Projecting the Second Half

by battlekow on Jul 17, 2008 5:28 PM CDT in Statistical Analysis Comment 14 comments

Sal Baxamusa at the Hardball Times recently published a very useful tool: downloadable Excel spreadsheets that produce Marcel projections for players. Marcels are basically weighted averages and are surprisingly competitive with advanced projection systems like PECOTA or ZiPS. For our purposes--predicting the most likely immediate future performance to evaluate, for instance, whether we should get a new second baseman--Marcel will do just fine. So, let's see what we've got:

Jason Kendall

Period	AVG	OBP	SLG
Current	.258	.339	.333
Balance	.259	.330	.329
Total	.258	.334	.331

Prince Fielder

Period	AVG	OBP	SLG
Current	.270	.357	.488
Balance	.282	.374	.531
Total	.275	.364	.507

Rickie Weeks

Period	AVG	OBP	SLG
Current	.217	.320	.367
Balance	.247	.357	.409
Total	.231	.335	.386

Bill Hall

Period	AVG	OBP	SLG
Current	.234	.294	.431
Balance	.260	.326	.465
Total	.246	.308	.446

Russell Branyan

Period	AVG	OBP	SLG
Current	.245	.331	.604
Balance	.236	.332	.469
Total	.239	.332	.515

J.J. Hardy

Period	AVG	OBP	SLG
Current	.283	.352	.476
Balance	.273	.335	.448
Total	.279	.343	.463

Ryan Braun

Period	AVG	OBP	SLG
Current	.286	.324	.549
Balance	.303	.354	.567
Total	.293	.337	.556

Mike Cameron

Period	AVG	OBP	SLG
Current	.231	.320	.481
Balance	.246	.329	.443
Total	.240	.325	.461

Gabe Kapler

Period	AVG	OBP	SLG
Current	.315	.352	.517
Balance	.272	.332	.425
Total	.294	.341	.471

Corey Hart

Period	AVG	OBP	SLG
Current	.289	.327	.504
Balance	.285	.340	.496
Total	.287	.331	.501

Ben Sheets

Period	ERA	IP	H	K	BB	WHIP
Current	2.85	123	109	108	28	1.11
Balance	3.47	87	81	71	22	1.18
Total	3.11	210	190	179	50	1.14

CC Sabathia

Period	ERA	IP	H	K	BB	WHIP
Current	3.67	137.3	130	137	40	124
Balance	3.21	98	94	86	25	1.22
Total	3.48	235	224	223	65	1.23

Jeff Suppan

Period	ERA	IP	H	K	BB	WHIP
Current	4.71	101.3	118	55	44	1.60
Balance	4.44	72	82	42	27	1.52
Total	4.60	173	200	97	71	1.57

Manny Parra

Period	ERA	IP	H	K	BB	WHIP
Current	3.78	100	99	78	50	1.49
Balance	3.98	71	70	57	31	1.42
Total	3.86	171	169	135	81	1.46

Dave Bush

Period	ERA	IP	H	K	BB	WHIP
Current	4.39	106.7	97	66	26	1.15
Balance	4.19	76	76	52	19	1.25
Total	4.30	182	173	118	45	1.19

Son of Clung

Period	ERA	IP	H	K	BB	WHIP
Current	4.16	75.7	66	64	36	1.35
Balance	4.53	54	51	40	24	1.40
Total	4.32	129	117	104	60	1.37

4 recs | Comment 14 comments

Story-email Email Printer Print

Comments

Great work

thanks for much for doing this for all of our guys.

One note: This isn’t a criticism of Marcel, since it is what it is, but the one thing that most differentiates it from other systems (to me, anyway) is that it doesn’t use any minor league data. For someone with less than 3 full years in the majors, it just regresses to the mean more. In a system like mine (or ZiPS, or PECOTA) I don’t think Parra would look as good, because while his MILB results were good, they wouldn’t translate to the equivalent of an average major leaguer. Parra would still fare just fine…but it’s a caveat worth considering for anybody who has spent a good portion of the last three years not in the majors.

Also, cheese.

by Jeff Sackmann on Jul 17, 2008 6:28 PM CDT reply actions 0 recs

awesome, this should be a required post for all team blogs. thanks.

by Sky Kalkman on Jul 17, 2008 7:05 PM CDT reply actions 0 recs

maybe you saw this, but

Jessica did it for the Mets, too.

At some point-probably not until next season, sadly-I think I can build this into minorleaguesplits, so you can see a current projection for every player every day of the season.

Also, cheese.

by Jeff Sackmann on Jul 17, 2008 7:07 PM CDT up reply actions 0 recs

cool, thanks.

David Appelman’s working on figuring out how to code a daily updated Marcel forecast into Fangraphs, too.

by Sky Kalkman on Jul 17, 2008 7:20 PM CDT up reply actions 0 recs

with the data presentation he has...

it should be really easy. it actually wouldn’t be that hard for me at MLS, either, but it would require harnessing a bunch of data I don’t yet have in the same place. and it would require a bunch of my time, which isn’t exactly available in abundance.

Also, cheese.

by Jeff Sackmann on Jul 17, 2008 7:44 PM CDT up reply actions 0 recs

Nice work

As a side note/statistical anomaly, Fielder this year has 15 HRs at home and 3 on the road.

And neck size to baby eating ratio.

by Jordan M on Jul 17, 2008 11:42 PM CDT reply actions 0 recs

You know what goes well with offensive projections?

Defensive projections. It’s simply a weighted average of 2005-2008 zone rating data, with a little regression to the mean thrown in. Plays and Runs are presuming a full season’s of chances, not the chances published per player.

I has a blog. | Ask a sabermagician!

by cwyers on Jul 18, 2008 12:37 AM CDT reply actions 1 recs

Thanks Colin

I’m going to use Craig Counsell as a proxy for Alcides Escobar here, simply to illustrate how much swapping out Escobar for Hardy would improve the team defensively.

Counsell, according to Colin’s ratings, would make 6.85 extra plays per 161 at-bats, or 13.87 extra plays per Hardy’s 326 chances. Over those same 326 chances, Hardy would make 8.51 less plays than average. The run value of those eight and and a half lost plays is (negative) 6.41, while Counsell’s almost fourteen extra plays net 10.44 runs—that’s a difference of almost 17 runs between the two players, meaning Whitefish Bay’s finest has a glove almost two wins better than James Jerry.

BCB's "very own marginally deserving all-star!"

by battlekow on Jul 18, 2008 5:40 PM CDT up reply actions 0 recs

Runs are prorated out to a full season's playing time...

...on the spreadsheet, not to the chances listed. (Chances are balls in zone – in this case they’re just an artifact of the weighted averages, and not a projection of playing time.)

Really, looking at the chart, Weeks is the guy who you’d replace with Counsell. I’d have to check, but can’t Counsell hit as well as Weeks has this year?

I has a blog. | Ask a sabermagician!

by cwyers on Jul 18, 2008 10:17 PM CDT up reply actions 0 recs

Ah, thanks for the correction

Either way, the difference is large between Hardy and Counsell, who I am assuming is a fairly good defensive comp for Escobar.

BCB's "very own marginally deserving all-star!"

by battlekow on Jul 18, 2008 10:28 PM CDT up reply actions 0 recs

Rickie

.247 batting with an OBP of .357

I hope so. Given Marcel’s stated accuracy, maybe yes, maybe no.

by ol Pete on Jul 18, 2008 10:19 AM CDT reply actions 0 recs

Haha, awesome

Battlekow at Brew Crew Ball put together a projection of each player on the Brewers using the Marcel system. It’s a fine system to use if one expects the players to perform exactly how they did in the first half, but it does not take trends into example at all. What about Dave Bush’s sub-3.00 ERA since May? Nope. Rickie Weeks’ career OBP above .360 after July? Nope. Manny Parra gaining more and more confidence with every start? Not a chance.

Who wants to take a guess on which is more predictive: Manny Parra’s pitching line over the last two years, or a blogger’s impression of his confidence? Anyone? Bueller?

BCB's "very own marginally deserving all-star!"

by battlekow on Jul 18, 2008 5:44 PM CDT reply actions 0 recs

hilarious

“It’s a fine system to use if one expects the players to perform exactly how they did in the first half, but it does not take trends into example at all.”

One, the point is that Marcel most certainly does not expect players to perform exactly how they did in the first half. Two, ignoring trends is a good thing. A GREAT thing.

by Sky Kalkman on Jul 19, 2008 4:19 PM CDT up reply actions 0 recs

Marcel does not ignore trends.

Marcel considers this year’s data more to be more important than last year’s.

I’ve done day-by-day Marcels in the past (search my archive on THT), it generally turns out the same as the year-by-year Marcels.

Children, until we have taught them better, will be perfectly happy with a seasonal round of games in which conkers succeeds hopscotch.

by salb918 on Jul 22, 2008 9:16 AM CDT up reply actions 0 recs

Comments For This Post Are Closed

Search

FanPosts

Community blog posts and discussion.

Recent FanPosts

+ New FanPost All FanPosts >

NL Central Standings

	W	L	PCT	GB	STRK
St. Louis Viva El Birdos	91	71	.561	0	Lost 6
Chicago Bleed Cubbie Blue	83	78	.515	7.5	Lost 1
Milwaukee Brew Crew Ball	80	82	.493	11	Won 3
Cincinnati Red Reporter	78	84	.481	13	Won 2
Houston The Crawfish Boxes	74	88	.456	17	Lost 3
Pittsburgh Bucs Dugout	62	99	.385	28.5	Lost 2