Hopeless Forecasts and the Stereotype Threat

CLEVELAND — This spring, I’ve briefly inhabited the clubhouses of some teams that aren’t expected to do very well in 2018. I’ve been in Sarasota, Florida, to visit the Orioles. I dropped by the road locker room at Progressive Field when the Tigers and Royals were guests there last week. There are no great expectations in Baltimore, Detroit, and Kansas City this spring.

The projection systems have given those clubs little chance at postseason contention. In fact, according to FanGraphs, those three clubs each featured a 0% chance of winning the World Series as of Opening Day. The same was true for a handful of other teams, as well.

Of course, these prognostications aren’t available only to the interested public. They reach the ears of on-field personnel, too. PECOTA forecasts appear on MLB Network’s preseason coverage. Some players even visit this very web site. Our projections have the Royals winning 71 games, the Tigers 70, and the White Sox 65 in the AL Central — or 25, 26, and 31 games, respectively, behind the Indians.

In an era increasingly populated almost entirely of super teams and tanking teams, there is theoretically less possibility of contention, less reason to hope, for teams forecast to finish lower in the standings.

Eight teams posted run differentials of 100 runs or greater last season. For perspective, in 2011, 2012, and 2014, only four teams had exceeded 100-run-plus differentials. Nine teams, meanwhile, recorded run differentials of -90 or worse last season, the most since at least 2003. The NBA style of tanking to secure better draft picks, fetch more amateur bonus dollars, and clear payroll space is only expected to increase in depth and popularity after the Cubs and Astros won back-to-back World Series using such approaches. Moreover, consider this finding from The Ringer’s Zach Kram: “As a group, last year’s playoff teams are returning 91.7 percent of their positive WAR in 2018, a proportion the league hasn’t seen since the 1980s.”

There are always going to be surprises, of course, and the Mets and Pirates have our attention early, but if there was something of a “hope index” for teams expected not to contend, such indices would figure to be hovering at 30-year lows entering this season. Perhaps that’s one reason, as Forbes’ Maury Brown noted last week, that attendance has been down, although frigid weather and some interesting ticket strategies in Toronto to combat the secondary market — and perhaps elsewhere — are playing roles, too. Last Monday, 10 games featured attendances below 20,000 and three below 10,000.

We generally talk about the projections here from the viewpoint of dispassionate observers. As I’ve said, though, these numbers are available to everyone. So, I was curious: what’s it like for a player to be told, over and over, how awful his team will be? Does it lead to something like the Stereotype Threat, which finds that certain demographic groups, when reminded of their gender or race before a test, perform below actual skill and/or knowledge levels. Do teams that are projected to perform poorly play even worse than expected when they are constantly reminded of how awful they are expected to be?

Jeff Sullivan has compiled all our preseason projections since 2005. While the methodology has been tweaked and while projections are far from perfect, the figures still give us a reasonably accurate portrayal of how FanGraphs expected teams to perform before a particular season began. I examined all the teams FanGraphs has forecast to win 72 or fewer games since 2005 and compared that to the actual wins:

Projected vs. Actual Result for Rebuilding/Tanking Teams
Team Year Projected Actual Difference
Royals 2005 68 56 -12
Pirates 2005 69 67 -2
Rays 2005 70 67 -3
DBacks 2005 71 77 6
Reds 2005 72 73 1
Royals 2006 65 62 -3
Rockies 2006 71 76 5
Marlins 2006 71 78 7
Rays 2006 71 61 -10
Reds 2006 72 80 8
Royals 2007 65 69 4
Nationals 2007 70 73 3
Rays 2007 71 66 -5
Orioles 2008 67 68 1
Marlins 2008 68 84 16
Pirates 2008 70 67 -3
Nationals 2008 70 59 -11
Pirates 2009 70 62 -8
Rangers 2009 72 87 15
Blue Jays 2010 65 85 20
Astros 2010 69 76 7
Royals 2010 72 67 -5
Pirates 2010 72 57 -15
Astros 2011 66 56 -10
Royals 2011 68 71 3
Pirates 2011 70 72 2
Nationals 2011 72 80 8
Astros 2012 64 55 -9
Orioles 2012 70 93 23
Twins 2012 71 66 -5
Cubs 2012 71 61 -10
Pirates 2012 72 79 7
Astros 2013 60 51 -9
Twins 2013 67 66 -1
Marlins 2013 69 62 -7
Astros 2014 67 70 3
Twins 2014 71 70 -1
Cubs 2014 71 73 2
Phillies 2015 66 63 -3
Braves 2015 71 67 -4
Phillies 2016 64 71 7
Braves 2016 68 68 0
Brewers 2016 69 73 4
Reds 2016 71 68 -3
Padres 2017 66 71 5
Reds 2017 68 68 1
White Sox 2017 68 67 -1
Brewers 2017 70 86 16
Phillies 2017 72 66 -6
Braves 2017 72 72 0
Totals 3457 3482 25
Highlighted cells designate a winning record.

As you can see, the projections have been remarkably accurate at the aggregate level. That said, teams expected to lose 90-plus game have actually outperformed their forecasts slightly, recording 69.6 actual wins versus 69.1 projected ones. In general, there seems to be no Stereotype Threat-like effect. A team forecast to perform poorly does not underperform its already modest expectations — or, doesn’t underperform FanGraphs’ expectations, at least.

Still, I asked players if it was difficult at all having an awareness of the projections, if it sapped energy and enthusiasm from the clubhouse even before the first pitch of the season was thrown. Or, possibly, was there some other kind of other effect?

If anything, the players polled seemed to indicate there was more energy. The projections can offer motivational fodder. And there is the “You Never Know, It’s Baseball” belief that anything can happen.

Tigers ace Michael Fulmer, for example, believes the clubhouse spirit was better this spring.

“When we went into camp this spring, there were 55 guys in the clubhouse and you had to introduce yourself to a lot of them,” Fulmer said. “But you also see them play and they are that much more hungry to make the team. There is that good source of competitiveness between guys. No one’s spot is safe. I thought it was really fun. It kind of brought me back to ’16 when I was trying to make the team out of camp in the bullpen or anyway I could. I remember that time, that competitiveness I had in trying to beat someone out for a spot.”

Baseball is a team sport played by individuals. A projection, meanwhile, is an accumulation of individual forecasts. So even if a team is not expected to perform well, players still have individual incentives to pursue. Fulmer is probably on to something.

New Tigers manager Ron Gardenhire was asked about how he approaches his job while leading a rebuilding effort. Gardenhire is definitely from the old school. There was a box of cigars on his desk in the visiting manager’s office at Progressive Field on Tuesday — modest incentives to compete each night — but there were also two open laptops. Managers have more and more become extensions of the front office, even gray-haired managers. Front offices are increasingly stripping themselves of emotion and operating with sober, realistic outlooks. In fact, if a team is tanking, its front-office group might be rooting for draft position and bonus pools more than team wins.

“It’s not about ‘We’re in are rebuilding mode,’” said Gardenhire of his message. “It’s about letting guys play that need to grow… We are going to try and beat you every day, and that is development, too.

“So my job is still the same, but I understand where we are at.”

And where that is, is at a place with virtually no chance at the postseason.

Out of the 50 teams projected to lose 90 or more games since 2005, only five ultimately recorded winning records. Only the 2012 Orioles, forecast for 70 wins, made the postseason — in this case, with 93 actual victories.

Said longtime Royal Alex Gordon: “Obviously you guys do your research and make a projection, but at the end of the day, it really doesn’t mean anything. It’s just what you guys think.”

But the forecasts aren’t the product of what anyone at FanGraphs thinks, per se. Rather, our projected standings — which take into account how talents and skills grow and age, etc — are calculated with a series of simulations. It’s an algorithmic stew.

It’s one thing to prove a subjective human filled with biases to be wrong, but what about a heartless, emotionless computer? I asked Tigers pitcher Daniel Norris — a thoughtful, curious player who knows his spin rate — if it’s different when a computer projection is forecasting one thing versus, say, a media member giving a prediction. Is that more difficult to refute? Is it more defeating?

Norris had an interesting thought.

“One day a guy could go out there and throw a bullpen and figure out a two-seamer that could propel him to another echelon, and that’s just not projected,” Norris said. “What if he does get better? Every year you see it: guys breakout and they break out for reasons of [fundamental improvement]. So imagine if we had four or five breakout guys?”

Projection forecasts do not know about launch angles and swing changes. They do not know about new pitches. We are in an era of experimentation. Projection systems will perhaps be more and more challenged going forward. Still, as the Tigers know, their odds are long. Competing seems, to this author, more about professionalism and personal incentive than it does about entertaining long-shot odds at October. And if there are too many clubs in similar situations in an era of super teams and tanking teams trying to become super teams, there’s a debate to be had on whether this is good for the sport. The misery index is up, and it’s only in part due to the weather.

A Cleveland native, FanGraphs writer Travis Sawchik is the author of the New York Times bestselling book, Big Data Baseball. He also contributes to The Athletic Cleveland, and has written for the Pittsburgh Tribune-Review, among other outlets. Follow him on Twitter @Travis_Sawchik.

newest oldest most voted

Off topic but does anyone have a clue what’s going with xwOBA this year? Up a ton from .321 to .332, Exit Velocity up from 86.7 to 88.2. The 2016-2017 difference that produced the sizable run increase was only .316 to .321. Is there a huge run surge coming once it warms up or are there possible calibration issues with Statcast?


Sac bunts are down? Sac Bunts are an exit velo killer…
Edit: they are down! From about .006 per PA to .004(unless i did something wrong)


As far as xwOBA, I believe they tweaked it to begin accounting for a player’s speed.


If they tweaked it, that would apply retroactively would it not? So the increase in xwOBA would still be the same. Still left with how to explain the Exit Velocity increase.

Also HR/FB rate down significantly vs April of last year, 12.8% to 11.6. Clearly the cold explains some but seems steep with the corresponding EV increase