Beyond the Box Score: An SB Nation Community

Navigation: Jump to content areas:



Sports blogs for fans, by fans.
New Blog: Backing the Pack for NC State Fans!


Help Me Hash Out WAR Spreadsheet Details

I'm hoping I can get some input from smart readers and bloggers who have played around with the WAR spreadsheet:

  1. Offensive outs targets.  Assuming that a team's wOBA is their OBP (a mediocre assumption), we know how many batting outs they make.  Not included are outs of other kinds and outs made by pitchers in the NL.  What would be a good way of measuring how many of hitting-only outs there should be per team?  My idea is to take the league-average non-pitcher PAs and wOBA (OBP, by definition) for NL and AL teams from 2008 and go from there.  Anything better?  Anyone wan to work on that?  Or is there a totally different, better way to make sure playing time estimates add up reasonably well?
  2. Team win baselines.  A .300 team will win 48 games.  AL teams are better than NL teams, on average.  But NL teams compete against each other, so they still average nearly 81 wins (maybe more like 80 because of interleague play).  So instead of a baseline of 48 wins, NL teams should have a baseline of 50-51 wins and AL teams should be at 45-46 wins, given 0 WAR.  Also, NL teams don't have a DH to add to their offensive WAR total, but this doesn't make them a worse team (significantly).  How does this effect the baseline?
  3. Hitter replacement level.  This one is tied in to #2 a bit.  On one level it doesn't really matter, since the team baseline can be adjusted to make teams average out to 81 wins.  Individually, I've used 2.0 WAR for NL position players and 2.5 WAR for AL position players, because it would be nice to also be able to compare projected player value across leagues once we have a community effort.  Some of you have decreased those individual numbers so that the team totals looks better (i.e. lower), but I'd urge you to adjust #2 instead.  Is that reasonable?  Any problems with the rep level of 2.25?  Any problems with weighting it 2.0 for NL and 2.5 for AL to make projecting raw stats easier?
  4. Pitcher WAR. The average position player is worth about 2.25 WAR over 700 PAs.  Average starters (with a 4.30 ERA and 180 IP) are coming out to 2.2 WAR (NL) and 2.6 WAR (AL).  I'd always thought the average pitcher was LESS valuables than the average position player.  Any thoughts?
  5. Labeling the ERA projection.  Some folks don't like using the label "ERA" because ERA is a bad stat.  What you really want to use is a fielding- and park-independent ERA projection (based on FIP or tRA or something like that).  Is there a better label to use or should I just better explain that part in the directions?

Also, if anyone doesn't have Excel and wants to use the spreadsheet, I recommend either OpenOffice (which contains an Excel-like application and is available for any operating system) or EditGrid (which is an online spreadsheet tool similar to GoogleDocs, but better.)

Updates: As I find good solutions to things, I'll list them here.

  1. [PAs by lineup slot from jhmoore:]

    1. 777.6
    2. 758.16
    3. 735.48
    4. 722.52
    5. 703.08
    6. 685.26
    7. 664.2
    8. 644.76
    9. 625.32

    Based on The Book’s section on lineups for NL… the AL one is negligibly higher (like 2% higher or something. Although in this run environment, lgOBP is some 10 to 15 points lower, you might want to subtract 10 from all of them or something.  [Sky agrees.  Also, teams with lower OBPs will have fewer PAs and teams with higher OBPs will have more PAs, although I don't think it's a huge difference.]

0 recs | Comment 16 comments

Story-email Email Printer Print

Comments

Display:

Offensive out targets

In 2008, based on my analysis of MLB play by play data, the total number of baserunning outs (caught stealing and other outs on the base paths) for a given team varied from 125 (Rays) to 68 (Padres). Average was 94 throughout the majors. Because this number is dependent on aggressiveness on the base paths, there would probably be some correlation from year to year for a team (although I have not confirmed this). Also, GIDP could be considered, average was 129 per team, ranging from 157 (White Sox) to 97 (Brewers).

To account for playing time, I considered adjusting for expected position in the lineup, as I’ve seen projections use about a 20 PA difference between spots in the order. That would have meant defining multiple potential lineups for each team, so I just used 700 PA for each position.

by Adam Peterson on Jan 9, 2009 2:15 PM EST reply reply actions actions   0 recs

Is 700 PAs per position even accurate?

I mean, is it more like 720? 740? 680?

What about NL teams, including PHs, but not pitchers?

Someone (I’ll get to it eventually) should look at team PA totals from 2008.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 9, 2009 2:26 PM EST to parent up reply reply actions actions   0 recs

Total plate appearances per team

according to ESPN was 6259 last year. Comes out to 695 per slot in the batting order. Using 700 is probably OK. Total PA per team in the NL was 6250, so about the same per slot.

As pitchers, averages were 23 and 349 PA in teh AL and NL, respectively.

by Adam Peterson on Jan 9, 2009 2:35 PM EST to parent up reply reply actions actions   0 recs

Excellent, thanks.

So 6260 is a good goal, more for high-OBP teams. For the NL, subtract 350 PAs from that.

Given a difference of about 20 PAs per spot in the lineup, you could do something like this:

760
740
720
700
680
660
640
620
600

Give or take. I’m not fully confident in that.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 9, 2009 2:47 PM EST to parent up reply reply actions actions   0 recs

Oh I am ALL over this one.

1. 777.6
2. 758.16
3. 735.48
4. 722.52
5. 703.08
6. 685.26
7. 664.2
8. 644.76
9. 625.32

Based on The Book’s section on lineups for NL… the AL one is negligibly higher (like 2% higher or something. Although in this run environment, lgOBP is some 10 to 15 points lower, you might want to subtract 10 from all of them or something.

---
Juuuust a bit outside!!
http://www.rightfieldbleachers.com

by jhmoore on Jan 9, 2009 4:02 PM EST to parent up reply reply actions actions   1 recs

Another issue:

How much does that vary based on team OBP? Will great offense create 10 more PAs per lineup slot? 40 more? I’m guessing it’s not that big of a difference.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 9, 2009 4:21 PM EST to parent up reply reply actions actions   0 recs

that's a good question

I guess we should just look at teams and how many PAs each team gets and how that correlates to OBP. The difference in PA/Game for each slot is very close to the .11 you’d expect (because there’s a 1 in 9 chance that the game ends at each spot in the order)

---
Juuuust a bit outside!!
http://www.rightfieldbleachers.com

by jhmoore on Jan 9, 2009 4:27 PM EST to parent up reply reply actions actions   0 recs

The Rockies were 4-10 PA's above the NL average last season

And they had a team OBP of .336 compared to .331 for the NL
The Rockies last season were:

1. 771/764
2. 756/746
3. 733/729
4. 718/712
5. 705/696
6. 686/678
7. 664/660
8. 649/643
9. 630/623

It averages out to exactly 6 more PA’s/lineup slot for that .005 difference in OBP.

by Rox Girl on Jan 9, 2009 5:43 PM EST to parent up reply reply actions actions   0 recs

Cool.

The Cubs had a .354 OBP and 6384 PAs. That’s 709 PA’s per lineup spot. The NL average, according to jhmoore’s data was 702. Huh.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 9, 2009 6:41 PM EST to parent up reply reply actions actions   0 recs

According to BB Ref, the NL average last season 694.45

Which would actually make the Cubs 15 over the average, at least according to the NL batting splits page.

by Rox Girl on Jan 9, 2009 6:58 PM EST to parent up reply reply actions actions   0 recs

Ok, that makes more sense.

jhmoore’s data was from The Book, and he even suggested subtracting ten-ish PAs from the 702 number.

So the Cubs were 15 over per slot and .023 points of OBP above average. Per .005 points of OBP, that’s 3.2 PAs per .005 points of OBP.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 9, 2009 7:35 PM EST to parent up reply reply actions actions   0 recs

Here's how I handle this.

If you really want to be precise – go ahead and actually set your lineup in order, top to bottom. Figure out the percentage of team PAs per lineup spot. Then figure OBP as a WEIGHTED average of the projected performance of each player, based upon their lineup spot. Use that OBP figure to figure team PAs based upon a set number of outs.

by cwyers on Jan 10, 2009 1:40 AM EST to parent up reply reply actions actions   0 recs

Baserunning outs will also vary based on team OBA

Along with the aggressiveness, so you’ll want to account for that as well.

by Dan Turkenkopf on Jan 9, 2009 10:04 PM EST to parent up reply reply actions actions   0 recs

Pitcher WAR

In 2008, the average ERA for a starting pitcher was 4.50 in the AL and 4.43 in the NL. If you calculate WAR using these ERA and 180 IP, what do you get? I also wonder if 180 IP is correct for the “average” SP…
 

by Adam Peterson on Jan 9, 2009 2:31 PM EST reply reply actions actions   0 recs

I'm using 4.30 as league-average ERA and not 4.50 (starter-average ERA)

The replacement-level winning percentages account for the role. Not that comparing 4.5 to 4.5 is any different from comparing 4.3 to 4.3.

You get 2.2 and 2.6 WAR depending on league.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 9, 2009 2:50 PM EST to parent up reply reply actions actions   0 recs

Comments For This Post Are Closed


User Tools

We use numbers and stuff.
Community Guidelines
Start posting on Beyond the Box Score »

Join SB Nation and dive into communities focused on all your favorite teams.

FanPosts

Community blog posts and discussion.

Recent FanPosts

P1_iannetta_small
Matt Herges: The reason we look at peripherals
Small
SLG and Speed
Small
Interleague Attendance Nonsense
Limes_125_small
A Note About Becoming a BtB Author: Contributing to the Community Helps
Small
Is Adrian Beltre underrated...?
Limes_125_small
How Do You Like the New Daily Link Roundup Posts?
Stlouiscardinals_small
Depth Charts Help
Zorilla_small
How Do You Measure a Pitching Coach?
Limes_125_small
Looking For SQL & Tech Geek Help For Collaborative Projects
Small
When do MLBers get paid...?

Post_icon New FanPost All FanPosts Carrot-mini

FanShots

Quick hits of video, photos, quotes, chats, links and lists that you find around the web.

Recent FanShots

There's your human element.  Why, when the technology is readily available, are humans still calling balls and strikes?
Fire Jim Leyland: Fu-Te Ni Follow Up; Concern Over Big Three?
Rany Gets Banned By the Royals
Contract Retrospective: Vernon Wells' 7-year, $126 Million Contract
The Rockets are innovative
Flip Flop Fly Ball
Yanks Considered Trading Rivera For Wells In '95: MLB Rumors - MLBTradeRumors.com
Bullpen Usuage Charts for Last 5 days
MiLB Game of the Week
Drive Mechanics Looks at Chris Perez's Mechanics and Pitch F/X numbers

Post_icon New FanShot All FanShots Carrot-mini

Most Commented

Subscribe to BtB via Email

Enter your email address:

Delivered by FeedBurner

BtB Goes Social

BtB on Facebook

BtB_Sky on Twitter


Managers

Mos-def-the-ecstatic_small R.J. Anderson

Limes_125_small Sky Kalkman

Editors

Rawlings_baseball_bigger_small Dan Turkenkopf

Face_small Harry Pavlidis

770insig_small Jeff Zimmerman (TucsonRoyal)

Rickstache_small erik

Authors

Jinaz-reds-avatar_small JinAZ

Hms_surprise_small Graham

Wisc19cropped2_small jhmoore

Raysring1_small Tommy Rancel

E52205a2_small tbsmkdn

Official Partner of Yahoo! Sports