WAR - it is good for something after all (Part 2)

OK, in the last post I gave a quick primer on the concept of WAR so let's dive into this topic....
How did WAR do as a model, based on the final win totals for 2009?

The concept of WAR gauges how each player can do against his potential replacement by a AAAA-level player...a player that is too good for AAA, but never good enough to play in the majors.

It is assumed that an entire 25 man roster of these replacement level players would win 48 games over the course of the major league season. So the 2003 Tigers (43-119) actually would have been better just playing their whole AAA team!

On fangraphs.com you can find WAR by a whole team for both batting and pitching. So I went through a added each team's batting WAR to its pitching WAR and then added in the baseline 48 wins to find what WAR predicted each team to finish. Listed below are the results, with each team's actual wins listed first and their WAR win total calculated (baseline + bat + pitch, in that order). The number in parentheses after the WAR total is the difference between the two.

NYY 103 (48 baseline + 38.2 bat + 18.7) = 104.9 (-1.9 wins)
BOS 95 (48 + 27.4 + 23.6) = 99 (-4 wins)
TB 84 (48 + 34.1 + 16.9) = 99 (-15 wins)
TOR 75 (48 + 21 + 18.1) = 87.1 (-12.1 wins)
BAL 64 (48 + 15.5 + 7.4) = 70.9 (-6.9 wins)

MIN 87 (48 + 21.7 + 16.4) = 86.1 (+0.9 wins)
DET 86 (48 + 21.3 + 16.5) = 85.8 (+0.2 wins)
CHW 79 (48 + 10.8 + 22.6) = 81.4 (-2.4 wins)
CLE 65 (48 + 19.6 + 10.5) = 78.1 (-13.1 wins)
KC 65 (48 + 6.8 + 19.9) = 74.7 (-9.7 wins)

LAA 97 (48 + 29.5 + 16.6) = 94.1 (+2.9 wins)
TEX 87 (48 + 22 + 18.5) = 88.5 (-1.5 wins)
SEA 85 (48 + 20.6 + 16.4) = 85 (0 wins)
OAK 75 (48 + 16.9 + 19.3) = 84.2 (-9.2 wins)

PHI 93 (48 + 27.5 + 13.3) = 88.8 (+4.2 wins)
FLA 87 (48 + 19.6 + 14.1) = 81.7 (+5.3 wins)
ATL 86 (48 + 17.9 + 23.4) = 89.3 (-3.3 wins)
NYM 70 (48 + 12.1 + 7.3 ) = 67.4 (+2.6 wins)
WAS 59 (48 + 16 + 3.7) = 67.7 (-8.7 wins)

STL 91 (48 + 18.8 + 19.3) = 86.1 (+4.9 wins)
CHC 83 (48 + 12.6 + 18) = 78.6 (+4.4 wins)
MIL 80 (48 + 25.8 + 3) = 76.8 (+3.2 wins)
CIN 78 (48 + 14.4 + 10.4) = 72.8 (+5.2 wins)
HOU 74 (48 + 12.1 + 10) = 70.1 (+3.9 wins)
PGH 62 (48 + 14.2 + 8.1) = 70.3 (-8.3 wins)

LAD 95 (48 + 23.9 + 19.4) = 91.3 (+3.7 wins)
COL 92 (48 + 18.7 + 23.6) = 90.3 (+1.7 wins)
SFG 88 (48 + 12.3 + 21.7) = 82 (+6 wins)
SD 75 (48 + 15.8 + 5.9) = 69.7 (+5.3 wins)
ARI 70 (48 + 15.6 + 17.9) = 81.5 (-11.5 wins)

A few thoughts...
1. 14 of the 30 teams were plus or minus within 4 wins of their actual totals. 4 wins either way is a 2.5% margin, very reasonable in my opinion. If you expand to +/- 5 wins, over half of the league (17 teams) meets this criteria.

2. WAR really missed on each last place team. It just can't account for the unmeasurable....some teams just can't win some years.

3. Take another look at the WAR win totals by division. Aside from ARI and ATL, WAR predicted the finish of each division. To me, that is the most amazing predictive use for this stat.

4. I wondered if each team's over or underachievement from the +/- 4 wins threshold was due to their defensive prowess (or gaffes). I looked at the Ultimate Zone Rating for each of these 16 teams and did not find a correlation. 8 of the teams that were outside the 4 wins threshold had a correlating good/bad UZR, while the other 8 did not. Meaning if you were the ARI Diamondbacks and underacheived by 11.5 wins, your defensive rating was a +2.5, meaning we can't blame your failures on not playing defense.

The final part of this will break down the 2010 Pirates and see what could be in store for them. Special preview -- it's better than you think.