So Let’s Talk About Alex Gordon
For most of the last few years, if you clicked on the Leaderboards tab here on FanGraphs, you’d find Mike Trout’s name at the very top. Today, that is not the case, as Trout has been surpassed in 2014-to-date WAR, slipping to #2 for the first time since late April. That isn’t necessarily controversial in and of itself, as it’s not that unusual for the best overall player in the game to not rate at the top of the WAR leaderboards every season, but what is somewhat controversial is the name of the player who has usurped Trout at the top of the list at this moment.
Alex Gordon, you see, is not exactly what most people think of as a superstar. He’s a corner outfielder who is hitting .286 with 13 home runs. Among 153 qualified Major League hitters this season, he’s ranked 36th in batting average, 32nd in on-base percentage, and 53rd in slugging percentage. Even using wOBA as a better evaluator of overall offensive performance, his .357 wOBA puts him in a tie for 33rd with Neil Walker and Jayson Werth. Add in park effects, and his wRC+ of 128 falls to 39th. As a hitter, he’s basically having the same season as Matt Kemp. This is the batting profile of the guy who currently leads all position players in WAR, and for many, that simply highlights the limitations of the model.
Even sabermetrically-inclined writers who live in Kanas City think this is weird.
Love Alex Gordon as a player. A legitimate star. The idea that he's the best player in baseball this year is absurd: http://t.co/Q6Vx9l2L00
— Jeff Passan (@JeffPassan) August 18, 2014
Passan, it should be noted, is arguing against a strawman, since I haven’t seen a single person argue that Alex Gordon is “the best player in baseball this year.” For one, even if you used WAR as the sole basis for determining “best player in baseball” — and you shouldn’t do that — then the answer would be Felix Hernandez (+6.2 WAR, a half-win ahead of Gordon), so the most aggressive argument you could make is that WAR has Gordon as the best position player so far.
But really, even that is a far too aggressive interpretation, since no one has ever rationally argued that WAR is precise to the decimal point. The reality is that WAR has always been best used for grouping players of similar levels of contribution, not for arguing that a 0.1 WAR difference means that Player X is having a better year than Player Y. No one actually argues for using WAR as a precise tool to measure minuscule differences. I’d suggest that what WAR is actually saying is that Alex Gordon, so far, is having one of the best seasons of any position player in baseball this year, and I don’t think that statement is at all absurd.
First, let’s start with just the less controversial offensive component, since we already went through his hitting numbers. We noted that by wRC+, Gordon ranks just 39th among MLB hitters this season, but then again, hitting isn’t the only way to produce offensive value. We know that players convert their number of times on base into runs at different rates, and that having Jarrod Dyson reach is more likely to lead to a run for the Royals than if Billy Butler reaches.
Gordon isn’t Dyson, but he is a very good baserunner, having taken 11 extra bases and only making two outs in the process this year. Once you combine baserunning value with hitting value, we find Gordon ranks 22nd in Offensive Runs Above Average this year, better than his raw hitting marks would suggest. He’s still more of a good offensive player than a great one, but simply looking at his BA/OBP/SLG marks will undersell his contributions to run scoring.
But, of course, that isn’t why Gordon ranks #1 in position player WAR at the moment. He ranks at the top because he’s #3 in MLB in Defensive Runs Above Average, coming in at +17 runs relative to a neutral defensive player. People are rightfully more skeptical of defensive metrics than they are of offensive metrics, and we absolutely have more uncertainty surrounding Gordon’s defensive performance this year than we do his offensive performance. But rather than saying that the defensive component rating Gordon as an elite player is absurd, we should instead ask what the magnitude of the measurement error might actually be, and how that should affect our view of his performance overall.
After all, Gordon isn’t exactly a defensive schlub. He moved to left field full time in 2011, and has won a Gold Glove in every season since. While there are all kinds of problems with the Gold Glove voting, Major League managers very quickly adapted to Gordon as a terrific defensive outfielder, so this isn’t just a trust-the-numbers-over-the-eyes situation. By the Fans Scouting Report, Gordon rated as an 82 last year, the fourth best mark of any player in the entire league. People who watch Alex Gordon play defense regularly think Alex Gordon is really good at defense, so we shouldn’t be too terribly surprised that Alex Gordon ranks as a very good defender.
But this is absolutely an outlier season for him in terms of UZR, which is the defensive component of WAR. Since moving to left field full time in 2011, here are Gordon’s UZR/150 numbers:
2011: +12
2012: +12
2013: +7
2014: +27
Over the last three years, Gordon has rated as a very good defensive left fielder, rating about 10 runs per season better than the average left fielder. This year, he’s pushing close to 30 runs better than the average left fielder, which is why he ranks #1 in WAR right now and he never has before. Alex Gordon is clearly not a true talent +27 defender in left field, and some skepticism about that number is entirely justified.
But again, let’s keep in mind that even a normal Alex Gordon defensive rating would still rank him as one of the best players in baseball this year. Since moving to left field full time in 2011, he’s ranked as +30 runs above an average defensive player, which includes the positional adjustment. Even if we put no weight on more recent data and simply use a straight average of the total, we’d expect him to have a defensive rating around +6 or +7 right now; instead, he’s at +17, meaning that the bump in defensive rating this year has given him credit for about 10 extra runs, or about one win.
Subtract a win off Gordon’s total, and instead of ranking #1 in seasonal WAR to date, he falls all the way to a tie for #9. That’s the magnitude of the difference. Using an overly regressed defensive assumption, Gordon is still a top 10 position player in Major League Baseball this year. If you were a bit more reasonable in your regression and weighted recent performance more heavily than past performance, you’d end up with a weighted average defensive value of closer to +9, and Gordon would again find himself in the top five among position players in Major League Baseball.
And here’s the thing; there is absolutely no reason to assume that defensive performance is more static than offensive performance. In fact, there are all kinds of reasons to believe the exact opposite, and to expect fluctuations in defensive performance of a greater degree than we find in offensive performance.
We all generally understand that performance variance decreases as sample sizes get larger, and not just on defense. Batting average over a full season is more credible than batting average over a month’s worth of games played. We don’t freak out when Josh Harrison leads the NL in wRC+ over the last 30 days, as he does now, as we know that stuff like that happens, even with metrics with very minimal measurement error.
And the reality is that one of the primary reasons why offensive statistics are more reliable is simply because the samples are larger. Over the course of a season, an everyday player will bat 600 to 700 times, allowing much of the small sample variance to wash out in the end. On the other hand, even a very good left fielder like Gordon averages about 300 putouts per year, and most of those are routine plays that any ambulatory Major Leaguer could have made, so they have no real effect on his defensive rating.
According to the Inside Edge data here on the site, 71% of the balls hit in Gordon’s direction this year have been “Routine”, meaning that they are converted into outs 90-100% of the time. Gordon has converted 99.6% of those plays, so even that range is likely too large. These are routine pop flies that basically everyone catches. In addition, another 42 plays have been labeled “Impossible”, meaning that no one ever catches balls hit at that location, angle, and velocity. That leaves just 53 plays this season ranging somewhere between “Remote” (1-10% chance of conversion) and “Likely” (60-90%), and those are the plays that determine 100% of defensive rating.
We’re really talking about evaluating a player based on his performance on something like 50 marginal plays throughout the course of the season. It would be ludicrous to expect performance over an N of 50 to be the same every single trial, especially when the result of the play made or not made has such a large swing in run value. Making a catch on one of these marginal plays in the outfield is often times the difference between saving an extra base hit or making an out, and the gap in run value between a double or a triple and an out is worth more than a full run. With just a few 50/50 balls going one way or another, a player’s defensive performance in runs saved can vary dramatically.
Think of defensive performance as similar to home run rate. If a player hits a ball one foot shy of the wall, he makes an out; if he hits one foot beyond the wall, he creates at least one and often times more than one run for his team. A very small difference in a player’s swing can have a very big difference in the outcome, and we’re not that surprised when we see things like Chris Davis‘ HR total going from 33 to 53 to 21 over the course of the last three years. When dealing with samples this size and high magnitudes of difference for an out versus a non-out, we should rationally expect fluctuations. Single year blips may be correct, or they may not be — we don’t know for sure — but deciding that a player’s defensive rating is incorrect because it fluctuates from his prior history is a very flawed way of thinking. The cliche is wrong; defense can slump just like anything.
There is absolutely an argument to be made that Gordon’s UZR may be incorrect — though interestingly, people only ever seem to assume that numbers are too extreme, ignoring the possibility that the measurement error could also mean that his defensive rating might be too low — and if you were trying to answer the question of who “the best player in baseball” is, you’d definitely want to use multi-season regressed defensive numbers. But even using those kinds of calculations, there’s no way to get Alex Gordon out of the top 5-10 position players in MLB this year. The only “absurd” argument would be that Gordon hasn’t been one of the best players in baseball this year.
The very best? WAR can’t tell you that. But the good news is that it’s not trying to. What WAR is trying to tell you, though, is that Alex Gordon is having a great season, and you should accept that conclusion even without putting as much faith in defensive metrics as you do in offensive ones.
Dave is the Managing Editor of FanGraphs.
FINALLY! R-E-S-P-E-C-T! Take that, Jazayerli!
“For one, even if you used WAR as the sole basis for determining “best player in baseball” — and you shouldn’t do that…”
But you can if you’re rating teams.
agreed, but there’s still a lot of unexplained variance between WAR and actual wins. There’s more in WAR to Ws to the pythagorean to Ws.
I think one of the big reasons why defensive stats get questioned so much is the relative lack of transparency.
We can easily see the result of every plate appearance, and the public has developed some sense of comfort in that. But can anyone say what Gordon’s UZR was on any of the plays he made in the field? Until we get there, the perception issue will remain.
Yes, a play log would make sense. Though it would quickly be disparaged by people pointing out specific cases where the credit/debit given didn’t match the eye-test, conveniently ignoring the fact that the same effectively happens on batted balls all the time.
I don’t actually buy the comparison? Yes, we can see the result of every plate appearance, but we can’t translate it into wOBA or wRC+. Yet people around here seem to be entirely comfortable comparing players by wRC+, as well they should be, despite the fact that the vast majority of us only have a good general sense of the process that produces it rather than a complete knowledge of the formula.
But all the components that go into wOBA are observed and recorded separately at the time of the event, whereas the components of UZR are not.
It would be pretty cool if you saw Alex Gordon make a great catch and it got scored right away as 10-30% catch or whatever and posted on the scoreboard. That might help people mentally keep track of defensive value a little better.
wOBA is pretty easy, though. Each possible outcome of a plate appearance has some value (an out is 0, a walk is about 0.7, a homer about 2.1, off the top of my head). You add up these values and divide by the total number of PAs, and that’s wOBA. Pretty transparent.
One thing about defense is that the eye test can get fooled pretty easily. A not very good defensive outfielder can make a spectacular looking play that a good defensive outfielder makes to look easy. It’s not always easy to see if an outfielder took the best route to the ball or how comparatively fast different players are. A player that wows by making really tough plays might be an average defender if he has repeated mental gaffs and misses easy plays.
It seems to me much less straightforward to assess than hitting.
You’re describing Yoenis Cespedes. 🙂
I see this all the time on Sports Center…they show an outfielder misplay a ball and have to make a tough catch while he’s awkwardly reaching across his body and falling backwards. Then they say “What an amazing catch! He’s a great defender!” It drives me crazy every time.
We have it for inside edge already – go to the spray charts on each player’s page. You can look up video and compare if you don’t trust the scouts. I think UZR is proprietary however, so you can’t see the run values assigned to each play by UZR. Here’s Gordon’s page: http://www.fangraphs.com/spraycharts.aspx?playerid=5209&position=3B/OF&type=fielding
Should you have trust issues with the way the run values are calculated, you could calculate your own using inside edge data (again, the BIS data that UZR is calculated from is proprietary http://www2.baseballinfosolutions.com) – I’m going to go out on a limb and say that they are going to be in the same ballpark.
When defense slumps though, isn’t it largely slumping due to opportunity, not talent?
What would make you think that?
He would think that because unlike your guaranteed 3 at bats a game, you are not guaranteed 3, 1, 100, whatever defensive chances a game. You are also not guaranteed whether those chances are routine plays or difficult plays which UZR likes to award additional points for.
This is one of the reasons UZR requires a bit of time to become somewhat stable.
But some guys get a lot more PAs than others because they are on teams that go through the order faster. We don’t care about that. Value is value. If the ball is not hit your way, you don’t add value. If it is, and you do well, you do add value. It’s the MVP, not the highest true talent player.
53 plays from “remote” to “likely” that drive value.
What if only 30 were hit in that zone?
Runs saved is a counting stat, yes?
It is a counting stat, but one that can have a positive or negative value, meaning that it’s not quite “slumping” that would be caused by a lack of opportunity, but, rather, “dampening,” that is, tending towards 0.
DRS per 150 is better at this, but (without having looked at the stats) I’d imagine that there’s some pretty significant variation in the total number of chances (more specifically, the total number of “remote” to “likely” chances) over 150 games.
What we’d really want would be this:
Every play gets a Raw %, representing the % of times a major leaguer at that position would make that play. Sum up the percentages, to get the “expected” number of plays that the fielder should have made, and compare to the total number of plays that fielder made. You could then scale that to the number of opportunities, giving you a defensive rate stat. That would account for variation in number of chances and distribution in difficulty among those chances. Of course, with the eye test or anything we have now, assigning a single percentage to any particular play would be a fool’s errand, meaning that the perfect defensive rate stat is, of course, unattainable at present. But it’s interesting in theory
@jruby while you would expect a dampening effect with runs saved when a player has fewer opportunities, I suspect you’d have a wider variance of UZR/150 in a smaller sample.
Agreed, Costanza. I was just addressing Marco’s “runs saved is a counting stat” comment.
When I said DRS/150 is “better at this,” I should have explained that I meant “eliminates this effect”
removing catchers and running a regression of plays against defense gives an alright correlation (though still pretty low R2), but i dont know well enough how “plays” is defined to have any sense as to whether players with great range really have a chance to get meaningfully more of them. not to mention whether there are other common cause possibilities…
http://imgur.com/zh6BMMM
Great article, Dave! I almost asked a question during Sunday afternoon’s Q&A that touched on this.
For some reason, people really seem to struggle with the idea that defensive performance metrics are just that, performance metrics. They seem to want to jump directly to interpreting it as talent and then freak out because there’s no way he’s a (+20) defense. People have no problem accepting that a guy can have an outlier offensive performance, especially over a relatively small sample, but they don’t pretend like the .750 OPS guy didn’t really put up that .900 OPS in the first half of the season. They say it’s unsustainable, but they don’t take it away from him.
It’s one thing to claim that UZR or DRS is subject to immense amounts of measurement error and that it’s likely that a +20 is really a bad measurement of a +10 season. But it’s quite another to take away 10 runs from a season-to-date performance analysis simply because you don’t think he’s going to sustain it.
For some reason, people really struggle with the idea that the defensive metrics we have don’t directly measure performance in the same way that the readily available offensive stats do.
They don’t. They are the results of data being run through a known-to-be flawed model. They are the best thing we have, and can measure true-talent fairly reliably over a large enough sample size, but the methodology is not there yet to claim that a single season of defensive data necessarily has anything to do with what happened on the baseball field.
To take an example, Fangraphs uses FIP for WAR rather that RA/9. I always assumed the justification for this was that they want to regress BABIP-driven (and sequencing-driven) performance. Why reward someone for something that’s not sustainable? If UZR values are not stable over a course of a season, shouldn’t we regress that somehow?
Terrific article, I love how you explain the difference between “best player in baseball” and “best season in baseball”. Regressing UZR may be helpful in the former, but we don’t do the same for HR’s or BABIP for offensive stata in the MVP discussion because MVP is about the best season.
However, we do knock RBI because they are an opportunity statistic. Should we do the same for UZR?
The difference being that they are an opportunity statistic in large part dependent on the quality of the hitters in front of you, which remains (relatively) static over the course of a season. Even accounting for relative divisional strengths, clubs face every other team in their league, and plenty of teams in the other league, meaning that defensive opportunities–obviously with outliers–will tend to even out in a way that RBI will not.
I love the way you wrote that.
Opportunistic? Yeah, there will be tons more chances when you have Jeremy Guthrie on your team.
The opportunities in fielding are not biased in the same way RBI opportunities are. However, there is still plenty of random variation in opportunity from one player to the next and one season to the next. You say it will “tend to even out”. I guess. But if you read about UZR you’ll see it evens out over a 3 year period. This is just a different way data can be skewed in favor of a particular player. It’s a valid criticism.
Without a doubt Gordon is having a great year and is an elite outfielder. But isn’t there a history of defensive metrics going nuts over left fielders with good range? (Carl Crawford in his Tampa heyday, Brett Gardner in 2010-11). It seems to me if you put Trout back in left field he would be way ahead of Gordon and we wouldn’t be having this discussion.
“It seems to me if you put Trout back in left field he would be way ahead of Gordon and we wouldn’t be having this discussion.”
If that’s the case doesn’t that represent a significant flaw in the positional adjustment system? If the same player would have a drastically different rating just by switching positions, two positions (LF/CF) that don’t require different skills really, that seems to be a red flag to me.
I feel like there might be something to this–maybe because a lot of genuinely bad outfielders are attempting to hide in left field? So the actual good ones get major credit? Would love to know if that’s backed up by the stats.
It wouldn’t surprise me, tbh. It seems like there may be a wide range of player types in LF. There are some guys like Nelson Cruz, Michael Morse, Jason Kubel and Mark Trumbo that are basically 1B/DH types that have to play LF because either 1B is taken alot of days or there is no DH in their league. Then there are guys like Heisey, Gardner, and Ackley that are pretty athletic and likely wouldn’t be too bad playing centerfield every day. So, IMO, that wouldn’t be too surprising at all.
Gardner’s got a career UZR/150 of 10.8 in CF, so yeah, he probably wouldn’t be too bad out there.
Yet Ellsbury and his horrendous arm (and 6.6 UZR/150) is their actual CFer. When MLB calls a stat shit I listen.
It’s a good thing Ells has got great instincts and is an 80 runner, then.
Nope. That’s what the positional adjustment is for. Because he plays LF, he has to be much MORE above average relative to his position to score as well as he does.
But like the above guy said (BenRevereDoesSteroids), being way above average against Nelson Cruz and Mark Trumbo are a hell of a lot easier. There aren’t any 1B players playing CF, like there are in LF.
Which — to repeat — is what that great big positional adjustment is for.
Agreed, but I guess my point in the ‘flaw in the system’ comment was that maybe the positional adjustment should give MORE credit for CF than it currently does. If Trout loses overall WAR by switching to a more difficult position, but doesn’t necessarily play worse (not sure if this is actually true, but he can’t be playing THAT much worse in CF than he was in LF), that seems to represent a flaw in a player valuation system. Am I missing something?
Maybe, maybe not. The positional adjustment is a value based upon the ability to play (and actually spending time) a specific position.
Depending upon a player’s skillset, he may be more or less valuable at his original position than his new position. Old pos. adj. + UZR/DRS at the old position will not necessarily equal New pos. adj. +UZR/DRS at that new position.
Let’s say both Jason Heyward and Jose Bautista move to CF.
With a 27.8 UZR/150 in RF, you may expect Heyward to be worth roughly 17.8 UZR/150 in CF, because the positional adjustment between the two is 10 runs. In actuality, he may even exceed that value because his range (+20.9 UZR to date) is good enough to mitigate the positional adjustment loss of 10 runs.
Bautista’s 3.3 UZR/150 would become -6.7 UZR/150 from positional adjustment alone, but he has below-average range for a RF (-2.2 UZR to date). In the end, this lack of range could significantly hurt him in relation to his peers, and he may end up with a significantly worse overall value.
The positional adjustment is a good starting place to see how valuable it is to be able to play a certain position, but you can’t expect it to be a perfect relationship across the board.
Trout played a lot of LF last year, and his rating was considerably lower than 2012, when he played CF almost exclusively.
Btw, anyone note that Trout is now ahead of Gordon again? And that he always has been at BBRef and at BBPro? Referring to more than one site helps to smooth out some of this variation.
“The reality is that WAR has always been best used for grouping players of similar levels of contribution, not for arguing that a 0.1 WAR difference means that Player X is having a better year than Player Y.”
Then why provide these nice sortable tables where WAR is listed to a single decimal point accuracy? Why not instead round these value into these “groups of similar production”?
But where do you draw the line of distinction? Rounding to the nearest whole number? Then the 4.5 and the 5.49 all round to 5, and there *is* a distinct difference in their performance.
I like it presented as is, and we can all divide the list into tiers as big or as granular as our own choosing.
For position players I’d round it up to the nearest half a point.
You draw the line where it is accurate. If it is 0.1, okay. If it is 0.5, okay. The problem is showing one thing for everyone to use then trying to say people should not be using it how it is shown.
I’ld be okay with that, but then the numbers would say someone who is worth 4.8 WAR (rounded to 5) is significantly better than someone worth 4.7 WAR (rounded to 4.5). Would that be okay with you? I think that is the give and take with rounding.
Or you just provide a 95% CI to the actual value.
Just wondering aloud – since folks seem to have more quibbles (rightly) with single-season defensive metrics, would the players who derive more value from their defense have wider confidence intervals? It seems like they should, no?
I don’t think so because all players are being evaluated on the single season UZR values. Whether or not they have a good value, the stat shouldn’t be any more or less reliable for an average defender as it is for an elite defender.
Just because the statistic is accurate to a particular amount doesn’t mean you want to round everything.
Let’s pretend there’s two guys, one 5.24 WAR and the other 5.76 WAR. Rounding it, you go to 5 and 6 WAR, respectively, and their +/- 0.5 meet at only one point, 5.5.
But, in reality, one guy has a sphere of WAR ranging from 4.74 WAR to 5.74 WAR while the other sits between 5.26 and 6.26 WAR. They’re actually far more comparable than what your rounding error would leave us to believe.
Rounding to the accuracy does not improve the accuracy. In fact, the opposite is true. The truth is that statistics are only as good as the people using them.
…probably if you compounded the error on the park effects, positional adjustments, defensive value and base running, we’d find that WAR is almost never useful for comparing between two players. …but we don’t know the error on any of those things, so it is all just speculation.
I agree. If Dave continues to insist in chat and articles that WAR isn’t precise to the first decimal place, why doesn’t he talk with other Dave to change how WAR is displayed in the leaderboards?
The same reason you have the top 3 hitters batting (making this up) .337, .335, and .333, and not “around .330 or .340”.
AVG is three decimals because that is how it has been for ~100 years (not necessarily a good reason). It is also a much simpler calculation than WAR which results in more significant digits (mathmatically speaking). I could see a good arguement to shorten AVG to two decimal places, even though it won’t happen.
Avergae has always been displayed as 3 decimals, but the way it is used, it actually is not shortened at all. It is to three decimals, because that has always been enough to break the ties. For instance, in 2003 Pujols won the batting title over Helton .359 to .358. Pujols was actually .35871 to Helton’s .35849. If they both would have been tied at .358 then they would look at the next decimal, and the decimal after that. The only way to tie would have equivalent fractions.
My point was that you typically display the data to the precision you have (within reason), not rounded to a point to remove noise.
It’s the same reason why the AL HR leaders have 32, 31, and 29 and not 30, 30, and 30.
Nobody claims that there’s a statistically significant difference between 32 and 31 HR, or someone getting 200 H in 600 AB vs 201 hits, but that’s what happened, so that’s what goes in the leaderboard.
If you were making projections, THEN you could make a strong argument that all batting averages should only go to 2 decimal places, HR totals should be multiples of 5, WAR should be divisible by .5, and so on.
Simply put: if you’re just giving a record of what already happened, you provide all the data you have. If you’re using the data to make a prediction, you need to know what’s significant and what’s not.
But the argument is that the error bars we have in our measurement of WAR are too large to know that a guy with 4.7 WAR wasn’t actually more valuable than a guy with 4.8 WAR, in the same way that we know a guy with a .334 average got a hit in a greater percentage of at bats than a guy with a .332 average.
However you round it, you don’t get away from that. You don’t know everyone who falls into the 5 WAR bucket is better than everyone who falls into the 4 WAR bucket, because the best estimate for some of the 5WAR guys will be in the 4.5-4.6 range, and some of the 4 guys will be at 4.4-4.5. Putting them in buckets makes it appear at first glance like there’s a real difference. It also makes it look like someone who is at 5.4 and someone at 4.6 are the same.
The fact that there are significant margins for error doesn’t mean that giving less accurate estimates is helpful. It just means we have to remember that there are error margins in the data. We could give confidence intervals, but so many people would misinterpret them that it may not actually help.
You’d go by halves, not full wins.
So a 4.3 WAR guy and a 4.7 WAR guy would both be ~4.5 WAR guys, and a 4.8 and a 5.2 would both be ~5 WAR guys.
Again, “The fact that there are significant margins for error doesn’t mean that giving less accurate estimates is helpful. It just means we have to remember that there are error margins in the data. We could give confidence intervals, but so many people would misinterpret them that it may not actually help.”
So, basically, Gordon has been roughly as helpful to his baseball team as Mike Trout has been to his, to this point in the season, but helpful in different ways, and more helpful to his baseball team than just about any other everyday player.
Well said!
I think the argument is that Gordon hasn’t been as valuable as Trout because he is getting too much credit for his defense.
How did you get there from here??
I think from the “even if you regressed those defensive stats” type of statements in Dave’s writeup.
(Which of course was not Dave saying that those YTD defensive numbers are invalid or heavily suspect, just that you can regress them if you don’t trust them and he’s still a super-elite, top 10 player.)
I’d say it’s more:
“So, basically, if you take the UZR numbers saying that Trout has been as bad a CF as Derek Jeter is an SS, and that Gordon is as good an LF as Simmons is an SS as wholly accurate, Gordon has been roughly as helpful to his baseball team as Mike Trout has been to his, to this point in the season, but helpful in different ways, and more helpful to his baseball team than just about any other everyday player.”
No, we assume he’s been roughly as helpful because we recognize that the “UZR numbers aren’t wholly accurate.
If UZR isn’t accurate, it shouldn’t be published.
WAR is a descriptive stat. It isn’t concerned with stability, but every stat is concerned with accuracy.
No.
The two are only even remotely close if you think that Gordon is a +30 LF and Trout is a -10 CF.
If you believe that Trout is a -5 CF and Alex Gordon is a +25 LF, there’s a full win gap.
I would love it Gordon and Trout finish 1-2 in end-of-season WAR totals, respectively, while Trout wins the Triple Crown as well as the MVP.
Triple Crown? Trout isn’t leading in any of the categories and isn’t even close on batting average (I don’t think his .291 avg will end up anywhere near Jose Altuve’s .339)
Woosh.
I tend to be believe Baseball Info Solutions’ Defensive Runs Saved metric more than UZR.
In this case, it tells a fairly stable story: 19 DRS in 2011, 24 DRS in 2012, 16 DRS in 2013 and 20 DRS year-to-date in 2014.
Good point. It is helpful to describe the differences between DRS and UZR in Gordon’s defensive rating. They essentially agree on the impact of his arm, worth on average about 8-9 runs per season (even this year when his assists are down, presumably because the word is out and runners are holding an awful lot).
UZR says that his range this year has been spectacular and totally average the previous 3 years. DRS says that it has been essentially flat, and consistently very good. It may be that DRS slightly overrates his range. If DRS is at +12 for range over the last 3 years and UZR is at +6, you could probably safely split the difference at +9 and get a fair rating. This would equate to +17-+18/150G rather than +27, as UZR has it for this season.
That would be roughly 4.7 WAR or roughly the 9th best position player in baseball. It should be noted that there has been no standout position player season this year. It may be that no one will get to 8 WAR, and usually there are one or two. There has been at least one position player who has reached 8 WAR every year from 2004 to 2013.
And really, Trout’s only behind because his own defense is in the negative this year by UZR, first time that’s ever happened.
This infers stability as a function of the fact that the differences between small numbers are small numbers. Yes, from 2011 to 2012 it only increased by 5, and 5 is a small number. But that’s 5 runs over a prior year’s 19, or an increase of over 25%. That is also a 5-run increase against an expected player’s average of 0. His 2012 DRS was 50% higher than his 2013 DRS. If someone’s total production in another counting stat dropped by 33%, you might not call it stable.
And that’s something to keep in mind: DRS is a counting stat. His LF innings totals for the four years in question are 1309, 1424.1, 1364.1, and 1042.2 (to date). On a per Inning basis (*1000 for purposes of comparison) that comes out to: 15.28, 16.85, 11.73, and 19.18 — while closer than the gross numbers, it remains that 5 runs over a season is no dust for sneezing, and it reinforces how much of a jump has been made this year.
Incidentally, his UZR ratings over that same span are 12.2, 14.6, 8.6, and 22.4, for UZR/150 scores of 11.6, 12.1, 7.3, and 27.2 — all of which is awfully similar to the DRS results, with the exception being that this year’s total to date looks is just a little more extreme. The numbers tend to agree more than they disagree. And in any case, why would year-to-year correlation imply greater credibility of measurement?
Great article. I have no problem saying that he has been the most valuable (purely in terms of runs produced/saved) position player thus far in 2014. The same way I have no problem saying the same thing about Erik Kratz for last night.
Passan should know much much better than to draw the conclusion that anyone is calling Gordon the “best player in baseball.”
Great article – I thought at first it was going to be another Dave-Cameron-hates-The-Royals post. I agree with tz too: defensive stats are subject to so many unquantifiable factors (positioning, imprecise batted-ball outcomes, etc.), that bias and preconceived notions of greatness can’t help but sneak their way in. All that said, Alex Gordon is a terrific outfielder period, but even a rabid KC fan couldn’t argue that Gordo is better than Mike Trout.
It’s almost like Cameron doesn’t hate the Royals, and you, the Royals fan, are the biased one.
Nah…that’s pure nonsense.
It’s been well established that DC hates all teams equally.
Oops, sorry – forgot that team fandom is a big no-no in the sabermetrics community.
lol nice strawman
It’s more that bitching that a writer hates your team is a big no-no among the adult community.
Go almost 30 years without seeing your favorite team enjoy any real success, then see if it doesn’t bring out some bitchiness in you too, “Cool” Lester.
I would have thought it would bring out a happy feeling; I am MUCH more prone to being bitchy if my team is 50-80 than if they’re 80-50.
Damn, first you bitch about the writer hating your team, then you miss a Wire reference?
Step your game up, brother.
Lovely. Congrats for sitting on your butt for a DVD marathon and not giving yourself a nickname because you think you’re cool.
Are there park effects included on defensive play scorings?
Base running and defense all suffer from the same issue – sample size. The article about Trout and base running a few days ago was great. It highlighted the fact that not all first to thirds are created equally. Depending on where the ball was hit, how many outs, was he already stealing, etc the results can vary wildly. And when you are talking about only a dozen or two scenarios, those unique events can swing the numbers dramatically.
There may be a little bias in me but I have a hard time swallowing that Trout is a scratch base runner and a below average center fielder. That doesn’t match up with my eye test or any reports that I’ve seen from scouts.
well, he’s not. the 3-year sample size says as much. But, given the numbers, evidence says that he has been an average baserunner and below average CF in this year in particular. Same way above average hitter Evan Longoria has been average at the plate this year. See the difference?
“…there is absolutely no reason to believe that defensive performance is more static than offensive performance.”
Great news for Nelson Cruz!
Delmon Young’s ears perked up! (Okay, let’s not get carried away here. We have to assume that Dave is talking about guys with some ability to track and catch a baseball.)
I guess my issue with all of this is bigger than the reliability of defensive statistics. Even if there were a way to say, with confidence, that Alex Gordon is the best defensive left fielder in baseball, and/or one of the best defensive players in baseball, I still don’t know how that makes him more valuable than the 21 players having a better offensive season than him (to say nothing about how, to simplify, him taking 9 extra bases this year bumped him up 17 spots. Different issue though).
I would love to see a discussion over the relative weight assigned to defensive value in the calculation of WAR, because at the end of the day, I just find it to be too much.
well, runs prevented= runs created. of a player reaches a certain level of defensive production he begins to equivocate his overall production with that of a superior offensive player. (think of it as the reverse Adam Dunn Effect. when Dunn was in the NL and having monster years, his glove was so bad he was barely overcoming the runs given up with the runs produced)
for example: Dunn hit .267/.398/.529 for the Nats in 2009 with 38 bombs ,(and 32.1 offensive runs above avg) yet his defense was so bad (-44.8 defensive runs above avg) that he ended the season with 0.9 WAR. Now that’s an extreme example, but Gordon (and Heyward in the NL) are using defense to increase their total value and impact on overall run production.
But that’s my point. Why have we come to the conclusion that Dunn’s defense almost completely offset his offensive production? How do we get to 45 runs below average for him? It’s very easy to measure, and compare, offensive production. I just have a hard time believing that any player could cost his team 45 runs in the field (and, to get back to the original point, that Gordon’s defensive ability has been worth 30 runs over the major league average).
Dan, the run values are calculated the same way that offensive run values are calculated, which is by the average change in run probability across all base/out states.
…that’s not how offensive run values are calculated. They use wRAA, not RE24, in calculating WAR.
Also, UZR is nowhere near as accurate a measure as wOBA-derived statistics.
Ok, take two hypothetical leadoff hitters. One makes a completely ridiculous HR-saving catch on the first batter in the top of the 1st and makes an out in his PA in the bottom of the 1st. The other does not make the catch, but HRs himself. Who was more valuable? They’re obviously almost exactly the same. So why are you ok with that measure of value- a run saved is about the same as a run created- in that one PA context, but not over 4.5 months?
Because tradition!
The problem is that what if that “completely ridiculous HR saving catch” is a play that 75% of outfielders make? Or that a better outfielder wouldn’t make that play because he was positioned in a different place than another outfielder? There are a lot of assumptions built into the defensive ratings which may or may not be true (skill vs. positioning, opportunities, etc.).
Is the home run on a ball 75% of position players crush? Or that a better position player wouldn’t homer on that ball because he was looking fastball and not changeup? Etc.
Since we don’t keep track of that for batters why would you consider it for fielders? Or is that your point – that offensive stats are simply luck because batters have no control over the pitches thrown to them?
My point was that the same argument can be applied for hitting. Yes, that play can be something 75% of fielders make, but what about pitches 75% of hitters crush? Yes, a better outfielder might not make that play because he is position in a different place, but a better hitter might not hit that homer because he is looking for the wrong ball (IE he looks fastball and doesn’t adjust properly). The same argument seems applicable either way. The initial argument is a saved HR = a hit HR, your argument is basically that saving the HR doesn’t necessarily indicate it is a spectacular play/one better outfielders might miss, if I understand it right. My reply was basically that without some kind of backing it can be applied to both sides of the ball.
You could also argue that we should keep track of this for hitters. I, for one, would be interested to see what players mostly just crush pitches everyone does and which do not, could even be useful.
Dan, you can find many such discussions at Tangotiger’s blog, http://www.tangotiger.com/index.php. Also his old one, “Inside the Book”, http://www.insidethebook.com.
Start by Googling those sites with the term “positional adjustments”. You’ll probably have to search around a bit, but you’ll be able to find what you’re looking for, especially on his older site.
Runs are runs. There’s no artificial weighting involved.
The issue is that the error bars for defensive runs are very high. The error bar for offensive runs from wOBA-derived stats is zero.
It’s stupid to weight them the same.
Not true. Different ballparks and line ups have different linear weights (and I’m not talking about regular park factors). They’re very closely associated with the mean or standard linear weights, but they aren’t the same.
Was this a reply to me?
It is indeed. The relative value of a walk to double is different in different contexts (not base-out states). It’s impractical to measure all the different values associated with different environments.
Yes, there definitely is error in run assignments estimated by wOBA. wOBA assigns league-wide average run values for batting outcomes and these averages have their own distributions. It is possible to walk every single plate appearance and never score a run (batting in front of Jose Molina for example….). What there is zero error in is the actual batting outcomes that go into wOBA. We don’t make mistakes recording what actually happened in the game (of course, even here there is room to quibble, e.g. Jeter’s “hit” to surpass Honus Wagner was turned into an error well after the game had ended).
Because KC’s OF defense is rating so high, has anyone looked at how they are being positioned. If they do a better job of positioning themselves – much like infield shifts can take away groundball hits, maybe this is one of the causes of the Royals very above average defense numbers?
Hi
Some teams are shifting more than others, true, but the type of shifting Gordon is doing has been around forever.
It’s possible some teams might be using advanced video watching / whatever to get an edge. However, I wouldn’t be in a hurry to assign any smarts advantage of any kind to the Royals. So, many years of totally sucking, more than a quarter century of it.
—
Also, there is a subjective element to defensive stats – how good is the stringer? Adrian Beltre’s inflated defensive stats in Los Angeles are one easy example of this. It’s obvious when one looks at the rest of his career in context. Could it be happening in KC, with all the fans finally feeling it?
Part of it is also the pitching staff. KC is an extreme flyball staff (I think they have the 5th or 6th lowest GB% – too lazy to look it up here!).
I wish I could find the link but someone looked at the top 5 defensive OF’s over a 5-10 year period and it seemed like flyball staffs tended to show up quite often – not enough to draw any definitive correlation though. UZR is not a context free stats – it is still a counting stat and an above average fielder will post more value on a heavy flyball staff (obviously).
I think it’s a combination of this (large # of opportunities) and excellent range for each of the three primary OFs compared to their peers.
Yeah, UZR notoriously does not take positioning into account.
But it’s totally as accurate a measure of performance as offensive stats, guys! I swear!
Quite the sarcastic strawman argument.
I don’t think any reasonable person is going to argue that defensive metrics are anywhere close to as accurate as offensive metrics.
Take a look around this thread, at all the people saying that we shouldn’t question the accuracy of Gordon or Trout’s UZR data, any more than we would regress their offensive stats when asking who has been the most valuable this year, because “it happened.”
It’s infuriating.
No one’s saying that. You’re overinterpreting .
Trout has been a poor defensive OF this season. So it may in fact be fair to say Gordon is a better allround player than Trout this season.
Have you seen Trout play this season?
Looking at his page, and specifically at his Inside Edge fielding data, I noticed something interesting. Of the those ~50 (53 so far this season) BIP referenced that make up the difference between defenders, the only “bucket” in which Gordon is playing above his normal level is “Likely” (60-90%).
He is actually BELOW (though only slightly) his career averages in the other three representative categories (Remote, Unlikely, Even). The difference this year is that he has converted 95.5% of his “likely” opportunities, compared to his 86% average in that category. And this year accounts for 22 of his 50 career opportunities.
As mentioned in the article, when your career sample size is 50 BIP in that category, wild swings in performance are unsurprising.
As a followup question, is there anywhere I can look for the major league average in each of those buckets? For instance, the “routine” category says 90-100%, so then would league average be 95% by definition? I seriously doubt it’s that easy because defensive plays are not created equal. However, Gordon may be above average in that category, meaning that he picks up significant value for converting 99.6% of those opportunities.
Here are the 2014 league statistics.
http://www.fangraphs.com/leaders.aspx?pos=all&stats=fld&lg=all&qual=0&type=3&season=2014&month=0&season1=2014&ind=0&team=0,ss&rost=0&age=0&filter=&players=0
And looking more specifically, OF in 2014 have a 99.4% rating in the 90-100% category.
This is a flaw in the presentation of the current Inside Edge fielding data. It would be much more effective to separate the Routine bucket into “Routine (90% to 99%)” and “Automatic (99% to 100%”, similar to the “Impossible” and “Remote” buckets.
Even though no one’s ever made an Impossible play, and there will certainly be “Automatic” catches that are dropped, the fact that OFs are converting 90%-100% at a 99.4% rate demonstrates that the vast majority of OF opportunities above 90% are also above 99%.
I am a full on believer in WAR and all of its components, but this all just feels a little desperate.
Tell us more, we are all enchanted by such a meaningful statement.
Why?
A backpedaling, ‘rationalizing’ article in leaky defense of the WAR stat. So many important newfangled stats on the scrap heap in the last few. I handicap xFIP to get there before WAR, but maybe not.
“Passan, it should be noted, is arguing against a strawman, since I haven’t seen a single person argue that Alex Gordon is “the best player in baseball this year.””
What a fucking load. People use WAR to make this case all the time. There’s no walking back from Trout vs. Cabrera, and the silly faux superiority behind all that. As if Dave is at all clean on this, either – he can write himself clean, but that’s not the story.
—
And then,’Okay maybe Jeff Passan might be a reasonable voice’ but beware everyone else…Bullshit. It’s not that complicated and this is all FUN- for the game.
The idea of taking away the stats understanding from just the commmon everyday fans, yielding true baseball standings to a guild over overweight supernerds is beyond pathetic. The original Bill James question of objective understanding has been turned around. Don’t think so, dear reader? Then how about you explain in any type of terms to the person sitting next to you at the game about Gordon vs. Trout, and then see how log the conversation lasts when you bring forth this nuanced and conditional representation of WAR? Even the samermetrics fans are being set up to get behind information they themselves don’t understand.
Geez, straw man much? What are you really angry about? Cabrera lover perhaps? If sabermetrics aren’t your cup of tea, maybe the ESPN message boards are more your speed?
I find it no coincidence that you are ranting about this when you are unable to complete a logical fallacy free paragraph. Perhaps you should just stick to batting avg and being smug at home , to yourself?
Dave Cameron shot JFK!!! wait, what was this guy’s point?
“…He ranks at the top because he’s #3 in MLB in Defensive Runs Above Average, coming in at +17 runs relative to a neutral defensive player.’
What’s that you say? Gordon actually ranks 4th in DRS?! JUAN LAGARES ranks ahead of everyone else in MLB and in fewer innings on the field!? That can’t be!
http://www.fangraphs.com/leaders.aspx?pos=all&stats=bat&lg=all&qual=200&type=8&season=2014&month=0&season1=2014&ind=0&team=0&rost=0&age=0&filter=&players=0&sort=20,d
sorry, Def runs above avg to be precise.
I love how the stats/projections are quoted as fact when it supports the argument the writer is trying to make and then written off as limitations of the model or random variance anytime the numbers suggest something else.
Are you suggesting that sports writers push narratives?
I think you read something different from the rest of us.
No, it’s just that he’s read other things that David Cameron has written, and this.
In the article, Dave talks about how UZR is flawed, illustrating a point that Alex Gordon shouldn’t be higher on this list in WAR than Mike Trout. However, it seems like every other time WAR is used, the number provided is taken for gospel.
I think this is what has upset some of the naysayers here.
“illustrating a point that Alex Gordon shouldn’t be higher on this list in WAR than Mike Trout”
You made this up.
“every other time WAR is used, the number provided is taken for gospel.”
You made this up.
Arc beat me to it. In an effort to find fault, we read in things that aren’t there sometimes.
I just don’t see what legitimate point you’re trying to make. That is, I think I see the point you’re hinting at making, but it’s not legit.
In 2012 Trout was 3 WAR better than Cabrera.
In 2013 Trout was 3 WAR better than Cabrera.
That’s a significant total. It’s the difference between Trout and Chase Headley/Jimmy Rollins/ Melky Cabrera this season.
In those debates people were making the point, you can’t go strictly on WAR, but the gap is SO big between the two.
This article is putting a .1 WAR difference in context. 2012 and 2013 were about 3 full WAR. (That’s more additional value than an average player provides.)
Indeed (and the point I made to Passan on twitter (which he ignored)). The only thing “absurd” was Passan’s assertion that 0.2 WAR difference is an absolute. We pretty much agree that if you have 3 players worth 5.1, 5.3 and 5.5 WAR that there is probably not much difference between them. They are all 5+ WAR guys who provided darn near equal value. But people are so used to judging batting average and looking for “THE BEST!” that they struggle with these concepts. They think .328 “beats” .327 batting avg so 5.7 “beats” 5.6 WAR. when, most of us saber nerds know they are pretty much identical.
I thought he was talking about Dave’s recent Orioles post.
All these newfangled stats might not be absolutely perfect so I’m going to stick with RBIs and grit.
Saves, baby
Finally! Defense gets it’s due.
Passan, it should be noted, is arguing against a strawman, since I haven’t seen a single person argue that Alex Gordon is “the best player in baseball this year.”
If you had the misfortune to watch the Royals on TV, you would know that Ryan Lefevbre, Rex Hudler, and Steve Physioc make that point on a nightly basis. Jeff Passan, who follows the Royals more closely than many writers, would be privy to this. No strawmen here.
Do those guys actually use WAR to support the claim that Gordon is the best player in baseball?
“What WAR is trying to tell you, though, is that Alex Gordon is having a great season”
Thank God for WAR, because I couldn’t tell that Gordon was having a great season by watching him play baseball.
Thank God for WAR, and Wynona’s big brown beaver.
oooooooooooooof SALESMEN!
This is what a baseball website has fallen to?
This article mixes and matches Inside Egde data and UZR.
Inside Edge only (kind of) covers the range component of UZR and ignores the other two – errR and armR. This feeds the central premise of the argument (conveniently) – these are tiny samples and while it may not be a real talent measurement or sustainable value it can be real because the sample is so small and we could be only talking about an extra couple of plays.
The issue here is much more likely to be simple noise – maybe not all of it, but a significant part of it. Yes defensive stats can fluctuate just like any other, but sadly UZR is a complete black box and we have no idea what the random variation is or the potential error. Nor am I aware of any actual study of this; which is a bit odd for a stat based community.
Gordon’s 1st 4600 innings had a rngR of 3.1
Gordons 1000 innings this year has a rngR of 12
When you account for innings that is over a 16X improvement… sure that could be “real” and it could be Gordon just having a career defensive year, but I highly doubt it is simply that. Without an actual understanding/study of the variation in UZR it is hard to distinguish how much of that might be real and how much of it could be just noise in a still pretty crude defensive model.
I’m pretty sure the issue is that if anyone did a study of UZR’s potential error, no one would use it anymore.
Alex Gordon plays in the best defensive outfield in baseball. Think of an outfielder as a circle; inside the circle the play is made, outside it the play is not made. Jarrod Dyson and Lorenzo Cain are huge circles. Gordon is a pretty big circle himself, but every time he takes the field with his fellow circular giants he can shade a bit here for this guy and there for that guy, knowing that his back is covered.
Defensive ratings seem to individualize the synergistic effects of team defense, at least where range factor is concerned. To isolate Alex Gordon’s “true” defense you’d have to account for the quality of his outfield associates.
Bwahahaha, all you fanboys with sticky Mike Trout posters in the bathroom. He’s not the best player in baseball anymore? Don’t cry, mama will make you porridge.
who puts posters in the bathroom and how did they get sticky?
can you please explain?
I’m just glad Alex Gordon is being talked about: 6.6, 5.6, 3.5 and 5.7 (Last ongoing) WAR the past 4 years means he has been an elite player, but he doesn’t really do it in a way people remember. But he is actually 6th in WAR from 2011 on (Though basically tied with Pedroia). That’s pretty good.
This article includes one of the more sobering and meaningful interpretation of WAR that i’ve come across in some time. Well done, Dave. Appreciate the tone, analysis, and conclusions drawn.
Passan often argues with a strawman. It’s the only thing that usually has the patience to listen to his drivel.
I could be badly misinterpreting this, but would it be a reasonable assessment to say that generally defensive metrics over a small sample are good descriptive measurements but are not particularly predictive? I.e. Alex Gordon isn’t necessarily the best player in baseball going forward, but he has been the most valuable/best so far this year.
Here’s a fun game. Lets see how many people counter any argument they disagree on with “nice strawman”.
As best I can tell, the only reason people (Cameron included) think that WAR is accurate to the nearest win is because of the scale it is presented to us on. We all know what a win is, tenths of wins seem trivial at best, and hundredths of wins are just absurd. However, we actually have no idea what a significant difference in WAR between two players is because we have no clue how much error there is.
WAR has lots of well know problems. These include positional adjustments that have little justification, park factors that describe an average across players, but likely do not describe any individual player very well, relative batting outcome weightings that are estimated league wide, but do not pertain to individual players very well, massive subjective error in baserunning and defensive measurements, etc.
Given all these problems and the fact that the error for the final statistic is compounded by the error in each of these elements, I’d be shocked if WAR is actually reliable to the nearest win. But since we don’t actually know one way or the other, we should ignore it completely when trying to evaluate the relative contributions of two players. Basically, if one player is vastly greater than another, then you don’t need WAR to tell you, but if they are similar enough that there is some doubt, we can’t trust that WAR is meaningful at that scale to be informative. Therefore, at best, it is worse than useless, it is actually misleading.
One sentence translation: “I prefer to ignore the best models we have in lieu of worse models because the best models we have still have model error and sometimes the lesser models work just as well.”
That is not a translation, it is a projection.
By all means then, explain why and how your approach is better rather than repeat the flaws that have already been established.
I did explain it in the last paragraph.
You really didn’t. You said “Basically, if one player is vastly greater than another, then you don’t need WAR to tell you.”
I’m asking how and why you know.
You also said “Therefore, at best, it is worse than useless, it is actually misleading.”
The implication being literally anything of statistical value is better. I’m asking how so?
When the gap between player abilities is large enough, you don’t need any statistics to tell you so. In the days before meaningful statistics were done, the Babe Ruth’s were never mistaken for scrubs, and vice versa. When the gap is huge, the problem is easy. We don’t need WAR for that.
However, when the gap in ability is small, and the problem is difficult, we should never use WAR because the error is almost certainly greater than the differences we are trying to measure. Even if you think the error in WAR is smaller than the differences you are trying to measure, you still should not use it because you don’t actually know what the error is.
WAR is a stat that makes people feel good, because people like order. However the order is probably largely random for the only parts of the scale that are actually interesting. It is bubblegum. You pop it in your mouth and it tastes good for a few seconds, but then you are just left chewing on a useless block of rubber.
Well, there’s a pretty big gap between Babe Ruth and replacement scrub – like every other player in MLB history. But, ok, I assumed you rely on lesser statistics and really you prefer to ignore them all it seems. Reasonable people can disagree, and certainly you can enjoy baseball without WAR. But people who work in baseball cannot, and most of us want to understand what those people are thinking.
No, I do not ignore all statistics, and I certainly never implied any such thing. All I said was that WAR has no actual use because the error is probably greater than the differences you are trying to measure (think of trying to do carpentry when your only measuring device is a yard stick with no further degradations).
Maybe too much reliance on a single site? Here are the top 10 in WAR by average of FG, BBRef and BBPro:
Stanton 6.39
Trout 6.22
Seager 5.37
Donaldson 5.28
Heyward 5.13
Gordon 5.11
Lucroy 5.09
Cutch 5.06
Tulo 5.06
Cano 5.01
144 comments and nobody has raised also the replacement value issue in WAR:
Alex Gordon defense value is bumped also because he plays LF and is compared to guys like Mickael Morse or Domonic Brown
Mike Trout plays CF and is compared to other CF that are pretty much all elite defenders.
But Trout gets more positional runs. Do you really think he would be rated higher if he played LF? Didn’t seem to help much last year.
This right here.
Yup. We should be looking at why UZR thinks that Trout is as bad at CF as Jeter is at SS (since we know that’s incorrect), not complaining about positional adjustments.
Any time you have a peer-based system, you’re going to run into these issues.
The biggest issue for Trout is that players like Lagares, Dyson, Hamilton, JBJ, et al. are wrecking shop defensively in CF and out-of-position corner outfielders aren’t going to rank well in relation to them.
Previously-good CF aren’t ranking well in relation, either. There’s a huge influx of defensive talent at that position this season, and that is the biggest factor in why Trout’s value has dropped so much. And it should have because, relative to his peers, his defense has not been as valuable as it was in the past.
-C
Or you could look at the Inside Edge data we have available, see that he’s far above the average for a center fielder, even with this influx of defensive talent and realize that there’s something funky with his UZR rating.
By definition, not all CF can be elite CFs.
He wrote “elite defender” not elite center fielder.
WAR already accounts for the position difference.
Yes, the positional adjustment that WAR uses considers every center fielder an elite defender by placing them at the top of the defensive spectrum. That was the basis for his statement.
I guess you could argue that the current gap in defensive value between LFs and CFs is greater than 10 runs, because we have an exceptionally strong group of defensive CFs and an exceptionally weak group of defensive LFs (excluding Gordon)…
This is what I’ve been trying to get at in other comments. I’d venture to guess that if you put Gordon and Trout side by side in the OF this year (either LF or CF – since you are basically doing the same thing that requires the same set of skills at either position), that Trout would get to many more balls and be a more valuable defender. However, Trout gets dinged value on WAR because the positional adjustment apparently doesn’t account for this enough.
One thing that needs to be said. Statistics are based on past performance. They have no bearing on what the player can/will do in the future. They are a tool to be used, but only a tool. Weather, health, other players performances, officiating, luck, coaching etc will all impact and greatly influence the outcome. If not, let’s just use stats or EA Sports to determine the outcome of seasons or championships and save everyone a ton of time.
And then there are statistical models, such as UZR, which are different than statistics and cannot be relied on as an accurate measure of past performance.
Would be curious to know more about the particular instances during which Gordon made an “impossible” play. My understanding of defensive metrics isn’t so hot, but could it be that an outfielder’s willingness to lay out or dive for a ball might be effected by the score of the game or runners on base? Blanco’s play during Cain’s perfect game comes to mind seeing as how it was a blowout.