot of sense to separate them from ground fielders which I will do in the next update.

 

[viriya]

Ratings weightings are always going to be an educated guess - this applies to batting/bowling and not just fielding.

How much does SR matter in ODIs compared to runs? It's an educated guess at the end of the day. Same applies here.

 

[RossTaylorsBox]

Batting and bowling statistics are fundamentally different because the because the basic unit (a ball) occurs much more frequently and you have more aggregated data that can average out over time. I don't think the same methodology can be applied to fielding. It sounds like you're basing your weighting on the maximum catches per game. Wouldn't using the actual great/normal catch ratio make much more sense?

 

[NUFAN]

Viriya, you know last night in the England vs Australia game, how do you feel about Maxwell's run out of Anderson? Do you really think Maxwell deserves less praise for a direct hit in this situation when it doesn't matter who was out?

 

[Daemon]

Lol, some of you guys are such ****s. Criticise and be direct all you want but is the sneering and all the sarcastic comments really needed?

 

[OverratedSanity]

Yeah I'm with Daemon, I'm not a fan of viriya's methodology here either but you're being right ****s.

Also, you're ridiculing him for using an "arbitrary" 20% modifier? Guess what those kind of made-up modifiers are used in every single ratings system.

 

[RossTaylorsBox]

Nah, I don't think questioning how why he used a certain parameter is ridiculing at all. I even suggested another value.

 

[viriya]

Wouldn't using the actual great/normal catch ratio make much more sense?

That's actually not a bad idea - I'll look into it. My guess is that it would result in a much lower weight for regulation catches. What do you think 1 great catch is worth in terms of the number of regulation catches (assuming the same quality of batsman for each catch)?

 

[RossTaylorsBox]

I'd weight great catches according that ratio. If I'm making a statistical model, I'd want all my variables to have some kind basis rather than being arbitrary. Even though it will lower the weight of regulation catches I actually think this is preferable, Like I said previously, you want to concentrate on unusual events (dropping easy catches and taking difficult catches). Taking a bunch of easy catches shouldn't alter a ranking too much because it's the expectation of an average fielder.

 

[viriya]

Yes I see your point - you didn't answer my question though.. if you had to guess what would it be?

 

[RossTaylorsBox]

If I had to guess - 8 for all positions and all players. My interpretation of tough catches is probably strict though.

 

[RossTaylorsBox]

Ratio for dropped sitters (excluding keepers) would be a bit higher I reckon. About 10 per regulation catch? You'd expect maybe one per innings.

 

[viriya]

That Kohli drop would be ignored because of the "difficult chance" text in the commentary.