2007 ZiPS Brewers Projections

[Two Fisted Slobber]

While Baseball Think Factory is having issues with their site, you can still reach the Brewers' ZiPS projections here

 

ZiPS believes in Sheets, but doesn't show much love for anyone else.

[Two Fisted Slobber]

For reference's sake, the 2006 ZiPS Brewers' projections can be found here.

[RyDogg66]

Ha. The Fools at the Think Factory clearly do not know that Capuano is the Ace of the staff. Someone tell them what is up.

[endaround]

Fielders' numbers aren't bad. Well except for the part where his .280 BA leads the team.

 

ZIPS for our big haul:

Estrada: .277/.323/.404

Vargas ERA 5.03

Aquino ERA 4.98

 

That said I don't think the program handles platoons well making Koskie and Cirillo's numbers look real bad.

[nsimps]

The 2007 Brewers will hit like the wind .... Fielder (113), Hall (145), Weeks (120), Jenkins (130), Hart (99), and Braun (101).

[Scooter1027]

While interesting, these are just educated guesses like anybody else. Look at what they had projected in 2006 for cats like Bill Hall, Geoff Jenkins, and Jeff Cirillo to see my point.

[RedbirdRay]

I've been working on compiling the Brewer Bill James 2007 Projections for you guys but always mess up the tables. Will probably be finished with it by tomorrow. Does someone want to volunteer to post it for me pretty like if I message them to you?

[sbrylski]

Aren't ZiPS projections usually pretty inaccurate anyway?

[pogokat]

if those numbers were spot on, i think the brewers would be .500

 

though i think those are pretty pessimistic...Carlos V's line looks pretty good for a rookie...and Vargas matches up with Davis very well...

[RedbirdRay]

I think ZiPS usually has a pretty decent track record.

 

I thought, in general, it usually went...

 

ZiPS

PECOTA

PrOPS

Bill James

 

in order of accuracy, but not sure. All have a pretty good track record in general. There are, of course, anomalies in any system...but it's based on history and age progression.

[SoCalBrewfan]

ZiPS are decent, as much as any projection system can be. My sense is that they are way off on minor league pitchers a lot of the time...that's not terribly surprising I guess. But Hendrickson, for one example we know and love, did not do anything close to his ZiPS projections.

 

They do make certain assumptions, particularly with respect to BABIP. The pitching projections this year are actually pretty nice for our guys...I don't know what they're figuring for the Brewers defense but last year's was not so hot...

 

Meanwhile, I'd take the over on Hall, the under on Mench.

[Ennder]

ZiPs are terrible, especially for pitching, they are nowhere near the best system.

 

Also keep in mind projection systems always hedge their bets, thats why the playtime is always low for everyone.

[Two Fisted Slobber]

ZiPS is supposed to be more of a quick-and-dirty method than other projection methods like PECOTA. Kind of like how OPS isn't as accurate as something like Runs Created or Linear Weights, but it still does a pretty good job.

[SoCalBrewfan]

Do you have a reference for that? I haven't seen anything suggesting that ZiPS is terrible, though I freely admit that I might have missed something. The comparisons I'd seen had ZiPS in the middle of the pack...though IIRC Marcel was about as good as anything fancier, for those players with a major league track record at least.

 

Most systems have trouble with pitching, from everything I've seen.

 

ZiPS freely admits that it doesn't really attempt to project playing time, so I'm not sure what you mean by the second statement.

[TooLiveBrew]

glad to know I'm not the only one that thinks those numbers are straight garbage!

 

You'd do better merely speculating, and I can't put any stock in those numbers. Is there a formula behind this? I seriously can't stop thinking how poor those projections are (the 2006 numbers prove the infrequency of accurate projections - they got Fielder's about right, so I'm sure that makes them look better)

[saget]

Bush's #'s seem quite delicious. I'd take those any day

[1992casey]

I've gotten used to these things. Because a computer does it and it applies the same criteria to everyone, I'll simply say they make sense and I don't see any surprises.

[rluzinski]

I thought ZiPS was just a weighted average of the last few years, regressed and with age and park factors. Similar to Tangotigers "marcel" projections. It's obviously going to work better with vets and with guys who play full time.

 

No one has a crystal ball. This is just an educated guess.

[1992casey]

That's the way I've seen it, Russ. We're looking at approximately three years' performance and age.

 

Bill Hall, for instance, gets adjusted downward because of a crummy 2004. Fielder gets adjusted upward because of his age and his minor league numbers. Rookies are often projected for a bunch of playing time because they were full-timers in the minors.

 

The real key is to understand how the system works, then make your subjective adjustments to account for the things you can't ask a computer to do. I'd give Bill Hall more credit for the last two seasons while blowing off 2004. With rookies, you have to adjust playing time downward and give the projection a bunch of breathing room to account for the inevitable small sample you'll be dealing with.

 

These things are tools. If you know how to use them (as in not pounding in a nail with the handle end of a screwdriver), you should be fine.

[SoCalBrewfan]

Yeah, obviously you need to take things with a grain of salt, and recognize the limitations of the systems. Projection systems are ill-equipped to identify something like the gain Bill Hall made two years ago. Still, all told they tend to be reasonably effective, at least for hitters.

 

Here's a discussion of projections, including ZiPS, at THT...part 5 has links to the previous 4:

 

www.hardballtimes.com/mai...rt-1-of-5/

www.hardballtimes.com/mai...le-part-5/

 

Tango says in part 1:

 

My pre-initial statement was that the Marcels gives you an regression coefficient ® of .65, and the maximum you can reach is .75. The sophisticated engines get you to .70. I don't think anyone refutes that. (All for hitters.)

 

So it's not exactly throwing darts...and there's many baseball transactions that seem to defy the premise that the progression of a player's talent is at least somewhat predictable, particularly when you speak of established veteran hitters.

 

The nice thing about ZiPS and Marcel is that they made available for free, and not a lot worse than the systems that aren't.

[snyttri]

Since Estrada projected to .294 in AZ and .277 here, I guess that means that the 6 million dollar man who projected to .251 in AZ projects to .234 here.

 

tinyurl.com/yh98v6

[1992casey]

I see snyttri's link is to the D'backs projections.

 

What's too bad is that 'the trade' occurred between the dates of the two BTF links in this thread. I want to see Doug Davis, who isn't on either one of them.

 

Is there a master ZiPs link?

[rluzinski]

One other thing I'd like to point out about ZiPS or really any projection system. I always hear how a certain projection system doesn't project pitchers very well. I would counter that, if you a projected ERA is your aim, all projection systems are destined to fail. ERA is affected by many things more or less outside the control of the pitcher. Defense, of course, if a huge part of that equation. Just a small difference in BABIP (batting average of balls in play) has a significant change in ERA. Second, WHEN it's occur can have a huge effect on ERA. Many pitcher's LOB% (left on base %) jump all over the place from year to year, and so does their ERA. Finally, trying to project relievers from year to year is generally a waste of time. If a guy has less than 80 IP, luck plays a huge part in his final ERA numbers.

 

Just a nice early morning ramble to start the day http://forum.brewerfan.net/images/smilies/smile.gif

[RedbirdRay]

Well, the Cardinal ZiPS are up. I didn't want to create a new thread, because I realize most people probably don't care much about the Redbird projections. But, it's winter, I'm bored, and I figure a few of you do.

 

The coolest thing about this year's StL ZiPS, IMO, is discussion prompted Dan Szymborski to create a reliever-to-starter projection tool. He mentions it in the comments, then even goes on to give it out for free. Dan is the man! What prompted it was a request to project Wainright as a starter. Near the end of the comments, Dan projects Papelbon also.

 

I noticed after the Suppan signing, someone linked one of my favorite blogs to visit. There's a lot of saber-talk at this site...of course, in reference to the Cardinals. There are a couple pretty cool write-ups analyzing the ZiPS if anyone's interested. Here's the offense discussion. And, here's the pitching.

 

The StL chapter of Sabr.org also had very similar postings. A cool part of the offensive analysis is an estimate of all NL Central teams runs per game.

[rluzinski]

That lineup tool he uses to calculate runs/game is just plain wrong. I think I wrote about it on my blog somewhere but it just doesn't work and I'm not the only one to say that.

 

I've had a hard time translating projections to a runs/game value. You have to estimate AB for every player, also making sure PA translates to OBP. I ended up writing a very crude simulator in excel but it's not exactly ready for prime time. Tnagotiger was talking about releasing one publically a couple months ago... not sure what happened to that.