About a month ago we took a look at the LGT pre-season predictions and found some winners. You can review the results for hitters here.
Some expressed interest in seeing how the LGT community fared against some projection systems. For this I was able to track down the pre-season projections for Steamer and ZiPS to compare against our predictions. I also had Oliver, but being based on 600 PAs it was rendered basically useless for us.
If you can't remember how the scoring worked, please refer back to the old post.
Scoring stayed the same except that Steamer and ZiPS were added in and scored independently of each other, such that with the points system they could both earn the maximum since they were only competing against LGT and not each other. Also, points were tallied for how they fell inside our projection "system" (i.e., our actual predictions). This means that for comparison, no points were taken away from LGT predictors.
Below is the leader board from LGT only predictions. The point figures for the projections were: Steamer - 72.31 and ZiPS - 61.75. Tyler was able to best the ZiPS projections but not Steamer, which took the crown. We'll explore how...
Both projections picked up points for being less optimistic than LGT as a whole, but Steamer managed 10 points out of, a pretty significant amount driven by the OBP and SLG numbers.
The only points here were 5 to ZiPS for nailing Yan's OBP. On the whole, ZiPS fell pretty in line with LGT with Steamer falling behind, though neither group was as optimistic with Yan's BA and power numbers as they should have been. All hail, Yanimal!
Nick Swisher stunk of a very foul odor this year. None of the 3 systems thought he'd be anywhere close to as bad as he was, but ZiPS picked up 8 and Steamer 4 for thinking he'd have a slightly above average year. I think that shows you just how bad Bro was.
That stuff I said about Swisher? Just repeat it for Kipnis. Except this time the number were huge, with Steamer taking 25.5 and ZiPS taking 14.25. Both systems were less encouraged by Jason's 2013 than anyone at LGT and they covered a lot of ground because of it. I'd like to not though that Kipnis' 22 steals are impressive against the projections given his 60-100 less PAs and the huge downswing in OBP.
Steamer and ZiPS get knocked the same way for predicting a full season out for Cabrera, though Steamer still managed to walk out with 7.167 points for being the least optimistic of the 3.
And by the way, LGT, Steamer and ZiPS all saw .318 for Asdrubal's OBP. That he couldn't deliver is a modern day tragedy.
Steamer managed 5.5 and ZiPS picked up 4.5 for Chisenhall. A pretty average amount among a group that largely didn't see this kind of production coming. Everyone overshot HR expectation though, when you consider the increase in PAs. Isolated, however, ZiPS nailed it.
Neither Steamer nor ZiPS picked up any points for Brantley and neither of them deserved any. Not only do their numbers look like a total failure against actual numbers and against LGT, but both systems scored among the worst predictions versus individual LGT members in nearly every category. The only one that didn't totally screw up was ZiPS PA figure. Woopty-doo.
Steamer and ZiPS managed 2.5 and 8 points, respectively for Bourn. The slash lines weren't terribly far off, though, like everyone else, the stolen bases just weren't there. In a way, Michael Bourn's 2014 was like Jason Kipnis' 2014, but without the stolen bases. That's a weird sentence to type.
9.5 for Steamer and 8 for ZiPS here. The PAs and batting average figures were all fairly accurate, it's just what Magic Hands did with those hits wasn't quite so thrilling.
Steamer gets 8 and ZiPS gets 11.
Next year LGT will be introducing its own projection system - TYLER. It will combine, well... Tyler... and the Cleveland Indians... and a computer. Revolutionary.
Kidding aside, I didn't really expect either system to do as well as they did, given the wild fluctuations in offensive output for the 2014 Indians. That said, Steamer made most of its ground up on Jason Kipnis (35% of its point total) so maybe it is getting more credit than it deserves. Even with a muted point total in that category, both systems still would have placed in the top tier of predictors. That's not too shabby. It also seems that, on the whole, our community seems to be generally more optimistic. I guess that is natural.
There will be another article coming for the pitchers as well, but in the offseason I will be thinking about ways to grade predictions solely against the result rather than against other participants. Think of it like golf, competing against the course. If anyone has any input that they'd like to offer as a means to accomplish that, feel free to comment below or reach out to me personally.