KC Royals: Why Don’t Projection Systems Like Them?

Nov 3, 2015; Kansas City, MO, USA; A Kansas City Royals fans shows support while waiting on players to arrive at Union Station. Mandatory Credit: Denny Medley-USA TODAY Sports
Nov 3, 2015; Kansas City, MO, USA; A Kansas City Royals fans shows support while waiting on players to arrive at Union Station. Mandatory Credit: Denny Medley-USA TODAY Sports /

The KC Royals are again unloved by the top projections systems among the sabermetric crowd. But in 2016, the pundits are mostly ignoring them.

CBS Sports Matt Snyder and ESPN’s Buster Olney consider Kansas City the second best team in baseball behind pre-season darling Cubs. Bleacher Report’s Zachary Rymer see the KC Royals as the third best team in MLB. More interestingly, the entire Baseball Prospectus staff projects an average of 88 wins for the Kansas City Royals despite knowing that their own PECOTA system predicts 76 wins:

"Our staff didn’t want it. It’s not just that when we polled our writers for their own Royals predictions—before PECOTA had been run—not one of them went as low as 76. It’s not even that not one of them went lower than 80, or that only one of 27 responses was lower than 85, or that the plurality response was 90, or that the average was 88. It’s this: When I asked a follow-up question a few days later—“If I told you PECOTA projects them to win 76 games this year, does your answer change?”—the response was overwhelmingly “nah.” In fact… counting them out… 73 percent of staff said it didn’t change their answer at all. As one put it: “No, because the projections just seem to not like the Royals.” This seems awfully close to a crisis of confidence."

On the other hand, SI.com ranks the KC Royals 19th headed into spring training. Even so, there seems to be a strong pushback against the projection systems, even among sabermetric analysts, when it comes to the 2016 Kansas City Royals.

This response is actually quite rational. The projections systems have been wrong about the KC Royals for three years running. Fangraphs.com’s Steamer system projected 79 wins both in 2014 and 2015: years in which the Kansas City Royals won the American League pennant twice in a row with 89 and 95 regular season wins. And, oh yeah, the KC Royals earned the second championship in franchise history. Steamer was also wrong about the 2013 KC Royals, projecting 82 wins for a team that actually won 86.

However, the Kansas City Royals aren’t the only team Steamer has missed badly on the last three seasons. Instead, they’re just the most noticeable because they’ve advanced to two straight World Series. The Pirates have actually been the team that has outperformed Steamer to the largest extent between 2013-15. Here’s the full list (chart courtesy of Fangraphs.com)

wins-projected-wins-2013-2015 /

Notice that Steamer has missed on the Pirates, Royals, Cardinals, and Orioles by more than 20 total games over the last three seasons. After those four, there’s a rather steep drop-off in the difference between predicted versus actual performance.

More from Kings of Kauffman

Is there anything those four teams have in common?

I know this is a rather simple analysis. Really, it’s more about trying to form a hypothesis about what projections systems could be missing about recent high-performers. The piece in question made clear that neither the Pirates or Royals were anything unusual in terms of Steamer missing over three seasons.

I’m also not even looking at other systems like PECOTA, ZiPS, or Clay Davenport, so you can’t take anything that follows as a generalized flaw for sabermetric projection systems.

With that being said, the only real common denominators I could find between the four Steamer outliers is that they all ranked in the top eight in bullpen fWAR (Wins Above Replacement) over the last three seasons, and all four teams ranked in the top six in strand rate (for the entire pitching staff). These features seem related in that if a team has a good relief staff, they’re likely to strand more runners when their starters get in trouble.

Of course, I looked at other team statistics. But, the four teams’ Isolated Power (.ISO), OBP, strikeout rate (K%), Base-running (BsR), and starting pitching fWAR all showed considerable divergence. The only other strong similarity between the group is they all ranked in the top nine in Defensive Runs Saved, but that correlation did not hold up for other defensive metrics like Ultimate Zone Rating (UZR) or DEF.

I also saw that the KC Royals, Cardinals, and Orioles ranked in the top five in CLUTCH (situational hitting performance) from 2013-15, but the Pirates came in at no. 21 in this metric. Sabermetric types largely dismiss CLUTCH as a true skill because multiple studies have found little ability for either players, or teams, to repeat performance in this area.

Since reliever performance suffers from small sample size problems, team bullpen metrics can vary quite a bit season to season even if the players remain mostly constant. In short, the sabermetric types can point to variance to explain bullpen “overperformance” for all four teams, and luck with respect to CLUTCH for the Royals, Cards, and Orioles.

Next: Royals Continue To Befuddle Experts

At this point, however, many stat-heads have lost faith in the ability of projections systems to evaluate the KC Royals. If Kansas City defies predictions for the fourth straight year, we will see more people wonder if something has changed about the game that has rendered these models obsolete.