Do You See Something the Projections Don’t?
Last night I was out getting a drink with our own Matthew Kory. His favorite team is the Red Sox. My favorite team is the Mariners. The bar we went to was showing the Mariners game, and while the Mariners were actually winning, that did nothing to stem the tide of jokes at our own expense. They’re two very different teams in two very similar situations — they came in with a lot of hype and promise, some people labeling them World Series contenders, and to this point they’ve more or less sucked. I don’t know which team has been the bigger disappointment. There’s still time yet, but while that means things could get better, that means, also, things could get worse.
The conversation turned to looking ahead. It was just last week I wrote about the meaning of the standings through a couple months, relative to the meaning of the projections. The numbers suggested that the Sox and Mariners would be pretty good. They continue to suggest that, and, my brain knows it should believe that. But it can be difficult to fully accept, when you’re watching a team playing different from the expectations. It feels like a bad team is just a bad team. It feels like a good team has something special going on. There are feelings you’re supposed to feel, and feelings you actually feel. Actual feelings, you could say, are greatly prone to recency bias.
The conversation has led to this post. It’s another post with an assortment of polls, asking for your participation. The idea: do you see something, in the teams you follow, the projections don’t? Do you see reason to doubt the projected records? The polls will ask about five teams: the Red Sox, Mariners, Royals, Cardinals, and Nationals.
We all like to think we can spot things. We all like to think we can identify characteristics that formulas would miss. And it’s true, we can do those things. It’s more a matter of how often, and of how often there are false positives or whatever you want to call them. Just because you see something doesn’t mean it’s significant. Just because you see something doesn’t mean it’ll sustain. Seeing the future is complicated. It’s because understanding the present is complicated, and the present is presently happening.
Below, we have five teams you could say are under- or over-achieving. One way you could look at this would be to compare actual record against projected record. But I’m going a different way, using a particular FanGraphs feature. You, probably, know the Playoff Odds page, and there you can see projected rest-of-season record based on Steamer, ZiPS, and the depth charts. Nothing weird there — we refer to those numbers pretty often. But there’s a tab up there, that takes you to a similar-looking page titled “Season to Date Stats Mode.” On that page, you get what you could consider to be team projections, but instead of ZiPS and Steamer, they’re based on what the players have actually done in 2015. This makes for a fun comparison. This is the core of the post. On the regular Playoff Odds page, for example, the Red Sox are pretty good. Using season-to-date stats mode, the Red Sox are pretty bad. My question to you: which do you believe?
The five teams being polled are the five teams with the biggest differences between projected winning percentage and season-to-date stats mode projected winning percentage. The Red Sox, Mariners, and Nationals have under-achieved, while the Royals and Cardinals have over-achieved. For the sake of responding to the polls, consider how you feel now, and don’t assume future trades. In the Cardinals’ case, don’t assume future discipline. I just want to see where you guys come down. Thank you in advance for helping out.
Red Sox
Steamer/ZiPS projected win%: .534
Season-to-date mode projected win%: .403
Difference: .131
Quick summary: They were supposed to have good position players and good depth to make up for a mediocre pitching staff. The pitching staff has held up its end of the arrangement.
Mariners
Steamer/ZiPS projected win%: .533
Season-to-date mode projected win%: .432
Difference: .101
Quick summary: A year ago, the Mariners had a 93 wRC+, and then they added Nelson Cruz, who’s overall been solid. As a result of that upgrade, this year the Mariners have a 93 wRC+, but with a worse defense.
Royals
Steamer/ZiPS projected win%: .487
Season-to-date mode projected win%: .574
Difference: .087
Quick summary: It’s funny what happens to an offense when Mike Moustakas, Eric Hosmer, and Kendrys Morales decide collectively to not be bad. You don’t necessarily have to replace James Shields with pitching.
Cardinals
Steamer/ZiPS projected win%: .534
Season-to-date mode projected win%: .615
Difference: .081
Quick summary: Adversity doesn’t matter to these people. They’ve had a top-five rotation without Adam Wainwright, the bullpen has a sub-2 ERA, and the offense has been good enough with Jason Heyward hitting like Yadier Molina and Yadier Molina hitting like Ben Revere.
Nationals
Steamer/ZiPS projected win%: .574
Season-to-date mode projected win%: .498
Difference: .076
Quick summary: The unsinkable team has sunk to second place, and by some measures there’s been no worse starting pitcher in baseball than Stephen Strasburg. Injuries have been a problem, but even the players expected to be on the roster have largely under-performed, as shown by the numbers above. All this, somehow, despite Bryce Harper and Max Scherzer.
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.
The red sox have been bad and seemingly everybody, except for the projection systems used here, knew they had a very high probability of being bad.
not the way the audience voted here
http://www.fangraphs.com/blogs/the-team-projections-and-you-again/#redsox
not sure the poll sample there would be an objective one.
I’m willing to bet it’s at least as objective as whatever source you’re using.
What sample of polled people *would* be objective? (hint: the answer is none)
fair enough.
but asking fangraphs readers whether they trust fangraphs projections might be little more than confirmation bias.
also, given the sheer volume of pro-red sox articles on fangraphs in the offseason, this poll might be tantamount to asking viewers if they want a refreshing coke after you’ve inundated them with coke commercials.
also – even red sox complained about the number of anti-red sox comments under red sox articles. those might be a better indicator of what fangraphs readers felt about the projections.
I think your poll actually supports the argument that people thought the Red Sox would be worse than projected. Yes, the top answer was that the projection was “good”, but that was the case for the majority of teams, and I think some of this is because people just choose “good” when they have no strong opinion or don’t follow the team closely. The second highest vote in the poll did say that the projection was too high by 3 games.
I thought the projection was off by a lot and that the Red Sox were about a .500 baseball team. I fully expected their pitching to be awful, and it has been, but I thought the offense would be better.
If you honestly thought the Red Sox had a high probability to be this bad, please explain to me why, because I can’t figure out why you would believe that other than wishful thinking.
I thought they had a high probability of being mediocre. In fact, I still think they have a decent probability of ending mediocre.
But the contending projections for the red sox never passed my smell test, at least. There were a few very suspicious things that didn’t add up for me.
1. This roster of players was projected to be worth near 25% more WAR this year than the same group of players was worth last year. That always seemed extremely unlikely as a 50th percentile projection. This almost never happens.
2. They were projected for this mass improvement despite having one of the oldest rosters in the league, which moved it from extremely unlikely to unbelievably unlikely, for me.
3. They were projected to have an elite offense despite having zero projected elite hitters. That is an extremely weird projection.
4. They were projected to have elite depth even though their starting lineup included 2 raw 22yr olds and 2 part time 34yr old vets.
5. They were projected for elite defense despite having 3 OF with virtually zero professional experience in the outfield, which is obviously crazy, and half the infield was mediocre at best.
6. Despite not having one guy on the roster who was a fulltime starting OF in 2014, they were still projected to have one of the best two way outfields in baseball. That was always ridiculous.
I don’t have much problems with projections in general, but we know they are very fallible, and this projection had far too many warning signs imo.
And it was a bit annoying that people hid behind the projections for the red sox, but had no problem dismissing the very similar projections for the yankees. this always stuck in my craw – everyone was ignoring the yankee projectiins but then getting mad when anyone questioned the red sox projections.
In the end, I suspect the projections need to take better account of the error bars on individual projections – I suspect that a team of very old and very young guys like the sox has much larger error bars around their projections than other teams do.
I think we should all be wary of positive defensive projections for defenders making big shifts on the the defensive spectrum.
This is a fair assessment. I see nothing wrong with taking the under on projections and calling for the Red Sox to be a mediocre team. The projections at the beginning of the season were only calling for the Red Sox to finish slightly over .500 anyways, and now they project to finish slightly under .500. Nothing strange there. I just think if you believed the Red Sox would play at this pace and finish the season with a .400 winning percentage you were being a little disingenuous.
“everybody knew they had a very high probability of being bad.” You are a liar.
Underperformance is always a possibility but I don’t think anybody thought that they would be this apocalyptically bad
Yeah, even those of us who expected this performance from the rotation though that Sandoval and the Mookster would at least produce a 105 wRC+, and that Ortiz and Hanley would rake.
Their rotation was a bunch of very mediocre pitchers, except for Rick Porcello, who needs a good defense behind him. To support Porcello, they signed several terrible defenders and then moved the worst one to a position he’d never played before.
Mookie Betts had good projections and a short history of MLB success – just like Brad Miller and Wil Myers did prior to 2014.
This is all stuff that was known before the season.
Ramirez can’t play defense, but Sandoval was by all accounts nothing less than an average defender before coming to the Red Sox. I’m not sure who else could constitute the “several terrible defenders.”
No offense intended, “but Sandoval was by all accounts nothing less than an average defender before coming to the Red Sox” comes across as a glass is half full type perspective, there are many, many accounts that Sandoval was nothing more then an average defender who would (typically) get worse with age (although obviously this year he’s been even worse then most would have projected).
Fine, but I still wouldn’t have qualified him as terrible and I wouldn’t qualify one player as several.
If you knew they “had a high probability of being bad” then hopefully you made a fortune by betting heavily against them each game up to this point as your odds have been extremely favorable. If not, it’s safe to assume that either you didn’t really think they’d be this bad or you hate money.
There may or may not be some anti-Cardinals sentiment in the voting today…
Eh: they’ve suffered a ton of injuries, and appear to be propped up by numbers we think we know are rather random. They’re a good team, but I’d take the under on them playing .615 baseball from here out.
Well, I agree that the Cardinals probably aren’t the next coming of the 2001 Mariners, but they’ve flourished despite the injuries in large part due to some excellent depth (e.g. Reynolds, Grichuk, Garcia), and some of the injured players will be back before too long (Lynn, Walden, Holliday). Meanwhile, some of the breakout stars like Wong, Wacha, Martinez, Siegrist, and Rosenthal look to be for real.
I’m sure it’s a total coincidence that your pseudonym is “Cardinal” spelled backwards and then you actually made the claim that Mark Reynolds and Randal Grichuk constitute “excellent depth.” They aren’t, for the record. They’re functional, which is more than some teams can say about their backups, but they’re otherwise fairly ordinary reserves. Garcia, on the other hand, has always been talented enough to believe that he can maintain an above average performance as long as he remains healthy.
But thinking that the Cardinals will be healthy from here on out because they’ve already suffered their injuries is a variation of the gambler’s fallacy. Assuming that they experience a normal rate of missed days from here on for a major league team and that players outperforming their peripherals do regress (which includes Wacha), then .534 looks like a reasonable number. Heck, it might even be a little optimistic.
Well, since Reynolds and Grichuk are indeed better than the average bench player (functional is an understatement), that does make them very good depth pieces. Furthermore, Grichuk is already good enough to start for a lot of teams without injured outfielders and has a good shot to be the starting RF from the get go next year. Yes, health has always been Garcia’s issue, which is why he was originally only counted on as depth beyond the initially projected 5 starters, depth that has now paid off.
Yeah, there could be more injuries, but those odds are mostly cancelled out by the players returning from injury, including top prospect Marco Gonzales in AAA.
Yes, Wacha probably isn’t (yet) a true Cy Young candidate, but he’s still a well above-average starter, and regression works both ways, such as with Molina, Heyward, and Jay.
*facepalm* No, Randall Grichuk is most certainly not “already good enough to start for a lot of teams without injured outfielders.” His major league walk and strike out rates are 3.5% and 28.4% respectively. Marco Gonzales is also nothing special. The reality is that we didn’t expect the Cardinals to have a .615 winning percentage even before they lost players, so they really shouldn’t be anywhere close to that going forward. As noted, I would expect something around .500.
I disagree, I think it’s only rationale for people to not expect a team to continue with a .615 winning percentage, considering how rare that is and Waino is out.
Whoa, lot of Ryans in the comments today
What is going on with the A’s?
I realize this is an article about projections, but is what the A’s have been doing thus far this year historic? The have the second highest BaseRuns run differential in the majors, yet they have the most losses in the AL. Also have an actual run differential of +23 and are 12 games below .500.
Oh, and how do the projects account for #cardinalsdevilmagic?
i think it would be interesting to create some sort of mass poll that asked readers to make their projections vs the fangraphs projections. Many commentors often retroactively act smug about team’s performances and act like they knew all along, so it would be great to have some way of forcing them to put their money where their mouth is.
That’s basically what this was about
http://www.fangraphs.com/blogs/your-opinions-of-the-team-projections/
Jeff – you should have added one more field to the polls: are you a fan of the team? My hunch is that a fan would vote closer to actual results than projections than a non fan, but it would be interesting to see some data.
Sample of one, but I’m a fan of the Nats and voted for the projections going forward. Our pitchers have a long track record of success and will probably sort things out. Injuries are a pain but Bryce Harper. Bullpen is a legit mess but we have a rock solid closer (except when October comes…). I see no reason to distrust the projections despite the pessimism of many fans.
What the Cardinals are doing isn’t just over-achieving, it’s utterly absurd. They’ll exceed their projections because they’re the Cardinals, but even if they didn’t lose all those key players, they would have to fall back to Earth, because baseball teams simply do not win 2/3 of their games very often, even really good, really deep, really healthy teams.
Cardinals–> Red Birds–> “The Red Woman” Melisandre.
When things get bad they have just have a witch perform blood magic. They know the future dude.
These Nationals are baffling. They have a 20-5 stretch. Before and after that, they are 13-26.
As a nats fan I feel as though they deserve their current record or worse. The projections likely underestimate the effects of all of the lingering injuries (Stras, Zimmerman, Zimmermann, werth, span, and fister), even when these guys were playing they seemed to be hurt or timing to get their timing back.
Yes they totally “deserve” their current record, although there’s no way that Ian Desmond is this terrible even with his contact issues. But the question is whether they will continue to play so erratically and the answer has to be no, given that, as you point out, they have a lot of guys coming back from injury. Anyways, we freak out every season at around this time (since 2012 anyways), and then they heat up in July and light up the division the rest of way. The only reason they couldn’t pull it off in 2013 was because the hot streak didn’t start until August and the Braves were inexplicably good. Of course, if the injuries linger (or spread to our healthy guys, god forbid), then we might be in trouble.
Typical Mariners fan. Why aren’t the Rangers listed here? They are overperforming their preseason projections and have plenty of reasons for their success. And plenty of reasons that their success wouldn’t look sustainable for the rest of the season. Typical #6 Organization fan.
If you read the article, you’ll know that Jeff took the 5 teams that are furthest from their preseason projections. The Rangers are not in the top 5. But you didn’t read the article, because you care more about being a righteous commentator than about being a well-informed baseball fan.
I think it’s most likely somewhere in the middle with the exception of the Nats. They at least have a bunch of injuries that have hurt them. If they get healthy I don’t see any reason they can’t play to their projections.
I’m surprised you risked putting the Royals in an open poll.
I voted twice.
You still have 33. Quit slacking.
The Red Sox are as bad as they are for a reason. I go by what my eyes tell me. I’ve never seen as many batters rolling over on outside pitches for weak grounders to their pull side. Ortiz, Pedroia, and Sandoval stand out for me. Check the hard-hit ball data on the team as a whole. They have to be near the bottom.
Also, Betts and Swihart, as much as I see their potential, really needed a full year in AAA before they were brought up. Now they’re getting service time in for a team that’s a lost cause, putting them that much closer to arbitration. Why Ben, why?
Close! They’ve got the 8th-lowest “Hard” percentage, but the highest “Soft” percentage!
http://www.fangraphs.com/leaders.aspx?pos=all&stats=bat&lg=all&qual=0&type=2&season=2015&month=0&season1=2015&ind=0&team=0,ts&rost=0&age=0&filter=&players=0&sort=16,d
Are their wives leaving them?
How have rest of year projections v. actual performance done historically once you are this far in? For example, for the group where there is a large difference like above is there some bias? Is there an article on it?
Jeff referenced his article on this topic from last week – http://www.fangraphs.com/blogs/the-meaning-of-the-standings-so-far/
Basically projections are better at predicting rest of season win % than current standings.
By OPS margin (hitters OPS minus pitchers OPS allowed), the Mariners way overachieved last year. They are right about where they should be win-loss wise this year. I am a Mariners fan that predicted they would suck and I’m feeling like freaking Nostradamus right now. Without Felix and Cruz you would be looking at the ’61 Mets.
That’s pretty terrible, considering that the Mets were founded in 1962.
Thanks BG, trusting my memory when I shouldn’t be, it was indeed the ’62 Mets that lost 120 games. Twas I that was founded in 1961!
Are we acknowledging the eye test?
Are you new? The fans scouting report has been a thing for a long time.
An unused thing no one generally talks about or puts much stock in, but I guess it does technically exist.
Your stereotyping is showing.
I always thought the projections were selling the Royals short. They won 89 games last year despite some pretty terrible performances from supposed key players (Butler, Moustakas, Perez in the second half). If they could even improve a bit on offense (which seemed like a decent bet) they were probably going to be similar to last year.
And 87 the year before. And both years it was way more than their projections.
My personal opinion is that the projection systems simply have yet to figure out a team of that mold – defense, speed and bullpen first, no clear-cut stars on the roster, agressive batters, that sort of thing. It doesn’t fit the sabermetric mold, but it seems to work.
Projection systems take numbers and throw them into an algorithmic blender. They don’t have any real practical applications and you certainly shouldn’t base your expectations on them.
That’s a lot of baby you’re throwing out with that bathwater.
The internet savvy Cardinals after 64 games, have an opponents OPS with runners in scoring position of .554.
This is basically like having Richie Weeks and Roughened Odor bat every time somebody was on 2nd or 3rd base. Erick Sogard would be an upgrade.
So yeah, I think maybe this will not continue to be the greatest pitching staff in baseball history.