Fielding Independent Offense, Part 2

Dare to dream.

On Thursday, we looked at Fielding Independent Offense (FIO) — as well as the Should Hit formula — and decided to toss stolen bases into the equation. The result were, let’s say, brow-elevating.

Today, we are going to put that result — the FIO formula — into action.

In the timeless words of Sir Samuel Leroy Jackson: “Hold onto your butts!”

Let us begin by addressing the widely discussed — and rightly so — matter of BABIP’s inclusion in FIO. One of the first questions we have to confront is: Does the inclusion of BABIP make this defensive dependent?

Yes, most certainly. But our only point of including BABIP is to get rid of it. Moreover, a hitter’s BABIP (over the course of, say, a season or more) is not really an indication of the defense he has played against, but rather an indication — or the beginnings of an indication — of a player’s BABIP skill. Think about it, when was the last time we used BABIP to decide which defenses were best in the league? (Never?)

I have digressed. The chief purpose of FIO is to put a numerical value to the effect of a change in BABIP. That is why BABIP is in it’s formula, that is why its nearness to wRC+ is essential, and that is why we have gone through this delightful little exercise.

Because here’s the thing, wOBA and wRC+ do not have BABIP as an input. When a player’s BABIP goes down, we have no means of inserting an xBABIP or somesuch into the formula and say, “Well, lookey here! If his BABIP goes back to .300, then his wRC+ goes to 110!”

Enter FIO.

These are the guts of wRC+:

And this is FIO:

BABIP is not the only ball to touch a fielder’s glove in the FIO chart. Some of the great catcher research lately by the likes of Mike Fast and Max Marchi have clearly shown not even strikeouts and walks are defensive independent. I’m digressing again, I realize, but the truth of the matter is this: BABIP is only in there so that we can try to smelt the precious Luck element out of it and then promptly toss the remaining in-play slag aside.

Let’s look at some specific examples of players who had abnormal BABIPs in 2011. With FIO, we can better pin the source of their struggles. Let’s look at:

Adam Dunn
Ichiro Suzuki
Jayson Werth
Jeff Francoeur
Carl Crawford

Let’s start with Dunn:

We have three lines here: His career wRC+, his FIO, and his FIO using only his career BABIP (not his season BABIP). Notice how his CaB-FIO (Career BABIP-FIO) stays much smoother than the other lines. We are employing the assumption here (which is not always true!) that changes in Dunn’s BABIP are mere fluctuations, not changes in his talent level or BABIP abilities (again, this is not always the case with players).

Employing that assumption, we can actually numerically estimate the affect of when Dunn’s 2011 BABIP dropped to .240 in 2011 (from .329 in 2010). CaB-FIO tells us: If Dunn had hit is career-norm BABIP — .292 — then his wRC+ would have instead been 75. In other words,

FIO says Adam Dunn needs to change his approach. Bad BABIP luck cannot be not the only thing at play here.

Ichiro had a rough year too — but don’t go crazy just yet, Mr. Mariner!

Here we see the opposite of Adam Dunn. Ichiro’s CaB-FIO indicates almost all of his struggles in 2011 came from his BABIP!

Which confirms not only my previous ruminations, but also the recent Clubhouse Confidential segment in which they noted the only major change in Ichiro’s numbers was his infield singles — not even his steals or his speed score dropped.

(But now the Mariners are having him change his stance and hit third. I hope that doesn’t ruin everything.) :(

Jayson Werth also had a down year, but he kind of appears in between both Suzuki (all bad luck) and Dunn (all bad approach):

CaB-FIO says he should have hit above average, but still well beneath his career norms. Maybe a look at his near-miss homers might reveal more bad luck, but his walk rate was also down. If he matches last year’s total of 27 homers to go with a career average BABIP, then he’s still only in the 127 neighborhood — which is much less than the Nats paid for.

If Jeff Francoeur has truly changed his approach at the plate, then he needs to convince CaB-FIO:

FIO saw the same guy as last year despite the uptick in steals and BABIP. Maybe he can sustain this new BABIP? Stranger things… Presumably could theoretically happen.

Lastly, by request, Carl Crawford:

Crawford had a miserable 2011 campaign. CaB-FIO says parts were due to bad luck, but the rest may have been the big free agent acquisition pressing at the plate. He had a career-high K%, a pretty dang bad BB%, and his lowest home run total in three years.

FIO says a normalization of luck (presuming his knees can still manage a .328 BABIP) will only bring him near league average. Crawford must stop pressing (or whatever he’s doing to strikeout more) if he wants to rebound in 2012.

So, what now?

I present the FIO Fun! Tool:

Feel free to edit in the white cells! Your changes are not saved to the master copy, so get frisky with it!

And for those who prefer wOBA:

You can download this one if you want to see, like, EVERYTHING. Just click ^ there.

A final note on the FI wOBA: As per another request, I ran these regressions on wOBA as well — to even more brow-elevating results! An R-squard of .97 — which makes sense, given the absence of park factors — makes FI wOBA a more than worthy substitute for FIO. I personally prefer the more intuitive plus scale of FIO, but for accuracy, FI wOBA is bar-none the better choice.

Please enjoy these tools! If you know how to, feel free to embed these spreadsheets wherever — all I ask for in return is links leading people back here (in the faint, distant prayer that people would actually learn how to operate this heavy machinery before lopping off some player’s head).

Image source.

Print This Post

Bradley writes for FanGraphs and The Hardball Times. Follow him on Twitter @BradleyWoodrum.

32 Responses to “Fielding Independent Offense, Part 2”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Brad Johnson says:

    I came late to the last article and didn’t get my question answered so I’m going to repost.

    So‚ĶFIP can be recalculated very easily to adjust for different run environments. Can the same be done with FIO? As in something along the lines of…

    The basic formula for FIO is X and if you think the run environment is different than the default value Y, then you can easily adjust the formula by doing _______.

    Vote -1 Vote +1

    • My *hope* is that FIO will not need to be recalculated for run environments. That will be worth looking into more, but we can see here it sticks close to Adam Dunn and Co. even through the steroid era. If FIO does the same for, say, Jose Canseco or some other staple of the Steroid Era, then we should be gravy.

      I have no idea about FI wOBA. We only just met.

      Vote -1 Vote +1

      • Brad Johnson says:

        Well, to be fair, FIP just barely need to be recalculated year to year at the MLB level. If you didn’t recalculate at all, there’d hardly be an effect. The recalculation component is more useful for lower levels, like if a college coach wants FIP.

        Vote -1 Vote +1

  2. John says:

    I mean, there’s no possible way that luck ISN’T the only variable involved in determining BABIP, right?

    Vote -1 Vote +1

  3. sprot says:

    I think I just had a seizure…Make it stop!

    Vote -1 Vote +1

  4. Yirmiyahu says:

    Coupla things….

    1) You said “Think about it, when was the last time we used BABIP to decide which defenses were best in the league? (Never?)” Have you ever heard of “Defensive Efficiency”? It’s the team fielding version of BABIP, shows what percentage of balls-in-play the defense turned into outs, and is probably the best measure of team defense without getting into the realm of insanely-complicated proprietary formulas like UZR and DRS.

    2) If it has BABIP in it, and the main purpose of the stat is to stick in different BABIP’s to see how it affects the result… “Fielding Independent Offense” is almost exactly the opposite of what it is. I’d rather call it “Fielding-Dependent Offense”.

    3) What about using SB/times-on-base instead of SB/PA? I get that you want to keep it simple, but SB opportunities are highly dependent on BABIP. Can’t steal if you’re not getting on base.

    Vote -1 Vote +1

    • Yirmiyahu says:

      Also, I still don’t understand why you’re primarily using wRC+ over wOBA. You’ve demonstrated that you’re hurting the R-squared by comparing a park-adjusted stat to a not-park adjusted stat. And if you’re looking at an individual player who moved from, say, San Diego to Colorado, your tool is going to be pretty much useless.

      Vote -1 Vote +1

      • Can you think of a player that has moved from SD to Colorado so I can test it? I don’t imagine it would be useless, but at worst a little tougher to interpret.

        Since the chief value of FIO is pulling out the BABIP, though, and comparing it inward — within the player’s own stat line — wRC+ and wOBA do essentially the same thing. I like how wRC+ uses the intuitive 100 plus-scale, but if I need accuracy or want to compare across players (compare outwardly, not just inwardly) then I must use wOBA.

        If you’re asking why I used wRC+ in all the charts above, it’s because I made most of this post before regressing wOBA.

        Vote -1 Vote +1

      • DD says:

        Brad Hawpe.

        Vote -1 Vote +1

      • With Hawpe we can see a clear (and expected) overvaluing early in his career (while with COL).

        But, since the purpose is more to look at the difference between his FIO and CaB-FIO, we see nothing out of the ordinary beyond that — no lucky or unlucky seasons, per se. With Hawpe more than others, the point need to be clear: FIO is not meant to replace wRC+.

        Vote -1 Vote +1

      • Bj says:

        I think yirmi is right about the name being misleading. FIP actually used inputs that are independent of fielding, this is clearly not that. I think Fielding Adjusted Offense is the clearest name for what you’ve created here. Interesting stuff.

        Vote -1 Vote +1

      • J.Ro says:

        Brad Ausmus and Kevin Kouzmanoff played in both places; not sure how sufficient the sample sizes were.

        Vote -1 Vote +1

    • 1) I should have probably said “a hitter’s BABIP,” but your point is valid.

      2) At some point, it’s a logicial preference, I imagine. We call FIP fielding independent, but catchers are HIGHLY involved in that — at least until robot umps come around.

      FIO shares a lot of the elements of FIP, and my only purpose for using BABIP is to pull luck out of it. If the name is all that bothers you, then I think we can call the metric a success.

      3) You’ve hit — as Shakespeare said — the nail on the head. My hope was to use inputs that you can easily grab from a player’s Fangraphs page. I STRONGLY considered using a weighted sum of SB and CS divided by SB chances, but I felt the reward would be too little and the added complication would be considerable.

      Vote -1 Vote +1

      • Yirmiyahu says:

        Just minor complaints. The wOBA vs wRC+ thing bothers me, for some reason. Maybe because my first reaction to a stat like this is to come up with a list of players that are the most over/underrated by it, and see what the inherent biases are. Instead, that’s going to just be a list of guys who’ve spent their careers in extreme parks. And it seems like it’d be so easy to improve it by either using wOBA or by making FIO park-adjusted.

        Thanks for the responses. I should have prefaced everything by saying “great work, thanks for sharing”, so I didn’t sound so nitpicky.

        Vote -1 Vote +1

      • Haha! No problem Yirmi. I’m glad you offered your thoughts — that’s how stuff improves.

        Vote -1 Vote +1

  5. siggian says:

    I’d love to see Jose Bautista’s chart…

    Vote -1 Vote +1

    • If you D/L the chart above there, you can see how I made those charts. It would only take a few minutes, I hope, to make his chart.

      Vote -1 Vote +1

      • siggian says:

        Thanks. I did it and CaB-FIO suggests that Jose was a little above what was expected in 2011 (!) and he should have been more like he was in 2010, which is still pretty awesome. Since his green line wasn’t too far from the other two lines, 2012 should still be a good year.

        For the last 3 years

        wRC+ 102 166 181
        FIO 104 152 184
        CaB-FIO 104 170 169

        Vote -1 Vote +1

      • Good heavens that man is a beast.

        Vote -1 Vote +1

  6. Jonny's Bananas says:

    I think this is excellent work. Easy to use and understand, and fun to play around with. Perfect? Probably not, but neither are 95% of the other metrics on this site. All the haters should come up with their own unique analyses replete with graphs, embedded spreadsheets, and a well-written explanatory post. Great job!

    Vote -1 Vote +1

  7. Barkey Walker says:

    Perhaps I’m missing something.

    Another way to look at this stat is to realize that Woodrum dropped the unlikely stuff in his first arrow graph (HBP, CS, RBOE), then substituted other statistics for doubles/triples (presumably, given HRs, Ks, and BBs you could get a pretty good guess of doubles and triples). then divided everything by PAs.

    What does this tell me? Well, using Career-BABIP I can see if a slump/surge is caused by whatever this years BABIP is measuring. This is something I could have done before, but I would have had to make a singles/BIP, doubles/BIP, RBOE/BIP, … to do this.

    If I thought I knew what this years BABIP was measuring (like I do for pitchers), this would make me happy. However, I’m worried that it is a stand in for strength, swing mechanics… and so I don’t really know what to make of it. Because of that, I’m not willing to say this is worthless or a bad idea, I just don’t know what it is or how to interpret it.

    Vote -1 Vote +1

    • I’m not sure if I totally understand your quandary, but I should have mentioned in the article that we don’t need to use career BABIPs — and xBABIP would do just as good (if not better).

      Players lose speed, change their swing, or decline with age, so it’s unrealistic to expect their same career BABIP every year.

      And yes, you can get a similar product by estimating how many singles and double and etc. they would have had and then putting that into a wOBA formula, but that’s tough guesswork — plus it’s much more complicated.

      Vote -1 Vote +1

  8. Thanks for duo, I’m digesting.

    Vote -1 Vote +1

  9. Jack says:

    CaB-FIO seems to be a great predictor of wRC+ the next year (much like xFIP and SIERA with ERA), do you have leaderboards, this stat would be very useful for fantasy drafts. Keep up the good work and way to become the Voros Mccracken of offense! Here is the average year to year dif between wRC+ and itself, and CaB-FIO and wRC+ for the graphs shown.

    Dunn 18.1

    Ichiro 14.0

    Werth 19.8

    Francoeur 22.4

    Crawford 18.2

    Vote -1 Vote +1

  10. Madoff Withurmoni says:

    Would it be safe to say that this is a tool best used for players who aren’t moving to new parks or had a significant change in batted ball or plate discipline profiles last year…ie Albert Pujols?

    Vote -1 Vote +1

  11. Chris from Bothell says:

    Side note to your side note on Ichi: They’re making him bat 3rd and change his approach for reasons that have nothing to do with his 2011 performance.


    a) artificially making room for the corpse of Chone f’ing Figgins to bat leadoff, in a wacky experiment to get some value out of the rest of his contract
    b) when that inevitably fails by about early May or so, starting to get an early look at the next leadoff hitter for the M’s (Ackley?), as a hedge against the possibility that Ichi doesn’t negotiate another contract this coming fall and ends up hangin’ em up
    c) having a veteran start off the traditional ‘heart of the order’ since we finally have a lot more kids than usual in the projected lineup, and having the still-best-bet-to-be-perpetually-on-base guy in front of Smoak, Montero, Carp, etc. should give these guys a bunch of chances to really produce.

    Anyway. This is a nifty tool, though I’d wished the explanation of FIO was as simple at the start of the first article… it took me a moment to get how this would actually be used, rather than follow the thread of inquiry that led you to researching this. Having a tidy answer to “can I handwave this guy’s season / issues away with his BABIP?” is rather nice.

    Your real-world examples above were pitch-perfect, and an approach I wish more fangraphs writers would use when laying out their arguments. (“here’s player foo…. here’s the chart of what I’m measuring… here’s what each trend line says… assertion based on stat in bold… advice to player foo in plain english… dread warning of the possible future if this is not heeded”). Very well done.

    And yes, in answer to previous people above, sabermetricians do frequently seem to use BABIP as a synonym for luck, in an effort to avoid having to actually say “it was random”, “I don’t know”, “no one can measure it”, etc. They seem a lot more comfortable in that scenario if they can attach an acronym and a number to it. ;)

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current day month ye@r *