.305 /.338 /.400

Scratchiro baffles yet another projection system.

Apparently the Yankees should make Shelley Duncan their DH.

Andy tracy? .804 OPS? Really?

I do like michael taylor's projection, though.

Apparently the Yankees should make Shelley Duncan their DH.

I was thinking Russell Branyan.

Wow. CHONE hates every Sox hitter except Pedroia.

Wow. CHONE hates every Sox hitter except Pedroia.

Yeah, just look at that Youkilis projection. Saying a 31-year-old career .292/.391/.487 hitter will hit .280/.381/.477 is just abject loathing, the sort of hatred that decades-long blood feuds are made of.

It's how the thirty years war started.

.305 /.338 /.400

Scratchiro baffles yet another projection system.

Bill James has him at .319, which seems more reasonable. Dunno what it is about Ichiro which makes people craft such poor projections (he did hit .302 one year).

Wow. CHONE hates every Sox hitter except Pedroia.

The only one I find curious is Drew. VMart's is because of '08 for sure.

Wow, those Royals projections really have me looking forward to the season.

Scratchiro baffles yet another projection system.

He's always baffled my system. I had about the same projection for him last year. I'm hoping to add an upgrade to the system that helps that a bit, using batted ball data and some formuli I've been working on, but haven't gathered all the data yet.

Any idea when you'll have 2009 WAR up?

Justin Upton: 9 RAA/150
Alex Avila: 8 RAA/150

Upton is a few months younger. He has outhit Avila by a country mile in each of the past two years (i.e. Avila's whole career). Moreover, Upton has spent that time in the major leagues, whereas Avila has split it between A and AA with just a brief big league stint at the end of this year, so presumably, Upton's performance is being regressed to a higher mean than Avila's. How can these two possibly project as equals or anything close to it?

And neither of these guys are particularly unique in terms of their statistical profile. I understand difficulties when it comes to projecting an outlier like Ichiro, but that's not what we're dealing with here. Strange.

Good question on #13. Could be an error. I'll have to look into it.

2009 WAR: Probably soon after retrosheet posts the 2009 PBP files.

boy, that Soriano projection just makes me all warm and happy.

Jeff Fiorentino, average MLB player?

Hitting: +2
Fielding: +5 in LF or -5 in CF (random guess)
Positional adjustment: -7.5 in LF or +2.5 in CF

I see that he hit .312/.387/.510 in 422 PA at AAA-Norfolk this year (age 26). As a free agent, he'll surely be cheap. What do people think? Can he keep up his production in the big leagues?

Nick Punto .242 Delmon Young .288. Seems like a remarkably generous system to me. Cuddyer 21 HR, 85 RBI. Oh, wait, Mauer .323.

I'd be curious to see a projection on Jesus Montero.

Has anyone studied the accuracy of the various projection systems in 2009? Thanks.

Edit: And many thanks for releasing these, AROM.

I second the Montero request as well. Was just going to post it actually. A top 5 prospect with time in AA should be included (in these free projections for which even ingrates like me or thankful for).

Danny

There were a couple threads over at Tango's blog. Iirc, CHONES were solid, but not as good as year's previous. PECOTA was very bad. Something like that.

Maybe check the forecast tagged posts.

UM, HI. MY NAME IS TOLAXOR!!!

WE NOW HAVE A FEW YEARS OF THESE BILL JAMES/ZIPS/PECOTA/CHONE/OTHERS PROJECTION SYSTEMS!!! I THINK THEY'RE NEAT!!!

I WENT A-GOOGLIN' AND I COULDN'T FIND AN ANSWER, SO LET ME ASK YOU FOLKS HERE MY BASEBALL FRIENDS:

HAS ANYONE BEEN TRACKING BRIER SCORES FOR THESE THINGS???!!! EVER???!!!!!1!

I THINK THAT WOULD BE NEAT, TOO!!!1!

The Orioles would like to re-sign Fiorentino, but I wouldn’t be surprised to see him look for an organization with weaker OF depth. A minor league deal with a spring training invite should get the job done.

Fio had a brief fling as a real prospect a few years back when he showed some power, so maybe he’s the next Jason Werth.

boy, that Soriano projection just makes me all warm and happy.

It kind of made me throw up in my mouth a bit.

HAS ANYONE BEEN TRACKING BRIER SCORES FOR THESE THINGS???!!! EVER???!!!!!

How would you define a "success" vs a "failure" (which you need to do in order to calculate a Brier score)? If AROM nails the OPS projection but the performance shape is different - if AROM projects .350/.500 OBP/SLG but the player actually does .375/.475 - is that a "success" or a "failure"?

-- MWE

JUST GIVING IT 30 SECONDS OF THOUGHT, THERE MIGHT BE A COUPLE OF WAYS OF APPROACHING OPS AND OTHER METRICS THAT ARE DERIVED VALUES -

1.) JUST SAY THAT IN A DERIVED VALUE WHERE WE ARE COMBINING TWO DISCREET MEASUREMENTS TO SUGGEST A QUALITATIVE ASSESSMENT - WE CAN ONLY USE THE PRIORS - THE MEANING IN THE POSTERIOR OUTCOME IS "TOO BROAD" TO BE INFORMATIVE. I PERSONALLY DON'T LIKE THIS BECAUSE IT SMELLS OF FREQUINTISM'S TENDENCY TO THROW AWAY MEANINGFUL DATA B/C IT DOESN'T MEET SOME SUBJ. NOTION OF "FIT".

2.) SUGGEST AN ACCEPTABLE RANGE OF VARIANCE (THESE ARE PROJECTIONS, NOT ENGINEERED VALUES, I KNOW, BUT THAT WOULD MEAN THE VARIANCE TOLERANCE WOULD BE BROAD, THAT'S ALL) WITHIN THE PRIORS - SO IF EXPECTED OPS WAS .850 BASED ON .350 AND .500 - WE MIGHT ACCEPT .375 AND .475, BUT NOT, SAY, .300/.550.

Thanks, Philly. For anyone else interested, I found a forecaster evaluation thread here. And, yeah, terrible year for PECOTA.

I personally am less interested in "which system did better/worse" as I am in "who did each system miss on, and why". If everyone is missing on the same players, that's one thing; if some systems project certain types of players better than they do others, that's quite another.

-- MWE

If everyone is missing on the same players, that's one thing; if some systems project certain types of players better than they do others, that's quite another.

everyone misses on Ichiro every year :-)

except Marcel the Monkey

That's not really instructive

I think one of the ideas behind Pecota was that by using comps to generate projections you would eliminate biases as to types of players - which fails WRT Ichiro because there are no good post WWII comps for him.

I also think there is a selection problem- it's hard to find a comp for Mark Reynolds, because in the past Mark Reynolds would not get a chance at all- or would be forced to radically alter his approach

Thanks for pointing out Avila and Upton. There were errors. When I update next, Justin should be +10 (minor error with the age on his 2006 batting line being wrong) and Avila will be -6 (big error excluded some of his minor league data). I must stress that these are works in progress. I'll probably repost these after I update the primary position data. On Upton, +10 still seems low to me but the system is working as intended here. I think it's the high strikeout rates that think there's a chance he follows a strong season with something that looks like the last 2 years BJ put up. That and he's played in good hitter environments, and the Runs+ column is park and league adjusted.

The CHONE projections didn't do so great in Tango's contest, but I don't know if that is more due to problems with rate stats or with projected playing time. If I participate again, I will try to combine my forecasts with his fans playing time projections. If I was trying to win that contest, I would probably just stop trying to project any minor leaguers other than the Wieters/Longoria types who you know will have an immediate big league impact. One of my goals is to estimate the current talent level of minor league players, so I have depth charts to evaluate what an organization should do (sign free agent or replace from within) and allow people to create simulation disks with minor league rosters.

Mike, interesting idea. If someone wants to evaluate rate stats perhaps we should be looking at absolute error of OBP + absolute error of SLG instead of just OPS error.

On Ichiro I've got a few problems. One is that regression based models aren't giving him enough credit for getting hits on balls in play. This I hope to fix a bit with batted ball data, but that has not been added to the model yet. I think a proper projection would have shown him as a .330 hitter over the years - so you'd miss equally if he hits .310 or .350

Another problem is that he has so far refused to age. If he were heading into his age 29 season with the exact same historical data, I'd project him at .317 instead of .305. Hitting for high average is a skill that is not supposed to age well, but here he's 36 and has shown no dropoff at all.

AROM, do your R/150 numbers include SB/CS? Thanks.

CHONES were solid

Then maybe he should wash them!

Those do not include SB/CS, just batting events.

AROM, you have Ryan Church still on the Mets.

So do the Mets?

Andy tracy? .804 OPS? Really?

Haven't you heard of the "35-year-old ex-punter" effect?

This is the year it kicks in for Andy.

So do the Mets?

Don't tell Francoeur that trade never happened - he'd be crushed. He's been trying to help the Mets this whole time, not knowing it was a hoax.

Those do not include SB/CS, just batting events.
Why on earth is that the case?

****

Anyhoo, so the downloadable spreadsheet is out - so I came up with the top 10 in baseball (pos. players and pitchers) for '10, by CHONE WAR.

Position players:
Mauer 72
Pujols 68
H-Ram 59
Longoria 52
Tulo 51
Utley 47
M-Cab 47
Sizemore 46
Petunia 46
Teixeira 46
(swapping total WAR for WAR/150 would cause Cabrera, Pedroia, and Tex and to drop out, whilst A-Rod, Zobrist, and Yunel enter)

Rotation guys:
Greinke 60
Lincecum 58
Halladay 57
Sabathia 56
Felix 50
Verlander 49
Haren 47
JavyV 46
Lester 45
Lee 43

For kicks, here's a replacement level squad:
C Mike Redmond
1B Paul McAnulty
2B Mark Bellhorn or Miguel Cairo
3B Angel Berroa
SS John McDonald or Alex Cora or Orlando Cabrera
LF F-Cat
CF Jason Repko
RF Jose Guillen
DH Mike Jacobs
SP Jason Schmidt or Ben Hendrickson or Lance Broadway or VanBenschoten
RP Jesse Chavez or Danys Baez or Justin Miller or Guardado or Fernando Rodney or Freddy Dolsi or Mike McDougal or Disco Hayes

Isn't that the lineup for the 2010 Royals?

Very quick-n-dirty projected standings, using top 10 WAR hitters, top 6 WAR starters, and top 7 WAR relievers and replacement levels that get both leagues to ~81 wins:

NYA 98
BOS 95
TBA 89
BAL 77
TOR 71

MIN 83
CHA 81
KCA 79
CLE 78
DET 73

LAA 80
TEX 79
SEA 78
OAK 77

ATL 91
PHI 89
NYN 78
FLO 78
WAS 75

SLN 86
CHN 79
MIL 79
CIN 82
HOU 74
PIT 74

COL 86
LAN 85
ARI 80
SFN 79
SDN 77

Rankings by total positive WAR for players aged under 27 years:

Team WAR
OAK 26.8
TBA 26.4
CLE 23.3
BAL 22.5
COL 21.6
CIN 21.4
SFN 21.3
BOS 21.0
KCA 19.2
FLO 19.1
ARI 19.0
ATL 18.5
LAN 17.4
TEX 16.1
MIL 16.0
WAS 15.3
SEA 14.9
SDN 13.8
MIN 12.6
NYA 11.9
PIT 11.2
LAA 10.4
CHA 9.6
TOR 9.2
SLN 8.4
HOU 6.4
PHI 6.3
DET 6.0
NYN 5.0
CHN 1.3