Saber-Friendly Blogging 101: Post-dicting 2008

by Sky Kalkman on Jan 31, 2009 7:00 AM EST Comment 10 comments

This idea comes from Doghouse over at Federal Baseball and it's really quite simple if you buy into Wins Above Replacement. Basically, the idea is to plug 2008 data for your favorite team into the easy-to-use WAR Spreadsheet and see how closely the team's talent matches their actual win total. It's not a prediction, it's a "post-diction".

This analysis is a bit different from something like Pythagorean record, because we're not dealing with actual runs scored. On offense, for example, we're dealing with either wOBA or OBP/SLG plus baserunning. Teams may have been lucky or unlucky in the way their batting events combined to score runs. And, like with Pythag, we're removing the timing of runs scored and runs allowed, which is not much of a skill beyond properly leveraging a bullpen (and leverage is included in the spreadsheet).

If you go ahead and plug in 2008 data and write something comparing "post-jected" wins to actual wins, let me know and I'll record the numbers here and link to your article. (It also will be a good test to see if the baseline win total used in the spreadsheet is accurate enough.)

Nationals: 59 act, 62 Pythag, 63 WAR spreadsheet

Mets: 89 act, 89 Pythag, 87 WAR spreadsheet

Twins: 88 act, 89 Pythag, 82 WAR spreadsheet

0 recs | Comment 10 comments

Story-email Email Printer Print

Comments

Minnesota Twins are up here

A Look Back at 2008 – WAR Style

WAR has them at 82 wins, below 88 actual and 89 pythagorean.

by Adam Peterson on Jan 31, 2009 7:41 PM EST reply actions 0 recs

WAR

Is this the same thing that Dave Cameron did over at Fangraphs with the Rangers and Twins last week? I like the idea of comparing 2008 WAR totals, with 2008 pythagenpat and if there is disparity then seeing if pythagenpat was different due to situational hitting/pitching that was either lucky or unlucky. If I had the b-ref subscription I’d do it for the NL West. What I have is…

Team – 08 Wins – 08 WAR – 08 Pythag
LAD – 84 – 82.2 – 86
ARI – 82 – 82.8 – 81
COL – 74 – 69.6 – 73
SF – 72 – 80.9 – 69
SD – 63 – 69.7 – 67

Would be intersting to see if the Giants with their huge 8.9 win disparity between actual 08 wins and 08 WAR was due to poor situational hitting and pitching (bad luck). Same for the Padres, just not quite as bad. Opposite effect for Colorado, as they may have had good situational hitting/pitching. Perhaps someone with a subscription can look it up.

vr, Xei

by Xeifrank on Jan 31, 2009 10:40 PM EST reply actions 0 recs

Uh, yeah, it's the same thing.

It actually wouldn’t be too hard to just go to the team WAR leaderboards at Fangraphs and compile one big table. Maybe I’ll do that for an article.

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

by Sky Kalkman on Feb 1, 2009 3:15 PM EST up reply actions 0 recs

I already have said spreadsheet

If you want it, just drop me an email.

by davidcameron on Feb 1, 2009 5:13 PM EST up reply actions 0 recs

The Giants had horrible bullpen pitching last season

particularly in close games. Most teams, because they have their best pitchers pitching those innings, will pitch better than usual— the Giants were 5% WORSE than usual. And they were unclutch on top of being worse than usual (the game where they gave up a homer to Pat Burrell with 2 outs in the 9th alone cost them about 0.9 wins relative to WAR, and that wasn’t the only big late hit of the season by a long shot).

Many years from now, when his name's recalled
Everyone will say, "He should have passed the ball"
-- Al Stewart, "Football Hero"

by PaulThomas on Feb 1, 2009 5:55 PM EST up reply actions 0 recs

Re :Giants

Thanks, it would be interesting to see those “unclutch” numbers in terms of what Dave Cameron did over at Fangraphs. I think the tendancies is for people to put alot of faith on pythagenpat as opposed to WAR. Would be interesting to see which one correlates to actual wins better, WAR or Pythagenpat. Great site, and great discussion!!
vr, Xei

by Xeifrank on Feb 1, 2009 6:06 PM EST up reply actions 0 recs

I'm guessing Pythag correlates better, because it uses actual RS and RA.

But WAR would likely be a better predictor of future seasons.

Great idea for a study. I’ll probably run with this one, too.

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

by Sky Kalkman on Feb 1, 2009 6:11 PM EST up reply actions 0 recs

That's not necessarily true.

Remember, BP’s third-order wins uses EqR (essentially linear weights) to figure a team’s Pythag win percentage.

Most WAR approaches I’ve seen use some variation on Pythag to figure the runs to wins conversion, even if they’re not explicit (or unaware) that they’re doing so. Both approaches use Pythag.

I has a blog. | Ask a sabermagician!

by cwyers on Feb 2, 2009 12:00 AM EST up reply actions 0 recs