The Hidden Moves of the Offseason

The word “move” is used in the context of an offseason to denote any number of varying transaction types. A trade is a move. A free-agent signing is a move. A player being designated for assignment is a move, or claimed off waivers, or sold to Japan. Players coming and going from rosters are the moves of the winter, and they’re the means by which the public tends to evaluate a team’s offseason.

The calculus for the outlook of the upcoming season is constantly changing throughout the offseason as these myriad moves transpire. When a team signs a star free-agent pitcher, we know that that team is several wins better than they were the day before. When a rebuilding club trades away its slugger in the final year of his contract for prospects, we understand that they’ve dropped a couple wins for the upcoming season.

But there’s another sort of move that happens during the offseason that’s more subtle, and it, too, changes the calculus of the upcoming season, though it often seems to be overlooked. We spend so much time and effort analyzing who “won or lost” the offseason that it’s easy to forget how much change should be expected from a team’s returning players. The Rangers didn’t go out and sign Yu Darvish this offseason, but he is expected to be a valuable addition to this year’s roster, an extra four or so wins added without any kind of traditional offseason move. Without doing anything, the Rangers rotation looks significantly better than it did at the end of last year.

Six years ago, Dave Cameron wrote a short post on this site titled 2009 Is Not a Constant. I recommend you read it, and sub in “2015” for “2009” when applicable, but here’s a relevant passage anyway:

We all know about career years and how you have to expect regression after a player does something way outside the ordinary, but regression doesn’t just serve to bring players back to earth after a big year.

Regression “fixes” a lot of problem spots from the prior year, even if the team doesn’t make a serious effort to change out players. The Royals got a .253 wOBA out of their shortstops a year ago. I don’t care how bad you think Yuniesky Betancourt is, you have to expect that number to be higher this year. They didn’t do anything to improve their shortstop position this winter, but the level of production they got from the position in 2009 is not their expected level of production for 2010.

You cannot just look at a team’s prior year won loss record – or even their pythagorean record – make some adjustments for the off-season transactions, and presume that’s a good enough estimator of true talent for the 2010 team.

By now, you’ve got an opinion of most every team’s offseason. You know the Tigers have signed two players to nine-figure contracts, and you’ve adjusted your perception of next year’s Tigers accordingly. You know the Braves have traded away anyone with a major-league uniform left over from 2015, and you’ve adjusted your perception of next year’s Braves accordingly. What you’ve probably spent less time doing is factoring in how much is expected to change at the positions where teams stood pat.

So what I’ve done is taken projected WAR, by Steamer, for every player who’s expected to receive regular playing time this year. For position players, I called it 300 plate appearances, and for pitchers, 100 innings. For all those same players, I calculated the difference between their projected 2016 WAR, and last year’s actual WAR, and I only looked at players who remained on the same team. The methodology might not be perfect here — it’s a summed total, and some teams have eight regulars returning while others have 13 — but this post isn’t about the exact figures, or determining a precise order. This is just about finding teams who can likely expect an overall bounceback from their core group of returning players, or perhaps an overall regression. It’s about reminding ourselves not to forget the hidden part of the offseason’s calculus.

This is something worth thinking about for every team, but we’ll go over a few of the extreme examples below. The chart looks like this:

Moves

Largest expected gains from returning players

Washington Nationals

Despite a projected three-win regression from Bryce Harper, because even Bryce Harper shouldn’t be expected to repeat a 10-win season, the Nationals top the list. Which is fitting, because the Nationals were last season’s biggest disappointment, and yet the Nationals are again projected as the best team in the NL East, by a five-game margin. This is the big reason why.

Rendon shouldn’t be expected to miss half the season due to injuries, and assuming full health, he should also be better when he plays. Before the Nationals even made a move this offseason, they could factor in an extra ~3 WAR, just due to a clean slate from Rendon, who looked like an emerging superstar at 25 just a year ago. Strasburg pitched like Clayton Kershaw in the second half, and shouldn’t be expected to have an ERA over five through May again. Extra two wins there. A healthier season from Werth, a better BABIP and more normalized strikeout rate from Ramos and fewer expected home runs from Roark could mix in up to four more wins from the Nats returning regulars.

Boston Red Sox

The Red Sox are basically the American League’s version of the Nationals, what with their 92-win team projection that might appear bullish on the surface relative to their colossal 2015 disappointment. All of which is thanks to the players added during last year’s offseason.

Our opinions of Ramirez and Sandoval should absolutely have changed from where they were when the Red Sox acquired them a year ago. But I don’t think anybody actually believes Ramirez and Sandoval are two of the three worst players in baseball, as they were last season. If you think Ramirez and Sandoval turn in two-win seasons — essentially assuming they’re just league-average starters — you get a nearly eight-win upgrade from last season. A bounceback from Porcello — who underperformed his peripherals at a level that’s jarring, even for him — could add another couple wins.

The Red Sox made headlines by adding David Price and Craig Kimbrel this offseason without having to move any of their major league talent, but the bigger part of their projected improvement is due to the expected upgrades from within.

Largest expected dropoffs from returning players

Toronto Blue Jays

It’s not that Donaldson isn’t expected to continue being a star player. He’s still seen as a top-three position player in the game. It’s just that, like Harper, you can’t expect another MVP season, in January. You can’t expect +9 WAR. It’s more reasonable to expect something closer to +6 WAR, which is still amazing. Then, you’ve got Estrada, who can be expected to regress somewhat from a career year, though I do find Steamer’s projection of him to be overly pessimistic, personally. Encarnacion is 33, and power declines eventually. Steamer doesn’t buy Goins’ sudden spike in walk rate, or Pillar as a true-talent +15 center fielder. Again, that’s not to say Pillar isn’t projected as a plus defender in center — he is — we just can’t reasonably expect that kind of extreme defensive value after just one season as a regular.

San Francisco Giants

Steamer still thinks the Giants have a great infield, just maybe not the best infield in baseball again. Even if you think this looks a bit overly pessimistic, it’s easy to at least see where it’s coming from. Duffy came out of nowhere — an 18th-round draft pick who skipped Triple-A and didn’t do anything in his brief 2014 debut — and turned in a star performance last year. That Duffy is even projected as an above-average starter (+3 WAR) is remarkable, given where he was a year ago, but he isn’t seen as a star just yet. Panik was seen as a low-ceiling player who had never hit in the minors, and turned in a star performance last year. That he’s even projected as an above-average starter (+3 WAR) is remarkable, given where he was a year ago, but he isn’t seen as a star just yet.

As for Crawford, his projection seems particularly bearish. While last year’s power spike was unprecedented and likely unsustainable, he’s significantly outhit his projected wRC+ each of the last two seasons, and the defensive metrics never matched the eye test until this season, which are perhaps unfairly muting his defensive projection. Like Duffy and Panik, Crawford probably isn’t the star-level player he was last season, so you probably wouldn’t be wrong to confidently take the over on 2.2 WAR, but some regression should be expected.

* * *

It’s important not to get too hung up on the particulars of the figures. That’s not the point. The point is that what Dave wrote six years ago is just as true now as it was then. It’s not as simple as starting from a team’s 2015 record, factoring in the major offseason moves and calling it a day. You can’t look at Boston’s 78-win record from last year, and scoff at the notion that Price and Kimbrel will add 14 wins to that. They won’t. But there are reasons to believe the Red Sox will improve substantially in other areas, in areas where they didn’t need to make a move. The expectation shouldn’t be that the entire Nationals roster gets hurt or underperforms in May again. Josh Donaldson won’t be an MVP every season, and Matt Duffy won’t always be an All-Star. Most every team has examples of this. Not to say these things can’t happen again, but the surprises of last year were surprises for a reason. Remember the hidden moves of the offseason.

Note: While Brandon Crawford’s figures were reflected in the original calculation and chart, I accidentally passed him over in the write-up. The post has been edited to correct this.



Print This Post



August used to cover the Indians for MLB and ohio.com, but now he's here and thinks writing these in the third person is weird. So you can reach me on Twitter @AugustFG_ or e-mail at august.fagerstrom@fangraphs.com.


Sort by:   newest | oldest | most voted
jdbolick
Member
Member
4 months 9 days ago

This piece is very well done, yet I feel like it needs a caveat regarding Steamer’s optimistic regression. Those Rendon, Ramirez, Sandoval, and Porcello projections all appear to be on the high side. While those players should contribute more in 2016 than they did in 2015, and therefore are positive “moves” in the sense employed by this column, I doubt they will be as dramatic as the projections suggest.

TKDC
Member
Member
TKDC
4 months 9 days ago

I buy Rendon and Porcello, but Ramirez and Sandoval contributing 7.6 WAR is a hell of a pipe dream. Steamer doesn’t know that Ramirez won’t even try to be a good defensive first baseman and Sandoval won’t think to decline seconds at the Golden Corral, and will pay $1.99 for his “take home box.”

redsoxu571
Member
redsoxu571
4 months 9 days ago

To follow up on what August Fagerstrom said, both players were valued at significantly NEGATIVE WAR last seasons, so most of the plus value comes from just returning to being more than a net 0 player for each. And that’s not especially optimistic.

TKDC
Member
Member
TKDC
4 months 9 days ago

Yep, I missed that. Thanks.

Tom Cranker
Member
Tom Cranker
4 months 9 days ago

What kind of caveat? The article is basically just presenting what the projections are indicating. Is the author supposed to cherry pick which projections he doesn’t believe will end up being accurate? He could do that, but that’d be a different article. Are you saying the projections systematically are projecting too much positive regression? Again, that’d be fine to investigate, but different article.

Surrealistic Pillow
Member
Surrealistic Pillow
4 months 9 days ago

Caveat: jdbolick doubts the positive regression of certain of these players will be so dramatic

jdbolick
Member
Member
4 months 9 days ago

I’m saying that the value of regression as presented in the column (Steamer’s fault, not Mr. Fagerstrom’s) is slightly to significantly exaggerated. Steamer, like most projection systems, struggles to deal with players who dramatically overperform or underperform previously established levels, so they will tend to overstate the value lost or gained from subsequent regression. While the column’s point regarding overall value from players doing better or worse than the previous season is well taken, using the extremes probably will not yield realistic representations of that value.

Rendon’s projection looks a lot like his 2014 career season with a little less playing time despite that season appearing to be the outlier. Then as TKDC noted, either Steamer is still calculating Ramirez’s defensive value at third base or else it genuinely believes he will be one of the very best defensive first basemen in the league. Sandoval is projected for his best ISO since 2012 along with an extraordinary rebound in defensive value. Porcello is projected to post the second best ERA of his career, etc.

Nathaniel Dawson
Member
Nathaniel Dawson
4 months 9 days ago

Or perhaps some people overestimate the how much one good or bad season informs us of a players actual talent level?

Jetsy Extrano
Member
Jetsy Extrano
4 months 9 days ago

If Steamer does this routinely, you can pretty easily patch on your estimates and have an improved system, can’t you?

If that’s not easy, maybe Steamer is making a decent midrange guess after all.

jdbolick
Member
Member
4 months 9 days ago

It’s easy to do by hand, yes. It’s much more difficult without manual adjustment.

jdbolick
Member
Member
4 months 9 days ago

Or perhaps some people overestimate the how much one good or bad season informs us of a players actual talent level?

Systems like Steamer are set up to spit out the numbers based on previous performance without any human input. The advantage of that is to eliminate some degree of bias, but the downside is that you don’t adequately account for all factors. Steamer makes the assumption that Hanley Ramirez should be a very good first baseman based on being adequate at third and shortstop before that. People who actually watched Ramirez know that he’s extremely unlikely to be very good at first base.

mrmaddness
Member
mrmaddness
4 months 9 days ago

Are the Mets a victim of recency bias? I know that Steamer looks back a while, but it seems like it woefully under-projects a lot of their offensive players (Conforto most notably), but also a guy like Lucas Duda who has 3 win seasons each of the last 2 years, but in 2013 was still fighting with Ike Davis for playing time. Also, he played 20+ games in LF (would love to forget that), which really damages his defense.

Brian Reinhart
Member
Member
4 months 9 days ago

“You can’t expect +9 WAR.”

As if we needed more evidence that Mike Trout is not like anybody else in baseball…his steamer projection is 8.9 WAR.

MustBunique
Member
Member
4 months 9 days ago

You’ve taken something that I think most people qualitatively knew and put some numbers to it. Thanks for that. Great read.

redsoxu571
Member
redsoxu571
4 months 9 days ago

I’m not sure that that is totally true (even more so in the NFL, in fact). While I think the occasional team gets highlighted for this (think how a few wise minds were ready for a rebound for the 2013 Red Sox), for the most part I agree that people treat returning players as too much of a repeat constant.

I also think that people are a bit more aware of bad-luck players who likely will perform better, but have a HUGE blind spot for returning player regression.

LHPSU
Member
LHPSU
4 months 9 days ago

Indeed, Betancourt did crush the ball to the tune of a .302 wOBA in 2010.

Twitchy
Member
Twitchy
4 months 9 days ago

I think the Jays regression will be countered by things like having Stroman back for a full season (don’t know if this was included as you didn’t mention it). I think Steamer is wrong about Pillar’s defence & base running, as well as a significant EE decline. The Donaldson drop off seems particularly harsh. He had a 7.6 WAR and then 6.5 WAR season before the MVP breakout, and now sub 6? As you said though, it’s not about the individual players, but the fact that teams are gaining or losing certain players or performances.

I think ZIPS is much more accurate but some of these projected declines seem particularly harsh. If anything I thought Colabello would be one of the leading regression candidates for the Jays, not the guys listed here.

redsoxu571
Member
redsoxu571
4 months 9 days ago

If anyone takes any umbrage regarding the Red Sox outlook here, it’s worth pointing out that the projections for them include MAJOR steps back from Bogaerts, Swihart, and Bradley (plus a tiny step back for Betts), despite the fact that all four players took a HUGE step forward from pre-July or so to post-July. Considering their ages and pedigree, I think it’s pretty safe to say that the foursome should net at least equal production to their 2015 selves, which would be more than enough to make up for Ramirez and Sandoval if they do indeed fail to even be +2 WAR players.

jdbolick
Member
Member
4 months 9 days ago

Those “MAJOR steps back” are projected to be 0.6 WAR, 0.2 WAR, 0.5 WAR, and 0.0 WAR respectively. Hanley Ramirez by himself is projected for a gain of three times that combined amount. Pablo Sandoval’s projected gain is almost as large, while Porcello’s adds another 1.2.

chazzycat
Member
Member
chazzycat
4 months 9 days ago

Regarding the Giants, I think you forgot to mention Brandon Crawford who has the biggest projected drop of anyone (-2.5 WAR).

However, given that almost the entire roster is projected to decline, and they still are projected to make the playoffs (as wild card) I’m liking their chances. Sure some will decline, but I don’t think ALL of them will like the projections say.

troybruno
Member
Member
troybruno
4 months 8 days ago

“but I don’t think ALL of them will like the projections say”

I can guarantee that each of them will absolutely not decline like the projection say… But I’ll also bet you that the aggregate change is closer to Steamer than your forecast! :)

vivalajeter
Member
vivalajeter
4 months 9 days ago

Is there a link where we can see the data for individual teams? Or did you have to manually go team by team to put that chart together?

Noah Baron
Member
Noah Baron
4 months 9 days ago

Look, I really do like projections. I do. They’re very useful tools at APPROXIMATING team strength, give or take 10 wins. And that’s something that needs to be mentioned a lot more, because while the individual player projections are usually (somewhat) accurate, the standard deviation on the team projections is downright massive.

But there are so many problems with using them for projecting team strength that FanGraphs writers never seem to mention. I’ll list a few of them below:

1) They do not understand platoons. Plenty of teams employ platoons at multiple positions, a move that (quite obviously) improves the production of both players by only allowing them to face opposite-handed pitchers. The projections do not attempt to account for this in any way. I don’t necessarily blame them for not doing this, because projections are meant for individual players, not teams, but it’s important to realize this when looking at team projections.

Examples: Utley/Kike Hernandez platoon, Lagares/de Aza platoon, etc.

2) They have a limited understanding of how to project minor league players (and young players not far removed from minors). Again, I don’t fault the projections too much for this, because projecting minor league players is difficult. However, when a minor league player reaches the majors and does better than the projections thought, the projections are notoriously slow at updating their understanding of the player accordingly. Indeed, the projections are notoriously conservative with projecting rookies and young players, a fact that hurts younger teams (and benefits older teams with more established players).

Examples: Syndergaard, Schwarber, Bryant, and plenty of others from last year.

3) Perhaps most important: For hitters, projections seem to use a regressed weighted average of recent performance to make their projections instead of looking at more granular indicators (they are better with most pitchers, because projections do seem to understand DIPS theory). What do I mean? Take Kris Bryant. He’s projected for a 137 wRC+ next year, which is actually a slight improvement from his 2015 production. Seems reasonable, right?

A deeper look reveals the projection to be somewhat unreasonable, because Kris Bryant’s 137 wRC+ was influenced significantly by a .378 BABIP. While the projected .345 BABIP for Bryant looks reasonable enough at first glance, we actually have tools for estimating BABIP. You can run a linear regression on the important components (using Hard Hit%, LD%, GB%, FB%, Oppo%, GB%*Spd, etc.), like I did, or just look at the public xBABIP calculators.

Either way, you don’t get a BABIP anywhere near .345. My model churns out 0.304 (while he hits it hard, he has extreme FB tendencies). Even if you kept all else equal (which may or may not be a reasonable assumption), Bryant’s wOBA drops to 0.335, making him a roughly 110-115 wRC+ player. Still a good third basemen, no doubt, but not the superstar that the projections (and fans) seem to think he is.

And BABIP is just one of many other variables you can look at this way. HR/FB rate is another obvious one, as it fluctuates greatly from year to year and doesn’t always line up with how many home runs a player SHOULD have hit. For every Kris Bryant being overrated by BABIP, you have a Brandon Belt who’s underrated by HR/FB (mostly because of AT&T Park’s right field).

4) FIP (and by extension fWAR) still doesn’t understand a good 5-10% of pitchers. Who am I talking about? I’m talking about the FIP beaters. Tyler Clippard. Antonio Bastardo. Chris Young. Marco Estrada. Julio Teheran. RA Dickey.

Hint: They’re almost all either extreme fly ball pitchers or knuckleballers.

For these pitchers, we’re forced to ignore they’re projected fWAR and look at their projected RA-9 WAR, which is a pretty sad state of affairs.

It’s not that I don’t like the projections. I actually find them quite useful. I just wish that FanGraphs didn’t overemphasize them, which seems to be the trend here, and instead spend more time focusing on fixing these issues and analyzing how these issues affect individual players.

Jetsy Extrano
Member
Jetsy Extrano
4 months 9 days ago

You make fair points, but the performance in the end is noticeably better than “APPROXIMATING team strength, give or take 10 wins”. The mean error in seasonal record is about 9 wins, or we can say 10 for that. But a chunk of that is irreducible error from playing the games, even if you knew all team strengths exactly. About 6.5 wins worth by a binomial assumption. Leaving about 7.5 (arithmetic in the squares) for error in team strength.

Noah Baron
Member
Noah Baron
4 months 9 days ago

Sure. I just wish FanGraphs devoted more resources to digging as deep as possible into individual players and their projections instead of just taking them as gospel.

To be fair, they’ve gotten better about doing this recently. Last year they were so shackled to the projections that they stuck to their guns about the Mets having a bad rotation (resulting in them being projected to be worse than the Marlins), something that predictably backfired.

BigChief
Member
Member
BigChief
4 months 8 days ago

I don’t think there is a single one that is consistently better than using career numbers to estimate a players BABIP. If you do know of one please let me know. I typically think that xBABIP calculators are a lot better for understanding if their is something in a players peripherals that has caused a large deviation from a career BABIP or if it is luck.

Also, I’d just like to point out that Estrada’s career ERA is only .2 less than his FIP and Bastardo actually has a higher career ERA than FIP.

BTW I do think there is room for improvement regarding projection systems, but it may not be as trivial as people often think.

Samuel
Member
Samuel
4 months 9 days ago

Homer alert!

Joe Panik is not a star? Do you have any evidence of this?

Brians Sticky Sock
Member
Brians Sticky Sock
4 months 9 days ago

You mean the 25 year old with 4.5 bWAR in 719 career plate appearances?

piddy
Member
piddy
4 months 9 days ago

He’s just reading off from the projection system…

Damaso
Member
Damaso
4 months 9 days ago

It’s annoying how many smart analyses are depending on only one projection system.

Zips and Steamer project wildly differently for a team like the Jays. And for the Sox.

Damaso
Member
Damaso
4 months 9 days ago

If you make an apples to apples comparison of Zips and Steamer for the jays by projecting each starter to 600pa/200ip, and each reserve guy to 150pa/65ip, then if my calculations are correct Zips projections comes out to a full 6.7war higher than steamer.

what sense am I supposed to make of that difference?

Do I trust my gut which says the weird massive declines steamer projections just don’t make any sense?

Noah Baron
Member
Noah Baron
4 months 9 days ago

Exactly. The thing a lot of people don’t really understand about the projections is that really small, subtle differences in individual players can add up and really influence the team rankings.

And plenty of these are easily observable, systematic biases.

the wiener
Member
the wiener
4 months 9 days ago

‘Steamer doesn’t buy Goins’ sudden spike in walk rate’

This shouldn’t matter as he will only be a sub infielder, Travis will be playing second and his upgrade with the bat will be better than anything Goins has ever or will ever do

TommyLasordid
Member
TommyLasordid
4 months 9 days ago

Things I’m looking forward to in 2016:

The dramatic fall off of the offensive production of the Giants infield, and the Panda’s short hoppers skipping past Hanley and down the right field line.

ice_hawk10
Member
ice_hawk10
4 months 9 days ago

i had forgotten about that. do defensive stats at 1B account for scooping ability? cause damn that could be a pretty important thing for someone who’s never played the position. i’m sure they’ll try to get him to apply himself in spring training, but of course that worked out so well in the outfield…

Rishi
Member
4 months 9 days ago

Agree with Thomas.

Radermecher
Member
Radermecher
4 months 8 days ago

Great article,enjoyed 2009 is not a constant.

wpDiscuz