Inside Edge Fielding Data!

We have a new section on both the leaderboards and player pages that now displays Inside Edge fielding data.

Inside Edge scouts watch every play and grade how easy or difficult it is to successfully field on the following scale:

  • Impossible (0%)
  • Remote (1-10%)
  • Unlikely (10-40%)
  • About Even (40-60%)
  • Likely (60-90%)
  • Almost Certain / Certain (90-100%)

We’ve aggregated all the data and made it available for you to peruse on a player-by-player basis:

ie_fielding_trout

In addition, this data will be updated nightly during the 2014 season!



Print This Post



David Appelman is the creator of FanGraphs.


Sort by:   newest | oldest | most voted
Eric Garcia McKinley
Member
2 years 2 months ago

Fantastic.

Aaron
Guest
Aaron
2 years 2 months ago

This is cool stuff. I do wonder how biased some of these results are. For example, is an ‘unlikely’ play the same for Miguel Cabrera equal to an ‘unlikely’ play for Andrelton Simmons? I’d assume in theory the answer would be yes, but the subconscious mind can be a very powerful vehicle…

Chris from Bothell
Guest
Chris from Bothell
2 years 2 months ago

Perhaps an angle to answering that is to ask, what is the best way to correlate this data to the crowdsourced defensive data? I.e. can you cross-check the Inside Edge scouts to see if their impressions tend to be much different than the unwashed masses?

Roto Wizard
Member
Roto Wizard
2 years 2 months ago

This is going to fuel a whole bunch of new Andrelton memes.

scotman144
Member
Member
scotman144
2 years 2 months ago

I wonder if the definition of “impossible” is a ball Andrelton could not get to….

Jeff Long
Member
2 years 2 months ago

Except that Andrelton ranks 51st on the leaderboard for “Remote” plays at just 6.3%…

<a href="” title=”Link to Leaderboard”>

Jeff Long
Member
2 years 2 months ago

Accidentally entered the HTML tag twice there… whoops

MDL
Member
MDL
2 years 2 months ago

About time this website added an “edit” or a “preview” feature, eh?

Hurtlockertwo
Guest
Hurtlockertwo
2 years 2 months ago

But, But, But this is opinion, not statistics?? I like it!!!

Travis L
Guest
Travis L
2 years 2 months ago

I think it’s a bit of a strawman to assume you can’t aggregate and quantify data that is qualitative or subjective in nature.

Shauncore
Member
Shauncore
2 years 2 months ago

Aren’t “errors” an opinionated statistic at times?

dls
Guest
dls
2 years 2 months ago

Why are the # of 10-40% chances so GREATLY skewed to catchers?

The top 14 are all catchers???

Tim
Guest
Tim
2 years 2 months ago

Trying to throw out runners is historically in the 10-40% range, and very little else in baseball fielding works that way. So I think this may actually be true.

tripvm
Member
Member
tripvm
2 years 2 months ago

It will be interesting to see how this compares with BIS’s +/- system for example.

scotman144
Member
Member
scotman144
2 years 2 months ago

The leaderboards for these look strange until you realize just how small the samples are for everything but the 90-100% plays….

Shortstops for instance: The average SS had less than 100 combined plays in all of the <90% buckets while having 400-500 plays in the 90-100% bucket. The most statistically significant values are the 90-100% bucket where you'll see noted inconsistent defenders like Starlin Castro, Hanely Ramirez, Eduardo Nunez, Jose Reyes, and Ruben Tejada near the bottom of the league's SS population. This passes the sniff test to me having guys noted for defensive lapses making a relatively lower percent of what is being bucketed as "routine" plays.

tz
Guest
tz
2 years 2 months ago

The SS leaderboard also shows how Jhohnny Peralta ends up with good defensive grades. He gets to 98.5% of the 90-100% bucket (13-14 combined) vs. Cozart and Andrus (his nearest ranked SSs) getting only around 97% of that group. This makes up for Peralta’s very low % made on the lower probability buckets.

Jeter Too
Guest
Jeter Too
2 years 2 months ago

Small sample size compared to his career, but I imagine Jeter has a similar profile that “fools” the eye test. He does well with the routine balls, and pretty well with the Unlikely ones, but is awful with the Likely and About Even buckets, but due to the nature of camera work with ground balls, he makes those ones look really difficult.

Andy
Guest
Andy
2 years 2 months ago

Just taking CF as another example, in one sample, 93% of the plays were either routine or impossible. You wouldn’t expect to get much separation between players on this basis. Maybe there are a few inconsistent players that miss more routine plays than others, but you would expect most players would be distinguished by their performance on plays of in-between difficulty. But there aren’t enough of these to be statistically significant.

Doesn’t this call into question the reliability of defensive sabermetrics? Seems to me that with such small sample sizes, one would see big chances in ratings for any given player from year to year, as in fact we often do.

Andy
Guest
Andy
2 years 2 months ago

big changes not chances

DrFarmer
Guest
DrFarmer
2 years 2 months ago

The lack of precision by putting the majority of plays into a single bucket reduces the utility of this information. We probably need to have precision to a single percentage in the 90-100 range to be able to say anything meaningful.

ed
Guest
ed
2 years 11 days ago

You can’t make a meaningful distinction between a 90% and 91% play, though. I’m not sure how that would help, other than through false precision.

RichW
Member
RichW
1 year 3 months ago

Scotman, I must be looking at a different table than you for the 90-100% column. Ranking 12, 13 14 as Ramirez, Castro and Reyes do does not suggest bottom of the league. Considering JJ Hardy is 15th, this doesn’t tell me anything since I would have ranked Reyes very much lower with my flawed eyes. Also note that the difference between Desmond at 6 and Miller at 21 is less than 2% (98.0-96.2) which is 8 plays in 400 chances. I don’t know what a 96% made rate vs 98% made rate on routine plays tells me.

Ira
Guest
Ira
2 years 2 months ago

Totally unrelated, but would it be possible to include a “Fangraphs” projection on the Projections page which is the average of ZiPS and Steamer?

larry
Guest
larry
2 years 2 months ago

something seems off to me when it shows guys having 100+ “0%” plays.

tz
Guest
tz
2 years 2 months ago

I noticed this too, mostly for OFs. Must be well-hit base hits to the outfield, where there is no reasonable chance to catch the ball.

Darren
Guest
Darren
2 years 2 months ago

ya they must include every single batted ball which would mean line drives dropping 100 feet in front of them.

jim S.
Guest
jim S.
2 years 2 months ago

With MLBAM’s weekend announcement, isn’t this already a dinosaur?

walt526
Guest
walt526
2 years 2 months ago

In 2014, only three parks will have the new system.

Also, these data are publicly available. It remains to be seen to what extent the new MLBAM data will be accessible.

LaLoosh
Guest
2 years 2 months ago

how are they ranked?

LaLoosh
Guest
2 years 2 months ago

ah ok, so no one knows…

RMD
Guest
RMD
2 years 2 months ago

These are definitely tough standards, too. Heyward’s game saving grab shown in the field tracking demo wasn’t even listed as remote (10% or less).

Bill
Guest
Bill
2 years 2 months ago

Seems like most plays are either 0% or 90-100% plays with very little in between. That tells me that the play was either a clean hit given up by the pitcher, or an easy out for the fielder.

camdenhu
Member
camdenhu
2 years 2 months ago

Why does the data from Inside Edge disagree with BIZ/Plays/OOZ from Baseball Info Solutions by so much?

jwise224
Member
2 years 2 months ago

Mike Trout! Do I win a Hardball Times Annual?

pft
Guest
pft
2 years 2 months ago

I love the new information, don’t get me wrong, but one of the things I don’t like about defensive stats is we can not see PbP, or game logs, and can’t manipulate it (H-A splits) like offense and pitching. Just given an season aggregate that one must take on faith.

RA Rowe
Guest
RA Rowe
2 years 2 months ago

Goodbye last remaining productive moments

filihok
Member
2 years 2 months ago

This, like most of the stats on this site, would be better served as a +stat. That is, related to average.

Hanely Ramirez’ 95% of plays in the 90-100% range seems average until you know that SS’s make 97% of those plays.

LaLoosh
Guest
2 years 2 months ago

yeah, seems like this metric needs work with the interpreting part….

dls
Guest
dls
2 years 2 months ago

Shouldn’t nearly every player get a 0% chance on most plays? I guess what I’m saying is… shouldn’t they run this data as a type of “FABIP”? Fielding Average on Balls In Play?

I would be curious to see those numbers if they had them.

The problem of defensive stats is “small sample size”… yet we arbitrarily limit the sample to “plays the fielders should reach”. When technically, every player on the field COULD make every play, except for HRs-BBs-Ks.

And while the reality is most players do not make most plays, I’m guessing that the higher denominator would push down all players… but better defenders would get pushed down less… and would relatively expand the defensive differences between players… which is what we are after. Right?

dls
Guest
dls
2 years 2 months ago

So, every defensive player on the field gets a “play not made” on EVERY play they don’t make (duh!).

EVEN IF they had a ZERO realistic chance to “make the play”. eg. RF gets a “play not made” on a line drive out to the 3B.

Would love to see the raw stats for that.

After that, THEN I would like to see the “%” chance applied (as done here)… and then see how/if that changes things.

dennis
Guest
dennis
2 years 2 months ago

Is it possible for someone to actually make an ‘impossible’ play? Is the play in question counted when determining whether a play is impossible or remote?

For example, if Machado or Simmons makes a play that zero other players have made, does it count as ‘impossible’ or ‘remote?’

Adam
Guest
Adam
2 years 2 months ago

I’d like clarification on that, too. Is ‘impossible’ a play no one has made, or is it a play it’s impossible for that player to make at the position, eg a 3B won’t make a force out at first with a ball hit to the right side?

bstar
Guest
bstar
2 years 2 months ago

Impossible is 0%, Remote is 1-10%

I would change the first one to Virtually Impossible and make it 0-1%.

Adam
Guest
Adam
2 years 2 months ago

Rephrased: why the need to qualify Impossible? How can we accurately measure something that cannot be done?

A Virtually Impossible rating makes more sense.

dennis
Guest
dennis
2 years 2 months ago

I agree that this would make it simpler.

At these margins though I do see a problem with a great defender essentially devaluing himself in this sort of system, because by making more plays, he increases the percentage of made plays in difficult zones.

DrFarmer
Guest
DrFarmer
2 years 2 months ago

One thing that could be done with these data is to these buckets to errors/fielding percentage. One might expect that fielding percentage would correlate well with the routine play percentage.

BaconBall
Guest
BaconBall
2 years 2 months ago

Although I agree this is “fantastically cool stuff,” I must say it seems rather unwieldy due to the far too many divisions. For that reason I combined it to only three. By dividing it into 1-40 (let us call it “difficult”); 40-60 (“average”); and 60-100 (“routine”)-BTW, should that not be 41-60, & 61-100?-it is much more simple and easier to understand. I would also like to see the # of plays made given along with the %.
By combining the sections this way we find that MLB shortstops averaged 17.5% of the “difficult” plays; 47.9% of the “average” plays; and 97.4% of the “routine” plays. I then did the math for four shortstops:

Difficult Average Routine
Andrelton Simmons 29.0% 72.7 97.7
Alcides Escobar 42.3 63.6 96.8
Yunel Escobar 14.8 34.5 98.2
Troy Tulowitzki 22.2 64.7 97.8
This tells me much more than the way you have chosen to present the information. Even better is this:
Difficult % Average% Routine%
Andrelton Simmons +11.5 +24.8 +0.3
Alcides Escobar +24.8 +15.8 -0.6
Yunel Escobar -2.7 -13.4 +0.6
Troy Tulowitzki +4.7 +16.8 +0.4
I first played baseball at nine years of age at a Boys Club in 1960 and played for a decade. I have watched fa too many games in my life to have kept count. Last season I watched maybe 90% of the Braves games, as I did the previous season. I have never, ever, watched a SS who could do the things Andrelton has done. I have been blessed to see this man play baseball. Unless one is fortunate enough to watch him on an everyday basis it is difficult to describe how good he has been in his first two seasons. I must therefore question how it is possible Andrelton could possibly have made only 6.3%, or only one, of the 16 “remote” with which he handled. Obviously this is subjective, but he made at least that percentage of what to me, and many others, would be called “impossible” plays! He is, quite simply, the best human being I have ever seen play shortstop, and with all due respect to the Wizard of Oz and Cal Ironman, it is not even close. The best shortstops of all time are all a distant second.

Jim Kelley
Member
Jim Kelley
2 years 2 months ago

Is there any way to get the inversion of these stats, focusing on the hitter rather than the fielder?

I’m curious whether these bins might useful in predicting babip – whether there is a stable skill in hitting balls that are difficult to field. I would imagine that hitters with the highest BABIPs would have an above average of rate of their balls in play labeled as impossible, remote or unlikely to be fielded successfully. On the other hand, batters hitting into a shift would theoretically have more of their plays labeled as easy to field.

mgoetze
Member
mgoetze
1 year 10 months ago

After checking the stats for Jackie Bradley Jr., I must say that I have no confidence whatsoever in this data.

RichW
Member
RichW
1 year 3 months ago

Is there any way to have the default table list 30 players? Since there is a minimum of 30 players playing each position each day it would be useful to see when “starting” players don’t make the top 30 at certain positions.

Shane
Guest
Shane
10 months 2 days ago

Does Inside Edge Data take into account errant throws to first that the 1B has to scoop, pick, or jump for, and credit them to the 1B as well as the thrower?

Matt
Guest
Matt
9 months 20 days ago

It is counted as a made play if the first baseman catches the ball, or if the first baseman should have caught the ball. Also, there are ratings for every scoop that a first baseman makes with the same categories (likely, certain, etc.)

wpDiscuz