Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Friday, March 31, 2017

Statcast ranking of top defensive outfielders | MLB.com

More great stuff.

Jim Furtado Posted: March 31, 2017 at 08:43 AM | 27 comment(s) Login to Bookmark
  Tags: fielding, statcast

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. TDF, trained monkey Posted: March 31, 2017 at 09:33 AM (#5426550)
So the Reds had the best-fielding CF and RF last year; they also have a very good SS.

Imagine what that pitching staff would look like with more average fielders.
   2. Rally Posted: March 31, 2017 at 10:29 AM (#5426602)
Correction - best fielding CF and LF. Reds pitchers gave up 258 homers (37 more than 2nd worst) and walked 636 batters (33 more than #2). Thanks to defense they only allowed the 4th most runs.
   3. Mefisto Posted: March 31, 2017 at 11:09 AM (#5426649)
It seems to me that this metric doesn't account for positioning, so it defines "top defensive" too narrowly. What it does do is tell us who had the most range.
   4. Walt Davis Posted: March 31, 2017 at 06:19 PM (#5426954)
#3: That's a feature not a bug. This is based just on the talent of the player. Their teams may or may not do a good job of positioning them. It is a counting stat not a rate stat -- i.e. maybe the Red Sox did a good job of positioning Betts, giving him more opportunities ... or maybe they did a lousy job, creating more difficult chances for him.

This at least confirms (or feeds into my confirmation bias) that Heyward played RF as well as I've ever seen it played. I'm not sure if this includes his time in CF or not but +17 in just over 1000 innings (or exactly 1200) is pretty impressive. However, that +17 seems to disagree with their charts and star rankings ... i.e. when those were released, I looked at Heyward and he didn't have a LOT of 5-star catches and, eyeballing the list, I expected him to come out as above-average to good, not excellent.

These are plays made. Seems simple enough to convert to runs. Seems to me there are two types of pretty amazing plays for OFs -- charging in/across to steal a single and flying into the gap to steal a double/triple. The first one is worth about .75-.8 runs, the latter should be worth at least a run, maybe as much as 1.1 for a double.**

** Occurs to me triples are so rare these days and, in many cases, a function of batter speed not where the ball is hit, that maybe we should never expect a hit to result in a triple for the purpose of converting OF plays made to runs. Maybe these sorts of gap balls and wall-bangers should be scored at 2.1 bases or something. (Maybe they already are.)
   5. cardsfanboy Posted: March 31, 2017 at 06:32 PM (#5426960)
Agree with Walt, or to put it another way, this isn't a value metric, it's a skill metric. With this data, teams are going to be able to improve positioning for their players, if they embrace it. (I mean at some point someone is going to come up with a measure on positioning, and at it's simplest it's going to say what spots that you can position an average player and hope for the least amount of likely damage vs an average left hander, average right hander...and eventually it will get deeper, average "stlye" left hander---meaning fly ball, pull, slap or whatever...and of course with a large enough sample size, what is best positioning against that individual)

   6. This is going to be state of the art wall Posted: March 31, 2017 at 06:54 PM (#5426966)
Adam Eaton had such an amazing year. I will miss watching him in RF. Enjoy Nats fans, you gave up a lot, but you got a great player.
   7. 6 - 4 - 3 Posted: March 31, 2017 at 07:17 PM (#5426976)
Based on how poor a defensive 3B Duvall was when he was a Giants prospect, I'm really surprised that he's now a top defensive LFer. I'm happy for him that he's found success with the Reds, but his emergence to being an above-average player is pretty unexpected.
   8. Walt Davis Posted: April 01, 2017 at 12:53 AM (#5427030)
I feel morally obligated to add that I fine these numbers pretty hard to believe as well. I still don't see how an OF can catch 22 extra outs. According to Inside Edge, Heyward had only 35 opportunities in RF or 42 overall which were non-routine. He caught only 18 or 20 total. An average RF is still gonna catch some of those, maybe as much as 40%. (Too tired to do the math on Heyward's distribution buy eyeballing it that seems about right).

Anyway, somebody more patient than I am will have to try to reconcile those two. That said, statcast so far seems closer to UZR and Rfield than to IE so it's more likely IE is wrong and I have no investment in IE.
   9. Sunday silence Posted: April 01, 2017 at 05:39 AM (#5427038)
But Walter those numbers that they give in the linked article (there are links to more articles in the story) those arent individual catches; that is a cumulative number; summing up all the fractional catches he's made. NOt sure which numbers you're looking at but I am looking at those lists for LF CF and RF where for example Betts in RF has +22 and Hamilton in CF has +20. That is the sum total (rounded to whole number) of all the catches he made that season.

SO in your example maybe Heyward caught 100 balls that were of the routine (Say 80% likely) type. He would get credit for +20 on those catches alone. At least that is how I read that.

****

Even more interesting about the article: we have been programmed to believe that CF places a greater emphasis on fielding range that the two corner OF positions. Which superficially seems correct since the CF can range to his right or left to get flyballs. But as you compare the numbers given for RF and CF (LF apparently have less players play there on a full time basis) they are very similar; the best guys top out at about +20 and a number of them in the 10-12 range.

WHat is going on there? Well if you think about it, RF can just as obviously range to their left or right to make catches as well as the CF. After all you wouldnt position a RF standing on the RF foul line or (to stretch a point) have him stand along the edge of the rail in foul ground. No. RF can range just as much to their left as the CF its just that if they make a catch at the extreme limit of their ability moving to their left, its a catch in foul ground whereas a CF making an extreme catch will almost always be in fair territory.

2 more thoughts; why do CF traditionally have greater range factors that LF and RF? (i think its about 2.2. for cf vs 1.8 for RF for top defenders)

1) At first I was thinking perhaps you just player your fastest player in CF so that he naturally is going to have a better range factor anyhow; but now I think:

2) the CF is covering both power alleys in left center and right center which is normaly going to be greater area than a RF covering one power alley plus a shortened area near the foul pole in RF. The good example is how the Pirates were playing Marte and McCutheon last year owing to an extra large RF area in PNC park (was it? going from memory).

ANyhow its a really good article which makes me rethink how we view fielding range.
   10. Sunday silence Posted: April 01, 2017 at 05:55 AM (#5427039)
ANyhow following up on Walt's post I was thinking about how lots of times people will say they dont see how Heyward or Marte or whomever can be a plus 20 defender (20 defensive runs saved above average). I dont think Walt is saying exactly that just that his post reminded me of this...anyhow:

If we take the numbers they give for Heyward what is it +17 in RF? (I dont think they are counting CF there); we can assign run values based on what Walt said in post 4. NOTE +17 is not a number of catches its a cumulative sum of all his catches; after assigning a probability value to them. But we can still use them as whole numbers.


Lets assume Heyward catches 4 balls to his extreme left 4 to his extreme right 4 to his extreme forward and 4 to his extreme rear; and 1 more we can throw in later.

The 4 in his front are likely singles so 4 x 0.75 = 3 runs.
4 to his right are likely doubles so 4 x 1.1 = 4.4
4 to his left are more likely foul balls? lets say 2 fouls and 2 fair (assume 0.5 run value for catching a foul)
so thats 2.75 runs

4 balls over his head, lets say 3 are doubles and 1 HR? I guess that makes 4.7 runs (HR being 1.4 runs)

Plus one more catch to add lets just make it a single that totals 15.6 runs on catches.

Thats not counting holding runners to one base and any outfield assists. My guess from memory is that holding runners to one base is probably half the value of the catches so an exceptional arm like Barfield might save you 7 runs per season on holding runners.

And there's still OF assists which could also be another 10 runs in an extreme case?

SO it depends on how many skills you can bring, but its not hard to see a very good outfielder saving 20 runs on defense perhaps as many as 30 if he has exceptional skills an exceptional season.
   11. Ned Garvin: Male Prostitute Posted: April 01, 2017 at 02:19 PM (#5427134)
Re: #9

I think CF has the greater range factor primarily because more balls are hit to them than to the corner OF. In other words, while your two ideas may contribute, I think that even if you put 3 identical clones in the 3 OF positions, the CF would have the higher range factor due to the distribution of batted balls. That is why you put your rangiest OF there. This is similar to SS, which is basically identical to 2B (other than the slightly longer throw) but since more balls are hit that direction, you should put your rangier guy at SS and the worse guy at 2B, all else equal.
   12. Sunday silence Posted: April 01, 2017 at 05:35 PM (#5427205)
I think CF has the greater range factor primarily because more balls are hit to them than to the corner OF.


which is what I used to think but there's really no reason to think that is there? Theres no reason to think that more fllyballs fly straight then fly to right or left is there?

It seems to me that there should be a fairly even proportion of balls hit to the various directions in the outfield PROVIDED that we include foul territory. Of course if there is little room for foul territory down the lines say in OAK then theres less room for the RF to make plays there.

As I said above, I think the only reason there's a difference in range is that there is more territory in deep center and in both power alleys.

Plus what really got to me is why doesnt BIlly Hamilton have better numbers than say Mookie Betts or adam Eaton. I mean if what you say is true then why not?




****

Looking at the table that is linked to the original story there is some interesting stuff there. Ichiro! has just about the very same range ability as starling Marte. Nori Aoki is really good at balls hit at him. Hunter Pence has limited range and is not even good on balls hit right at him; maybe his days are numbered? If you dont count balls on the extreme edge of range Heyward might be the best he rarely misses on balls hit at him and catches a huge amount of those he has to run for. Bourgos is not too bad either.

***

NOte in the figures I gave in post 10, it should be 2.5 runs for fly balls hit to left of the RF not 2.75.

THe final value of 15.6 runs is correct; I was doing some editing and failed to change all the numbers.
   13. Walt Davis Posted: April 01, 2017 at 06:57 PM (#5427234)
#9 ... I'm well aware of that. But routine balls are caught 99.4% of the time. Not 80%, 99.4%. Basically the only time routine balls aren't caught is a brain fart error, getting lost in the sun, running into another OF, etc.

So yes, Heyward's accumulated +17 catches has to come almost entirely on the non-routine. Or there is something very wrong with how Inside Edge creates their categories such that Heyward's "routine" is like twice the range of Bautista's "routine" -- but that's not what they're trying to achieve. In 2016,the average RF faced almost exactly 6 routine plays for every one non-routine but Heyward's ratio was below that so it's not likely they were classifying his non-routine as routine.

It's possible that in IE's "likely" category (60-90% caught) that Heyward's happened to tilt toward the 60% range. But even if you add it up under an assumption that all of his chances in a category were the worst possible, then I get a max extra catches of 10.
   14. . . . . . . Posted: April 01, 2017 at 07:10 PM (#5427241)
I second the point someone made above highlighting Ichiros numbers. He is 43 and is competitive with the best defensive outfielders. He must've been an absolute monster in his prime. Shame we'll only be able to guess at it with pbp stats.
   15. Stormy JE Posted: April 01, 2017 at 07:12 PM (#5427245)
Left field
+12 Adam Duvall, Reds
+9 Christian Yelich, Marlins
+8 Brett Gardner, Yankees
+6 Angel Pagan, Giants
+5 Michael Conforto, Mets
Michael Conforto. Imagine that.
   16. Sunday silence Posted: April 02, 2017 at 05:46 AM (#5427362)
Walt I am not sure where you are having trouble with the math on this one, but it does seem to check out at least by making some rough estimates and then plugging the values in. Based on the tables for each player that the story links to.

NOTE: 5 star plays, are 0-25% probability, 4 star: 25-50%; 3 star 50-75; 2: 75-90; 1: 90-100
so the point being that the intervals are not all equal they are smaller intervals at the low end.

So I tried to be liberal in making a guess at the value of each category so that the numbers could be at least plausible. I make the following estimated values:

5 star value: 0.9
4 star: 0,7
3 star: 0.4
2 star 0.2
1 star: 0.1

obviously I tilted a little liberally in favor of tougher catches/higher reward but Im just trying to see if the numbers will fit. I tested for 3 players: Betts Hamilton Heyward. Here is the result:

There's 8 columns that break down as:

PLAYER.cumulative value 5 star plays .cum value 4 star ..cum val 3 star. cum. 2 star..cum. 1 star..My estimate total value..Value given in the article


BHam..6.5 7.0 5.2 1.2 2.0...21.9...20
JHey...1.0 5.9 4.8 4.2 2.9 ...18.8...17
MBet...0.5 4.2 9.6 1.4 5.0 ...20.7...22

you may be right in the case of Betts he's getting 5.0 catches in the one star column on the basis of going 59/60 and me giving him a very liberal 0.1 per catch. If it was only 0.05 per catch that value would be only 2.0 and his overall goes to approx 18.

But the other columns and other players the math seems to check out. Both Hamilton and HEyward came in about 2 catches more using my more liberal values so I dont think theres any problem with their values.

NOTE: I did this real fast literally on the back of paper w/o calculator so there may be slight error somewhere.
   17. Sunday silence Posted: April 02, 2017 at 06:29 AM (#5427364)
Oh hell this is so much fun lets do it for five more OFers. Keep the same values as above but this time a one star catch value = 0.05 (not 0.1) so if you miss a routine ball its -0.95 catches. Ok here we go


CoCrisp.....-0.2 2.8 0 -1.0 -1.4.....0.2
LoCain.... -1.8 3.2 4.0 0.4 -0.15....5.65
AMcCutch...-3.2 1.9 -2.2 -0.4 -5.65...-9.55
ISuzki...... 0.7 2.3 0 -0.7 0.65......2.95
HPence... 0.2 0.3 -1.0 -1.4 -0.75... -1.65

NOTE: here the article did not mention their estimated total value so the final value is the total value I came up with.

SOME MORE OBSERVATIONS:

1) the methodology they used and my argument above suggests that you can actually interchange the values of corner outfielders with centerfielders, which is pretty surprising and I dont think that's ever been suggested. I dont see any reason why there would be any difference in the number of balls hit to the extreme range of Suzuki whether he's in CF or RF. The only factor is how well he can get to such balls.

2) you can probably assign a run value per catch based on the weighted runs I suggest in post 10. So for corner outfielders I would assign 0.9 runs per catch and CF about 1.05 runs/catch. So for instance McCutcheon in CF cost the Pirates about 10 runs and if he played RF 8.6 runs.

3) If SUzuki played CF in PIT he would have saved the Pirates 12.5 catches or 13.1 runs on defense.
   18. Lassus Posted: April 02, 2017 at 06:55 AM (#5427365)
Michael Conforto. Imagine that.

I was going to mention that earlier, but was basically gobsmacked. I can see it, though.
   19. cardsfanboy Posted: April 02, 2017 at 11:21 AM (#5427417)
2 more thoughts; why do CF traditionally have greater range factors that LF and RF? (i think its about 2.2. for cf vs 1.8 for RF for top defenders)


I haven't read any posts after this, so if it's already been discussed, I will get caught up with it as I read on.

Anyway, traditionally the centerfielder is the captain of the outfield, and he is going to haul in any discriminate play, meaning if he has a chance at it, he's getting it. He's also traditionally the best quality fielder on the team, so of course they are going to let him catch it. Add in that traditionally the centerfielder is where you put your fastest player and as show with Hamilton's ratings on this, speed is the number one factor in regards to defense in the outfield(remember, Hamilton is a converted shortstop, he hasn't been playing in the outfield that long, and he did make a few really bad route errors when he started playing in center, he's gotten much better since then, but you have to applaud the Reds for sticking with him, which allowed him to not be a useless player)



Edit: Post 11 talked about higher percentage of balls hit to center, which is also a factor, but considering that batters are generally pull happy, I'm not sure that it's true, although balls hit to the gaps are probably more fieldable than balls hit down the line hard.
   20. cardsfanboy Posted: April 02, 2017 at 11:30 AM (#5427423)
Plus what really got to me is why doesnt BIlly Hamilton have better numbers than say Mookie Betts or adam Eaton. I mean if what you say is true then why not?


Because the players are being compared to players at the same position, not outfield overall. Just for illustrative purposes, most teams put their best defender in center, their best arm in right, and they use left for their best bat. Now it's probably not a stretch to assume that if outfield was evaluated as a whole, that cf average would be +10, right would be 0 and left would be -10. In this hypothetical moving someone like Hamilton to right would turn him from a +20 to a +30 and in left he would be a +40 (there are plenty of other factors involved including amount of terriotory to cover, which would limit Hamilton's biggest asset-speed)

Or put it another way, if you watch someone like Matt Holliday play outfield for a few years, you know he's just not that good at it, over his career by rField he is roughly -3 runs per full season played. It is not hard at all to imagine that in comparison to the worst everyday centerfielder in baseball, that he's easily -10 runs below them if he played their position.
   21. Sunday silence Posted: April 02, 2017 at 07:10 PM (#5427691)
CFB: I am not sure that's true. I mean that would be true for a system like Total zone or maybe BIS but I was referring to the stat cast numbers.

Until we know a little more about the statcast methodology I dont think you can say that they are comparing CF to CF etc. From the article it seems like they are assigning each fly ball a distance, a velocity, probably an angle etc. Then they are calculating that players distance from each fly ball and assigning it a probability.

If Betts standing in RF is 100 feet away from a ball with a 2 sec hang time I dont think its any different than Hamilton being 100 feet away from another ball with a 2 sec hang time. One ball might be in RF where only a RF can get it; and another might be in CF where only the CF can get it. I dont believe they are calculating that probability based on similar players.

At least I dont think so but until we know more about that we cant say for sure.
   22. cardsfanboy Posted: April 02, 2017 at 07:39 PM (#5427709)
Until we know a little more about the statcast methodology I dont think you can say that they are comparing CF to CF etc. From the article it seems like they are assigning each fly ball a distance, a velocity, probably an angle etc. Then they are calculating that players distance from each fly ball and assigning it a probability.


The probability isn't grabbed from thin air, it's grabbed from other centerfielders. Why would any system ever grab it from the entire field of fielders?
   23. Sunday silence Posted: April 02, 2017 at 08:17 PM (#5427732)
how do you know that? where in the article is it saying this?

I'll go back and check the article maybe I missed it.

EDIT: Ok it is saying that or something like that. so I think you are right.

Its real confusing how its worded:

(Remember that these numbers are specifically relative to each position, so while Betts' +22 is the highest number you see here, the bar is lower in right than it is in center, so when compared to all outfielders, Inciarte (+25) and Hamilton (+24) just barely edge him out.)


"... the bar is lower in right than it is in center

OK I get it he's being compared to other RF so maybe they are not as good as CF. We assume that's so maybe it is but whatever he's being compared to other RFs because they suggest that the bar is lower which can only mean that there's a different standard there...

"...when compared to all outfielders, Inciarte (+25) and Hamilton (+24) just barely edge him out


WTF is that supposed to mean? If CF are being compared to a higher standard then there is no telling how much better Inciarte and Hamilton are. Correct? Thus the wording "barely edge him out" is just either wrong or misleading or both.

I get what CFB is saying. The article seems to suggest that. But then other wording is really wishy washing. But also reading about the methodology and seeing the graphics that they posted in other videos I am not sure they dont just calculate speed and angle and distance and figure it from there.

Also or moreover they say the numbers are "specific relative to each position" which is not clear in my mind. If they are actually doing it like CFB is saying where they try to figure out what the average RF can get to or if that phrase means that Heyward gets to 30% vs 20% other RFers.

   24. bookbook Posted: April 02, 2017 at 09:24 PM (#5427803)
I can't get to the article. A spam sweepstakes takes over my device every time I try to click on it.
   25. Sunday silence Posted: April 03, 2017 at 07:56 AM (#5427941)
this link is to the sheet that has the tabulated values maybe you will have better luck:

https://baseballsavant.mlb.com/catch_probability_leaderboard
   26. TDF, trained monkey Posted: April 03, 2017 at 09:37 AM (#5427992)
"...when compared to all outfielders, Inciarte (+25) and Hamilton (+24) just barely edge him out


WTF is that supposed to mean? If CF are being compared to a higher standard then there is no telling how much better Inciarte and Hamilton are. Correct? Thus the wording "barely edge him out" is just either wrong or misleading or both.
Here's my reading: The key is "these numbers are specifically relative to each position".

A X-star play for a CF is not the same as a X-star play for a LF or RF, so (despite what the quote says) you can't really compare Betts' +22 to Inciarte's or Hamilton's score. In the Duval video in the linked article, he ran 44 feet in 3.1 seconds for a catch with a 14% chance for a LF; meanwhile, in another video Hamilton has to go 64 feet in 3.77 seconds for a similar 14% chance in CF.
   27. Sunday silence Posted: April 03, 2017 at 02:35 PM (#5428223)
can someone email the writer and clarify this? I tried to google his email but all got was linked in which Ive never really used even thoughIm set up for it and I clicked it and then it wanted a password and all that. I got a million things going on right now so if someone could do that it would be nice.

I think CFB is probably right about that but then again I can see a system where they simply plug in distances and times and produce probabilities. It wouldnt shock me since it might be way harder to plug in those numbers into various spots in the outfield. Or if they dont do it that way then they have to determine which guy made the play and what position is he playing when some guys can play more than one OF position.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

News

All News | Prime News

Old-School Newsstand


BBTF Partner

Dynasty League Baseball

Support BBTF

donate

Thanks to
Mike Emeigh
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogHowie Kendrick Named NLCS MVP After Nationals Sweep Cardinals
(16 - 10:20pm, Oct 16)
Last: SoSH U at work

NewsblogDo Giants' Buster Posey, Johnny Cueto have two of worst MLB contracts?
(8 - 10:20pm, Oct 16)
Last: Cooper Nielson

NewsblogCatch-All Pop Culture Extravaganza (October 2019)
(400 - 10:08pm, Oct 16)
Last: What did Billy Ripken have against ElRoy Face?

NewsblogHow Baseball Cards Got Weird
(32 - 9:50pm, Oct 16)
Last: Ziggy is done with Dominican discotheques

NewsblogAre the World Series-bound Nationals throwing shade at Bryce Harper?
(1 - 9:44pm, Oct 16)
Last: Bote Man sez Deivi is MoY

NewsblogAngels To Hire Joe Maddon
(32 - 9:35pm, Oct 16)
Last: Jack Sommers

NewsblogDog Pee or Scorpions? Cubs Prez Theo Epstein's Wild Landlord-Tenant Dispute
(18 - 9:30pm, Oct 16)
Last: What did Billy Ripken have against ElRoy Face?

NewsblogChampionship Series OMNICHATTER!
(769 - 9:02pm, Oct 16)
Last: Bote Man sez Deivi is MoY

NewsblogTyler Skaggs: Fentanyl, oxycodone, alcohol led to death| LA Times
(71 - 7:44pm, Oct 16)
Last: DanG

NewsblogInvestors Get Path to Buy Stakes in Major League Baseball Teams
(5 - 7:19pm, Oct 16)
Last: Brian C

NewsblogOT - NBA thread (pre-season)
(497 - 6:43pm, Oct 16)
Last: TFTIO is building his own mealworm farm

NewsblogESPN may re-work 'Sunday Night Baseball,' but A-Rod is safe
(31 - 5:29pm, Oct 16)
Last: Brian C

NewsblogDodgers to reassign pitching coach Rick Honeycutt, Mark Prior to take over
(28 - 4:37pm, Oct 16)
Last: Sweatpants

NewsblogIdea of Red Sox going with internal candidates may be gaining steam
(5 - 4:18pm, Oct 16)
Last: Bring Me the Head of Alfredo Griffin (Vlad)

NewsblogPrimer Dugout (and link of the day) 10-16-2019
(2 - 3:51pm, Oct 16)
Last: vortex of dissipation

Page rendered in 0.4778 seconds
46 querie(s) executed