Here’s Where the Projections Disagree
Good news! It looks like we have our ZiPS information uploaded. What that actually means — until now, whenever you looked at the projected standings, you were seeing Steamer projections, only. Now you should be seeing 50/50 Steamer and ZiPS projections. This is the way we’ve done it for a while, evenly compromising between the two systems.
So, first, here’s the current projected winning-percentage landscape, displayed poorly and with an arbitrarily-truncated y-axis.
I can’t imagine there’s too much that takes you by surprise. Cubs good. Padres bad. Dodgers good. Brewers bad. We have some understanding of pretty much everyone, and forecasting systems never disagree that dramatically.
But they can and do disagree! So we can quickly look at that. What we have right now are blended projections. What we had at the end of last week were the Steamer projections. It requires only some simple math to isolate the ZiPS projections, and from there, we can look at the differences between what ZiPS says and what Steamer says. In this plot, I’ve subtracted Steamer win% from ZiPS win%. The teams on the left are favored by ZiPS. The teams on the right are favored by Steamer.
I’ve done nothing here to examine the Why. That would be something for a larger article. But, Blue Jays fans might’ve noticed a bump in the projected standings today. That’s because ZiPS likes them about five wins better than Steamer does. Ditto the Orioles and, partially, the Mets. You also see the Cubs get a bump, with Steamer being a little higher on the Dodgers. But Steamer has the biggest positive disagreement in the directions of both the Royals and Yankees, who Steamer likes more by four and a half wins. Interesting! Again, I’ve done no further research. This whole thing took 10 minutes from start to finish.
This doesn’t mean ZiPS is more right, nor does it mean Steamer is more right. There’s a reason we fold them both in with equal weights. This is just to help explain why you might’ve seen some shifts in the standings. ZiPS has got opinions, see. It’s not just Steamer re-packaged.
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.
I wonder if ZIPS weights for team presidents. Randy Levine should be worth roughly negative 4.5 wins.
well, I’m looking forward to The Why that this was a teaser for
5 of the worst 6 teams, and 8 of the worst 11 teams are NL teams, yet among the top 10 teams, they are split evenly 5-5
The three non-NL teams in the bottom 11? All in the AL Central.
Could Detroit be a surprise wild card contender just because they get to play a third of their season against garbage?
People have been saying this for the past 3 years.
And it’s just as true now
A third against garbage, but also 19 against the number two team in the league.
I have to believe the discrepancy on the part of the Yankees is how the young players do and perhapsSteamer is more optimistic injury prone players or players coming off injuries stay healthy and/or bounce back.
Have to remind yourself that the STD error on these projections is +/- 6 wins. An 81 Win team can win 75-87 wins and still be a win for the projection
The standard deviation for a 162-trial binomial distribution at 50% success (coin flip) is 6.4, so there’s no reason to expect a team’s SD to be below that. A team’s true talent level fluctuates throughout the year due to injuries and actual improvement/regression, so it’s hard to imagine getting the SD within 7.5-8 wins.
On the Yankees part, that’s probably a big part of it. But also, the top three absolute difference teams are in the AL East (TOR, BAL, NYY). Might there be some kind of feedback loop for these three teams based on the weighted schedule?
ZiPS has never beaten Steamer in a head-to-head projection accuracy evaluation, so the 50/50 weighting seems too ZiPS-heavy to me.
Wild guess: Steamer hates the Jays’ rotation, loathes Estrada, is disgusted by the smell of roses.
Do both of the systems get their playing time projections from the depth charts? In other words, is the only point of distinction the projected player performance, and not projected playing time?
Correct.
I noticed Musgrove had his innings bumped from 46 to 47. Come on…give him some love!
“I’ve done nothing here to examine the Why. That would be something for a larger article.”
I hope that larger article gets written.
either there is a bug in the code or someone needs to explain to me how the phillies go from being almost 5 wins ahead of the braves in the team WAR list in the “Depth Charts” page sidebar to being 2 wins behind in the projected standings. cubs-dodgers order reversed as well, and several more i believe. i know that the WAR totals are 2 or 3 wins too high on average due to playing time inconsistencies, and i know that different opponents implies different win total for teams of equal strength, but a 6.5-7 win swing for two teams in the same division seems very implausible to me!
… ps: if there IS an error, my guess is that the projection page is using old rosters… several teams projections have been almost unchanged there all winter, even while the depth charts WAR totals were moving up and down after major transactions.
Looks like the Angels didn’t take as big of a hit from Steamer to ZIPS as everyone was expecting
Maybe ZiPS has figured out where the Orioles voodoo projection beating magic has come from over the past 5 years.
Steamer hates Estrada and adds a full run on his ERA compared to Zips. The five starters as a whole have a 4.15 ERA compared to 3.83 with Zips. I think Steamer uses velo in their projections so this might have something to do with it.