David Fung did a good story on 1/30 about the A's projected wins for 2012 using two different projection systems: MARCEL and Clay Davenport. MARCEL is a very basic projection system that weights past seasons in a 5/4/3 fashion, finds a players' league average, and uses a age adjustment to come up with a projection about player future performance. A basic level, this is really what all projection systems do: use some combination of past performance and age to attempt to find the next year's performance. Davenport, BP's PECOTA, RotoChamp, CAIRO, ZiPS, and Bill James are other projections you might have heard of.
I'm going to focus on ZiPS here because a) its data are readily available and complete and b) because I think it is by far the most pessimistic
And here they are:
What I did with that was put it into Excel and scrapped the guys who aren't on the 40 man roster. I know Dan Zymborski (the curator of the ZiPS model) errs on the side of more data, but it's really not helpful for me to see Grant Green or Stephen Parker on this list. So, they're gone.
What I'm going to try and do is try to point out where ZiPS might be too pessimistic on the A's. Let's start with the hitters:
The first thing to point out is how terrible ZiPS projects the offense as a whole. If sorted by OPS+, which is a measure of OPS (on-base + slugging) weighted to the league average (100 being average). Yes, your eyes aren't fooling you -- ZiPS does not project any one A's player to have a league average bat. What about the guys who are close, though?
Crisp is projected for 104 games. While certainly 2010 showed that Coco can be a fragile player, he also played 136 games last year while battling some nagging injuries. Barring unforeseen injury, I expect to see closer to 136 games from Coco in 2012
The same applies for Jemile Weeks, who is projected for 126 games. Again, while his MiLB history is riddled with injury, he was able to remain relatively healthy despite losing his helmet nearly every game. I expect closer to 145 games for Weeks in 2012.
These projections cannot take into account that there will likely be a winner of the 1B sweepstakes. I expect that guy to be Daric Barton over Brandon Allen, with Carter getting an occasional spot start over there. So, that's more ABs of a projected 94 OPS+ hitter over Allen (85).
Scott Sizemore's K% is projected to be somewhere around 26%. That was indeed around his K% last year, but it was also the highest of his professional career to date. This could go either way, of course, but as he gets used to the league and a new position, I would expect it to go down closer to 20% or so.
Starting pitching:
As expected, these projections are considerably more rosy than the hitters' projections.
Again, I think the weighting of prior season's data hurts the A's when the actual prospects look better. ZiPS puts Dallas Braden and Brandon McCarthy down for 21 and 16 starts respectively. Braden, most recently, has said he feels great after his surgery. McCarthy has said that with the A's, he's discovered a new bone doctor and has new techniques in his delivery to minimize the load to his reactive shoulder. Also, while Colon will be entering his age 39 season, last year he posted one of his best K/BB ratios of his career (only one better was in 2005 with the Angels). While his HR/9 ratio was a little higher than his career average of 1.11, he is also going to be starting half his games in Oakland. The more I think about it, the more I think he has one serviceable year left in his arm.
Most projection systems, including ZiPS, also produce defensive and reliever projections. The one thing that is projectable about these statistical groups is the volatility. Defense notoriously needs somewhere around 3 seasons of data to normalize. What's more, with Sizemore and Weeks both presumably improving their defensive prowess from last year, and a mixed bag of results from Barton, the projections become extremely variable. With only a couple truly established relievers on the A's, it also makes those projections extremely difficult to have any confidence in. So, I choose to ignore them.
In terms of where I think the A's will end up, I think close to 70 wins is likely. This is a little more optimistic than Dan Z., but much less so than PECOTA, which projects us closer to a .500 team.
No matter how many wins we will have though, I'm ready for real baseball. How about you?
1 recs | 28 comments
What stands out to me is that at .408,
Seth Smith is the only hitter projected to slug over .400. That would be…sad. And yes, I’m ready for baseball!
Nico - February 12, 2012
bad projections
Have to believe Smith, Sizemore, and Suzuki will have better years than projected.
coachmmm - February 12, 2012
Yes, everyone has a 50-50 chance of having a better year than his projection.
WaddellCanseco - February 12, 2012
Hmm
First of all, I find it interesting that they predict for Carter – if he wins the starting job – to hit 20 homers but bat .224 with 170 K’s. I envision around 15 homers, a .250 average, and 100 strikeouts, but that’s just me.
I see they predict a large drop in Weeks’ performance. I would put him more around at .290-.300, and while injuries hit him hard in the minors, I’ll give him between 140 and 150 games.
I also think Sizemore, Smith and Reddick, especially Reddick. i see potential in him and figure to see him hit around .280.
McCarthy and Braden will both do better then that, and I think McCarthy could emerge to be a bit of workhorse in 2012 (and he sorta was in 2011). As for the rookies, the ERA’s may be a bit lower.
Lastly, I highly doubt that Balfour will be so much worse after coming off two fantastic seasons. I’d give him 2.88/62.1 IP/73 K’s.
Sean Fortuna - February 12, 2012
I don't see how you can predict McCarthy as a "workhorse" with his injury history even thru 2011
WaddellCanseco - February 12, 2012
Well,
Perhaps I should’ve phrased it differently. McCarthy isn’t really a workhorse, but I saw him as a reliable guy who could go out and give you 7,8 maybe even 9 strong innings. He just had this solid “workhorse” feel to him.
But you’re right, it was the wrong term.
Sean Fortuna - February 12, 2012
So you meant within a game? OK I can see that. He's pretty economical with his pitches.
WaddellCanseco - February 12, 2012
100 strikeouts in a full season of Chris Carter is [Rob Lowe]LITERALLY the most optimistic prediction I’ve ever heard.[/Rob Lowe]
doctawojo - February 12, 2012
Well hello Ann Perkins!
Nico - February 12, 2012
I wouldn't take these projections too seriously
For most players, either they learn to hit MLB pitching, or they don’t. You won’t get someone like Chris Carter hitting .220 with 20 HR, for example. From what I’ve seen, it’s close to impossible to predict what a player will do in his first couple of years if he’s still learning the ropes. The same is mostly true of pitchers as well. Trying to predict, for example, what Anderson or Braden will do, given their return from injury, is impossible.
The Coco Crisp projection is probably pretty accurate as is the projection for Kurt Suzuki, but that’s because they’ve been around a while and unless they fall off a cliff or have a career year, this is what they’ll do.
But most of the rest
-forget it. And, of course, for relievers, the ability to predict a good season for any reliever with previous inconsistency is impossible, and most relievers are incredibly inconsistent year to year.The thing is: we can look at this team and realize that it’s extremely unlikely that more than a couple of position players will fully make the transition to the next level, whatever it is, and that’s why this team is going to suck.
But most of this looks to me like statistical masturbation.
richwol1 - February 12, 2012
This isn't really true. The projection systems are about as useful/useless for full time 1st year players as for anyone else
Where they are sometimes wildly off is when a first year player gets sporadic playing time, but that’s true for anyone as well. Even a 7th year player who gets sporadic playing time is likely to be worse than his projection.
Does any statistical analysis not look like that to you?
WaddellCanseco - February 12, 2012
No, sometimes analysis does tell you something
I don’t think this projection tells us anything though. The only way Chris Carter or Kia K. perform as expected here is if they come out with a bang and then fall apart. Otherwise they won’t get the at bats. You can predict the Yankees pretty well with these systems, and you can predict the Phillies because you’ve got a sufficient sample size to draw from. But here, they have hardly anything at all.
This winds up as guesswork.
richwol1 - February 12, 2012
Well sure, the projection is for what the player would get if they got full-time playing time.
It’s not a projection of how much playing time they will get.
WaddellCanseco - February 12, 2012
Yeah, but it's meaningless then
And on top of that, it’s pointless to say that Kila will hit 16 HR and bat just over .200 because, as I said, the only way that happens is if he starts strong and fades. Either of us could say, “Kila’s a long-shot” and we’ve said more than ZIPS.
richwol1 - February 12, 2012
Actually, Carter's projections seem just about the most likely on the team
His stats would similar to Cust’s early lines. Lots of HR, Lots of K’s, low BA. Some of the other projections seem worse than pessimistic. Weeks with a .267? REALLY?
Furyan - February 12, 2012
I wouldn't call it statistical masturbation
These are projections, not predictions. They’re not meant to hit any one player’s performance on the nose. Some of these young players will break out and vastly overperform the projections, while others will completely crash and burn. The goal is to get an overall picture of how good the team will be. That’s the use I see in this – setting a reasonable expectation level for the season.
I’d go further on relievers. I can’t think of a single one besides Mariano Rivera that is not incredibly inconsistent from year to year. That doesn’t mean we should ignore the projections as a whole.
laserbeams - February 13, 2012
ZiPS has this team as truly terrible. Not only are all the hitters bad, but if McCarthy
gets hurt, as he often does, then Tom Milone is the ace of the staff. Yes, Tom Milone.
Forget 70 wins, this team might not even win 60. They seem to be better than the worst recent team, the 2003 Tigers who had a collective 78 wRC+ and a 111 xFIP-, and won 43 games, but a sub-50 win season isn’t out of reach if they get unlucky.
WaddellCanseco - February 12, 2012
Of course ZIPS says this...
So few of these guys have track records, and the ones that do have track records that suck at the major league level, albeit with too small a sample size to be meaningful.
Yeah, this team will suck because there are so few high-rated prospects on it, and so few veterans with great track records. But you knew that going in.
richwol1 - February 12, 2012
So you're saying ZiPS is just confirming what we already thought? OK, that's generally
But there are sometimes surprises. For example I was surprised how well it predicted McCarthy to do last year given how crappy he had been before that in the majors, and lo and behold he was pretty good when not hurt.You’re really not interested in MIlone’s decent projection in light of Goldstein’s conclusion that he’s not one of the A’s Top 20 prospects? Did you really think the offense was THIS bad? It’s not all confirmation of pre-existing beliefs.
WaddellCanseco - February 12, 2012
Of course the offense could be this bad...
Most likely it won’t be, because most likely Beane would make moves to ensure it isn’t. But when you take a bunch of players and throw them to the wall and see if a couple stick, and ZIPS projects that none of them will stick because a statistical analysis won’t show you surprise successes, of course it’s going to look this bad.
richwol1 - February 12, 2012
I disagree that it's unlikely the offense will be this bad. It's 50-50 if you ask me.
WaddellCanseco - February 12, 2012
Actually, when I Monte Carlo it out all the percentiles and with estimated playign time, ZiPS has the A’s at 69 wins, on average. They have a lot of lousy hitters, but there’s actually pretty good depth of them and the pitching is good enough to keep from the downside being too horrific. I only had them with a 20%ish chance (don’t have the data in front of me) to lose 100 games.
D.Szymborski - February 12, 2012
Yay! We're going to win 69 games! Hopefully those will be the 69 that I watch.
WaddellCanseco - February 12, 2012
BTW Dan, how often does ZiPS project fewer than 69 wins? It can't be that often.
WaddellCanseco - February 12, 2012
Is this a joke?
under 60 wins? a chance to even miss out on 50 wins?
That is ridiculous. Even wihout McCarthy we have enough pitching depth to field an average pitching staff in the Coli. So you think these hitters are 30 wins worse than average? With a lot of players at least over 90 OPS+ and above average defense. It would be impossible to be half that bad.
Rio - February 13, 2012
I agree with you Alan, with 70 as a win total
Do you think ZiPS is pessimistic in general?
David Fung - February 12, 2012
How can it be pessimistic for everyone? It has to be optimistic for half and pessimistic for half, no?
WaddellCanseco - February 12, 2012
FWIW
(Apologies for anyone who follows me on Twitter, since this will be repetitive.)
PECOTA has the following hitters coming in above league average (overall, not for their positions): Suzuki, Barton, Allen, Carter, Sizemore, Smith, and Gomes. Reddick’s PECOTA is basically league-average.
The guys who are projected to be at or above the 2011 average for their position: Suzuki, Sizemore, Smith, Gomes (as LF), Carter (as DH).
In other words, PECOTA has a much rosier picture of the A’s offense than does ZiPS.
doctawojo - February 12, 2012
You must Login with your SB Nation account and be a member of Athletics Nation to post a comment.