The lack of any worthwhile fielding statistics has always bothered me.
So for this edition of the IPL I tried to do something about it. Ive spent the last few weeks reviewing every ball of the group stage of the tournament, creating 1,666 entries in a spreadsheet for all the fielding ‘events’.
It has been an absurd amount of fun.
Last night I finally finished and crunched the numbers. The results, in my opinion, are a curious mix of the expected and the surprising. No doubt my method could be refined and improved – I’m sure some people will disagree with it entirely – but I am brazen enough to declare I think it is on the right track. Please let me know what you think!
First, here are the results. (If you are interested in how I came up with these numbers, I’ve included a detailed summary below.)
This first table lists the top 10 fielders in the IPL (excluding wicket keepers). The ‘runs’ figure is the total number of runs across the tournament that each player’s fielding has been deemed to have been worth to his team.
And now, the bottom 10 fielders who have all been of negative value to their teams.
Here are the results for the eight main wicket keepers used in the tournament.
And finally, here are the team totals.
|2||Rising Pune Supergiant||114.0|
|3||Kings XI Punjab||79.0|
|7||Royal Challengers Bangalore||39.0|
|8||Kolkata Knight Riders||2.5|
What do you think? My first impression was that the fielders at the top and the bottom of the list look right, but that the numbers themselves look low.
Is Ben Stokes, at number one, really only worth 28 extra runs for his side, ie 2 runs per innings? You often hear about how such and such a player is worth 10 or 15 runs a game through their fielding so why are the numbers not higher?
I think that a significant factor is that we tend to overestimate how many touches players get in T20 matches. After all, with a maximum of 120 balls, how often will a fielder actually get involved? And even when they do, the majority of such instances will be standard pieces of fielding in which they have no opportunity to excel. Also, I think we overestimate how much a wicket is worth in T20 cricket (see below for more detail on this).
In actual fact I think these numbers are still very significant. Take the wicket keepers – the difference from best to worst is over three runs per game. In a typical tournament a side might lose one game by this margin, so the difference in Uthappa behind the stumps versus Saha could well be considerable.
And if I was assembling a side, I would want to know where a potential player fell on this list. If you have a few close calls, you might as well choose the players who will add a handful of runs to your side in the field rather than those who might actually cost you runs. This is all borne out in the team values with 115 runs separating first from eighth, in other words eight runs per game.
I’d love to know what you think. Below I outline exactly the approach I have taken. Warning: if you’re into this sort of thing you may enjoy this section. If not, you may find it a handy insomnia cure!
1. Chances (catches, stumpings and run outs)
I wanted to assign a runs value to every chance taken or missed. To do this there are two components to consider: how many runs is a wicket worth, and how difficult was the chance.
How many runs is a wicket worth?
I consulted Professor Steven Stern – the current custodian of the Duckworth Lewis Stern (DLS) method – and he helpfully supplied information about how much a wicket impacts the DLS target in an ‘average’ game (where the side batting first scores 150-160).
To my surprise, it was only about five runs.
Obviously in many instances a wicket is worth much more – for instance, dropping Kohli only to see him score 100 from 60 balls. Had Kohli been dismissed, it’s unlikely the batsmen after him would have scored so many off those 60.
But just as often a wicket is worth less: for example if a side is 120/2 chasing 130 with 30 balls remaining, a wicket or two lost will almost certainly have no impact whatsoever.
So after much thought, I decided to go with DLS and settled on five runs.
Having a standard number, also removes any ‘noise’ caused by factoring in the quality of the batsman dismissed. In the context of a given match, it is more significant if a fielder catches (or drops) a top batsman than a tailender, but in determining the quality of a fielder it is not: if Jadeja performs a brilliant run out why should the skill of the batsman dismissed play any role in assessing Jadeja?s effort?
How difficult was the chance?
To answer this, I ask, “how often would a standard fielder have taken the chance?” Obviously this is subjective, but as things stand that is unavoidable. (Hopefully in future, technology will be able to remove the subjectivity).
So I assigned a percentage value to each chance – to the nearest 10 percent. For example, a straightforward catch might get 80% or 90%, a sharp stumping 50% or 60%, and a direct hit with one stump to aim at 10%. An absolute sitter would get 100% and a miraculous catch would get 0%. I soon felt confident that I was able to apply these judgments quite consistently, game by game, chance by chance.
So if a fielder took a catch that I deemed only 20% of fielders might take, what happened then? Well, by accepting the chance, the fielder had saved their side five runs. A ‘standard’ fielder would have only caught it 20% of the time and so would have saved their side only 20% of five runs, ie one run. Therefore, the fielder in question has performed four runs better than a ‘standard’ fielder and so four runs get added to their tally.
What about if a fielder dropped a catch (or missed a stumping) that I assessed as a 50% chance? Well, they had cost their side five runs, while a normal fielder would have cost their side 50% of five runs ie 2.5 runs. So the fielder in question has cost their side 2.5 runs more than a ‘standard’ fielder would have and so they have 2.5 runs debited from their tally.
2. Ground fielding
Ground fielding was more straightforward. As an example imagine David Warner has stopped the ball from hitting the rope and the batsmen run two. I again consider what a ‘standard’ fielder would have done. If I judge that they would not have stopped it, then Warner has saved his side two runs but if I judge that a typical fielder would indeed have stopped it, Warner gets nothing.
On the flip side, if Warner is unable to stop it he either gets debited two runs if I think a standard fielder would have stopped it or no penalty if I think a standard fielder would not have stopped it.
I then applied this method to each instance of ground fielding, such as keepers stopping byes, overthrows, strength of arm, accuracy of throws, speed in retrieving balls etc.
Time constraints prevented me from watching every ball of the tournament. Instead I used Cricinfo’s ball-by-ball commentary as a starting point.
Most balls nothing of significance happens from a fielding point of view (eg the batsman defends, is beaten, is bowled, hits a massive six, taps a ball to fine leg and jogs a single etc).
But when something of note did happen I usually then navigated to the replay of the ball in question and watched it (often several times) and formed my judgment. Sometimes the Cricinfo description was definitive (eg ‘Smith fumbles and allows a single’) and I didn’t need to watch the footage.
After going through all of the 56 group games I then totalled up each player’s tally as well as a team tally.
As indicated above, this method is subjective and the amount of runs I’ve deemed a wicket to be worth is up for debate.
Furthermore, I am reliant on the quality of the Cricinfo commentary. For example, if David Warner has moved quickly on the boundary, picked up the ball and fired in a bullet throw maybe the batsmen didn’t even consider a second run, yet would have, had they been dealing with an inferior fielder. It’s quite possible that the Cricinfo commentary might not make mention of this – although I must say I think in general it is very informative.
The number of opportunities that a fielder has is also significant. For example, James Faulkner finished with a total of 20 from the 10 ‘opportunities’ he received in the field. This was a higher ratio than everyone above him in the list – so had he played more matches he may have finished on top. Potentially an index combining total with average-per-opportunity might produce a more accurate result – but I’ll leave that for another day!
As I said up the top, I have had a lot of fun compiling all of this information – I hope it provokes some discussion!