Jake Roth-US PRESSWIRE

Projection Systems Part I – The Offense


With all the major projection systems now published, it’s worth taking a look at how they feel about a lot of the names we’ve been throwing around the site here recently. But before we get into the Ms, it’s important to know how these systems work.

Rick Scuteri-US PRESSWIRE

Marcel is the most basic projection system out there, besides Eric Wedge’s gut instinct. Its 2012 predictions came out on Fangraphs last week, but the easiest thing to predict was all the hubbub about its pessimism. Why are only two players predicted to top 30 dingers when 20 players did just that last season? Why is no one expected to top 600 ABs when—to name a familiar face—Ichiro has done so for 11 consecutive seasons? These are valid questions, and highlight both strengths and weaknesses to this simple crystal ball.

Consider a group of 25 players. We’ll call them a team, a baseball team. We know there is a pretty good chance that at least one of them will miss significant time in 2012 due to injury (things happen…like IBS). Instead of attempting to guess which player is going to go down, Marcel just projects fewer ABs for every starter. What this means is that Marcel is likely to massively over-predict for the guy who tears his ACL, and slightly under-predict the other 24 players. Its strength is strength in numbers. It hits the averages pretty well.

The weakness is then obviously its inability to properly forecast the players that don’t get injured…right? Actually, Marcel—as simple as it is—has forecasted just about as well as many other projection systems that use a lot more inputs. One phenomenon that many people underestimate is that of regression to the mean. Marcel uses not only the past three years of data to project a player’s future, but also regresses that estimate slightly toward the league average. How is that fair, you ask? For whatever reason—and I have my own opinions on this—players tend back toward the league average each season, whether that be back up or back down.

So Marcel’s pessimism isn’t actually quite as pessimistic as we thought. In fact, it’s just as pessimistic for yesteryear’s great players as it is optimistic for previous busts. But its strength still lies in projecting for teams rather than individuals, in my opinion. So how do we look out there?

Marcel M’s 2011 M’s 2012 Angels Rangers Athletics
AVG 0.233 0.255 0.262 0.274 0.255
OBP 0.292 0.317 0.329 0.338 0.325
Slugging 0.348 0.390 0.420 0.438 0.399
wOBA 0.284 0.311 0.327 0.338 0.318
Runs 556 670 749 802 705

 

Clearly, we’re not going to be competing offensively with the Angels or the Rangers, but we didn’t need a projection system for such astute analysis. The key here is the positive regression expected of the Mariners as a team, and that doesn’t even account for a whole season of Jesus Montero! Again, just as players that are performing far better than expected come back down to earth, the opposite is true for players hitting very badly (see: Mariners).

As for the individual players’ forecasts, I’m going to combine each of the major projection systems. The players have been divided into “easy-to-project” and “hard-to-project” based on the deviation of their projections (in batting points), and I’ve recorded their average wOBA from Marcel, ZiPS, RotoChamp, Bill James, Steamer, and the FANS projections.

“Easy” Deviation wOBA
Brendan Ryan 4.17 0.285
John Jaso 4.67 0.317
Mike Carp 5.33 0.335
Franklin Gutierrez 5.67 0.297
Chone Figgins 6.00 0.293
“Hard” Deviation wOBA
Ichiro Suzuki 7.33 0.312
Jesus Montero 7.89 0.350
Miguel Olivo 9.00 0.288
Dustin Ackley 9.00 0.339
Kyle Seager 9.17 0.321
Justin Smoak 10.00 0.327
Michael Saunders 10.11 0.285
Casper Wells 10.78 0.326
Alex Liddi 16.32 0.312

 

First things first: Can we please start Jaso??? Jaso had a double-digit walk rate in the minors, and has maintained that in his 687 PAs in show, and look! He’s projected to hit 29 wOBA points better than Olivo. Additionally, new pitch-blocking data shows Olivo to be well below average. In fact, he’s the worst since the stat became a real thing.

Overall, young players tend to be the hardest to project. Saunders, Wells and Liddi had the most diverse projections, and they have combined for only a thousand PAs total. Other than Ichiro and Olivo, the “hard” ones to predict are pretty much first and second-year players (or the equivalent in terms of PAs).

In other words, this is a high variance team since so many young guys are going to be in and out of the lineup. With even two or three outperforming their median expectations, things could be a little more interesting this year come September, especially with an extra playoff spot up for grabs!

Pitcher projections to come next week…

Tags: 2012 Featured Marcel Mariners Popular Projections Regression