The Problem With Position Adjusted Stats

This came up in the Coco Crisp trade thread yesterday afternoon, but I felt like it’s worth a larger mention. It’s one of the things that I see repeated frequently among fans and analysts alike, but it just lacks the validity of it’s claim. I’m referring to the notion that a player who is moved from an up the middle defensive position to a corner defensive position loses a significant amount of value – you’ve probably seen this stated as something like the following:

David DeJesus is a very good center fielder, but his bat makes him just an average left fielder.”

These kinds of statements are mostly due, in my opinion, to the rise of metrics that compare players against an offensive baseline of other players who play the same position. Since the average hitting LF is significantly better than the average hitting CF, moving a player from CF to LF will make him look quite a bit worse. He’s now getting compared to a better crop of players, so his relative ranking falls.

The problem, however, is that the relative drop in offensive value is almost entirely offset by a relative increase in defensive value.

Let’s look at LF/CF, for instance. Last year, major league left fielders hit .269/.344/.442, while major league center fielders hit .268/.334/.420. Over a full season, the offensive difference between a corner OF and a CF equals about 12-14 runs. Over that same full season, the defensive difference between a corner OF and a CF equals about 9-10 runs.

The net difference in moving a quality defensive CF to a corner OF spot will be a loss of somewhere between 2-5 runs, thanks to the decreased amount of opportunities that CF will get playing in a corner. Of course, if you replace that CF with another CF who is significantly better with the glove, you can get those runs right back by getting the premium defender more opportunities.

We really need to get over this idea that guys like David DeJesus, Carl Crawford, and Ichiro Suzuki lose a significant amount of their value because they’re not playing center field. The difference in their value in CF vs a corner is pretty small, and there are many cases where it certainly makes sense to have a premium defender in an outfield corner.

We have to get away from this notion that a good defender is wasted in a corner. It’s just not true.




Print This Post



Dave is a co-founder of USSMariner.com and contributes to the Wall Street Journal.


22 Responses to “The Problem With Position Adjusted Stats”

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

    While you do make a point for the arguments sake, what if we consider what the Royals are losing offensively from having Crisp in center over DeJesus? Career OPS+ wise, we’re talking nearly 20 points. That is substantial.

    Also, why the assertion that DeJesus would flourish in left field? While he presumably will, is there not a chance that he moves from being an alright defensive CFer to being an alright defensive LFer?

    Vote -1 Vote +1

  2. Kurt says:

    DeJesus has played a lot in left field over the last few years since the royals have pick up Joey Gathright. Last year half of his time was in left field and half in center. BP has his rate2 last year at left field as 123 while center was 98.

    I think the reason people refer to the decrease of the value of a player when then change to a more offensive position is completely because of fantasy baseball. Fantasy baseball has made us compare players offensively based on their defensive position because that’s how you fill out your fantasy baseball team. And a lot of fantasy baseball is geared to offense only. Not that we shouldn’t compare players based on the position they play, we just need to look at the whole picture. I’ve seen that same claim many times and I just blindly agreed. Good Catch Dave!!

    Vote -1 Vote +1

  3. JWay says:

    I see how fantasy baseball sways the argument, but after Dave’s post a while back about the ’01 Mariners outfield which featured Winn/Cameron/Ichiro! The team had basically 3 centerfielders patrolling the grass. I would hate to be a hitter going up to bat and seeing that outfield.

    Vote -1 Vote +1

  4. JLP says:

    Brandon,

    The odds of DeJesus being better in LF than CF is probably due to the fact that he’d have less ground to cover, providing that his replacement in CF can cover what a centerfielder should be responsible for.

    And, when asking about offensive numbers, you should consider who the Royals are dropping from the outfield. I’d imagine Joey Gathright will get less time, and, let’s face it, the guy isn’t exactly golden at the plate. If anything, it’s an offensive upgrade.

    Vote -1 Vote +1

  5. JWay says:

    I would think playing CF would wear you out more too. Having to back two guys up instead of just one. Over a few games no big deal, but for 7 months it adds up.

    Vote -1 Vote +1

  6. BigRedMachine4Ever says:

    If we recognize that LFs are the worst defenders in baseball, and they are, then you’ve pretty much got to recognize that the players there (the bulk of whom you’d want at no other OF position) have pushed so close to the bottom that the differences are meaningless. Ambulatory with two legs is ambulatory with two legs and that describes most of the fielders at that position. As more time passes, it gets hard to treat fielding numbers in LF as anything but noise. For instance, Manny Ramirez did not improve in the field when he went to Los Angeles.

    The reality is LF is the home of guys who can mash and that almost none of them are worth a positive mention in the field. Some years the numbers swing their way, other years the numbers don’t. You’ve got two honest-to-goodness good defenders (Crawford and Eric Byrnes) who’ve been able to hold down a long-term job in LF and that’s about it. Most of the guys who can field a bit with just enough bat fade quickly to nothingness.

    I know you are a fan of defensive metrics that overvalue the defensive contributions of players. Defense in left field is a minor factor for a team’s overall defensive ability because there are so few balls hit to left field. Most of them are either routine plays that any fielder will make, or they are impossible plays that nobody makes. There are very few opportunities for a left fielder to make significant positive or negative defensive influence on a team’s record.

    The guys that spend all their time developing defensive metrics tend to exaggerate the importance of their results. Otherwise people wouldn’t pay much attention to them if they realized that defense was proven to be dwarfed in importance by hitting and pitching.

    Dunn’s defensive shortcomings (even when exaggerated) for instance do not come anywhere close to canceling out the excellent offensive production he provides.

    Dunn’s offensive production is elite. Therefore he is an elite player. Which players make the most money — great hitters or great fielders? Hitters do, because they have the greatest impact on the team’s chances of winning the game.

    Vote -1 Vote +1

    • Jack Str says:

      “Ambulatory with two legs is ambulatory with two legs and that describes most of the fielders at that position.”

      Not the case. A typical LFer is still a tremendous athlete.

      Vote -1 Vote +1

  7. Teej says:

    Dunn’s defensive shortcomings (even when exaggerated) for instance do not come anywhere close to canceling out the excellent offensive production he provides.

    Yes, they do. Dunn’s bat was worth about 25 runs more than average this year. Over the past three seasons, though, Dunn has given up an average of about 20 runs per season more than the average left fielder, according to Dewan’s +/-. So, with some pretty rough math, Dunn is barely above average as a full package.

    Dunn’s offensive production is elite. Therefore he is an elite player.

    No, his offensive production makes him an elite hitter. He is not an elite baseball player.

    Which players make the most money — great hitters or great fielders? Hitters do, because they have the greatest impact on the team’s chances of winning the game.

    Good hitters get paid more than good fielders because baseball as a whole doesn’t understand the value of defense. A run saved is 100% as important as a run scored.

    Vote -1 Vote +1

  8. Teej says:

    Clarification: Above, I should have said Dunn was an average of 20 plays worse than the average left fielder, not runs. I can’t recall the translation of plays to runs, but I think 20 plays is about 17 runs? (Dave can correct me on that.) But I’m positive that my point still stands: Adam Dunn’s defense takes away a pretty big chunk of his offensive value.

    Vote -1 Vote +1

  9. Ray says:

    Teej, your translation is correct.

    Vote -1 Vote +1

  10. DanDuke says:

    How is that ratio of plays to runs calculated?

    Vote -1 Vote +1

  11. BraveBronco0121 says:

    Does it go the other way, though? I’m ok with saying that a center fielder that moves to left is probably not going to lose any value by doing so, but I would think that, assuming equal offensive production, a perfectly average center fielder is more valuable than a +5 left fielder. My understanding is that a center fielder moving to a corner picks up about 10 runs over what he was worth defensively in center. This may be rather nit-picky as we’re only dealing with 5 runs difference in a hypothetical situation, but I tend to think that positional adjustments are still justified on some level.

    Catchers are of course unique animals and probably deserve a significant adjustment upwards, and I’m inclined to think that those on the right side of the spectrum still need a bit of downward adjustment based on their inability to move around the diamond and pick up defensive value at an easier position.

    It is an interesting argument, though, as positional adjustments are kinda taken for granted in the stats community.

    Vote -1 Vote +1

  12. Dave Cameron says:

    The odds of DeJesus being better in LF than CF is probably due to the fact that he’d have less ground to cover, providing that his replacement in CF can cover what a centerfielder should be responsible for.

    Actually, I think the reason CFs perform better in LF or RF is simply due to the change in who they’re being compared to. The rankings are relative to the player’s peers, and when you’re getting compared to Ibanez, Dunn, and Ramirez instead of Sizemore, Gomez, and Beltran, you’re going to look better.

    Defense in left field is a minor factor for a team’s overall defensive ability because there are so few balls hit to left field. Most of them are either routine plays that any fielder will make, or they are impossible plays that nobody makes.

    Why do you believe this is true? It’s clearly not, but I have no idea why you’d even think this.

    How is that ratio of plays to runs calculated?

    Each play made is about .8 runs.

    I tend to think that positional adjustments are still justified on some level.

    Oh, I’m not arguing that we just stop doing position adjustments. I am arguing that there’s a better way to do it than how VORP and stats of that ilk currently handle it.

    Vote -1 Vote +1

  13. lookatthosetwins says:

    As a twins fan I saw a lot of this where people would say that Denard Span is being “wasted” in right field. He definately should have been in left, with the much larger field of play in left in the metrodome, but either way, his defensive contribution in right was still substantially better than Cuddyer’s would have been. Just saying that Cuddyer could handle right doesn’t change the fact that Span made so many plays that Cuddyer wouldn’t have. When Cuddyer goes back to starting in right, Every flyball heavy pitcher on the twins staff (which is essentially every pitcher on the twins staff) suffers.

    Vote -1 Vote +1

  14. I love the article! Great thinking, I’ve always been vaguely uncomfortable with positional adjustments but didn’t know why.

    People, you need to get out and vote for Dave, he needs a LOT of votes!!! http://www.collegescholarships.org/blog/2008/11/06/vote-for-the-winner-of-the-2008-blogging-scholarship/

    As a clarification, when you note that each play made is worth about 0.8 runs, I assume that is for outfield plays, where a play not made most probably is a double and sometimes a triple. What is it for the infield? Even there, I assume there is a difference between MI and CI in that the CI plays are worth more as the missed plays go into the corners for doubles more often than for MI, I would think.

    I was going to examine the difference in positions by plays handled, but regular defensive stats don’t differentiate between putout by someone throwing to you or you taking it to the bag. Perhaps Dewan has that type of data.

    Vote -1 Vote +1

  15. JLP says:

    [i]Actually, I think the reason CFs perform better in LF or RF is simply due to the change in who they’re being compared to. The rankings are relative to the player’s peers, and when you’re getting compared to Ibanez, Dunn, and Ramirez instead of Sizemore, Gomez, and Beltran, you’re going to look better.[/i]

    Dave,

    While I agree that perception plays a large part, doesn’t skill also come into play? I’m sure we all agree that CF is a much more taxing position defensively than is LF. Therefore, you want a player that plays better defense in the position. Now, slide that highly-skilled defensive centerfielder to a corner outfield spot where they will have less ground to cover. You’re basically plugging CF defensive skills into LF, and defensive performance in LF should improve.

    Now, here’s hoping I got my italics right. Otherwise, I’m going to look like a fool.

    Vote -1 Vote +1

  16. Matt says:

    “…Of course if you replace that CF with a another CF who is significantly better with the glove…”

    I think that is where my problem with the argument comes in. It is more difficult to find a great defender who is good enough with the bat, than it is to find a guy that can hit. So you are losing runs in left field, although it may only be 5, and then you may lose runs in CF as well. So if the guy can play CF and hit it is more valuable as far as building a team long term. I see the value in having great defenders in the corners and certainly your argument is a good one, but positional adjustments are definately needed.

    I don’t know that VORP or other stats like it do it correctly (I certainly don’t understand the metrics well enough to know.) But there has to be some positional adjustments made.

    Vote -1 Vote +1

  17. Eric M. Van says:

    I’ve been critical of a lot of analysis posted on this site, so I have to chime in and opine how spot-on this piece is.

    It’s sort of paradoxical: a rule of thumb of good management is not to have guys playing positions they are overqualified for, because you lose value. The corollary, though, is not to be too slavish following the rule, because you don’t lose *a lot* of value. If there are other arguments in favor of the move — if you can regain the value at another position (because of a trade or because of payroll flexibility) or in the long run — go right ahead.

    The other very interesting corollary is that *the variance in defensive value lost by such a move is greater than the lost value.* IOW, there are players who are perfectly adequate at a given position who, when moved one notch easier on the defensive spectrum, gain more defensive value than usual, and more, in fact, than the offensive value lost. The classic example is Albert Pujols, who can almost certainly play an adequate 3B but whose defensive numbers are so high at 1B that it’s unlikely he would just lose the average amount of D value if moved to 3B. Another example appears to be Nick Swisher, who actually hasn’t been bad in CF (somewhat below average, but not inadequate) but has played a ton better in RF.

    If you think about it, *this has to be the case* or else there would never be any question as to what a guy’s best position would be (e.g., whether Dustin Pedroia and then Jed Lowrie could stay as SS). That it’s not always obvious where a guy should play pretty much proves Dave’s argument that the net loss in value as you move on the defensive spectrum is just a handful of runs per notch.

    OH, BTW, .8 runs per play is for singles. Fielding Bible Plus / Minus actually reports bases in addition to plays (it’s the difference between their “enhanced” and basic totals), and you can count those as .25 runs. For a system like PMR that just reports plays, you can count an OF play as closer to 1 run.

    Vote -1 Vote +1

  18. Blackadder says:

    Dave, if you weren’t right, then it would be a conceptual mistake to apply positional adjustments! The question, of course, ultimately comes down to how efficiently the resources of a team are allocated. All else being equal, it is worth switching a player to an easier position if and only if the relative improvement in his fielding more than offsets the higher nominal replacement level. There are some cases in which it is very clear: the gap in replacement level between 1B and SS is about, say, 25 runs. I would be willing to bet that Jason Giambi would be more than 25 runs worse at short than at first; conversely, it’s almost impossible to imagine Jimmy Rollins being able to make enough extra plays at first to make up the 25 run gap; there simply aren’t enough plays.

    These cases are of course easy. When teams actually move players, they seem, anecdotally, to be in cases where it is basically a push; Ichiro, for instance, seems to field just well enough in RF to precisely counter the change in replacement level from RF in CF. In these cases, the right thing to do is to move the player should it ever seem useful to do so.

    Vote -1 Vote +1

  19. philosofool says:

    Actually, I think the reason CFs perform better in LF or RF is simply due to the change in who they’re being compared to. The rankings are relative to the player’s peers, and when you’re getting compared to Ibanez, Dunn, and Ramirez instead of Sizemore, Gomez, and Beltran, you’re going to look better.

    This is a really interesting point. I did an extremely cursory comparison of performance by CFs that also played other positions in a given season using RZR (extremely cursory=clicked around for guys that played both in a given year and compared their performance at each position.) I didn’t actually see any evidence that OFs that play CF get more outs per chance in a corner.

    However, CFs as a group (within said cursory examination) perform better than players in a corner (median RZR about .025 higher than than RF and .035 higher than LF.) They also have a greater number of chances per inning than the corners.

    The increased value of CF defense, on this little study, seems to lie in it’s getting more action, and hence the greater value of reliability in each play. However, it’s hard to be sure that guys that usually play a corner and do center in a pinch would perform as well in center as they do in a corner.

    Vote -1 Vote +1

    • Barkey Walker says:

      I think this means that the 2-5 runs calculation is way too high. When you play in the corner, there is more kicking dandelions, less fielding. Hence the lower value. Given how small your result is, probably the naive view that is being criticized was actually a very good guess.

      Vote -1 Vote +1

  20. lookatthosetwins says:

    Ok so I’m a little late, but does anyone know how much the difference is between third and short? I’ve heard someone say that that Brendan Harris is an awful shortstop, but is fine if we play him at third. How does this make any sense? His lack of range will also show up at third, and his weak bat will be worse than most of his peers at that position. Obviously more balls are hit to short, but does that really account for someone going from awful to decent? Also, does A-Rod lose a lot of value by playing third? If he was playing short I would think every stathead would put him at the top of the MVP list every year…

    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>