My method for Non-Leveraged Runs Saved Above Replacement:
1. Find IP, tRA for each pitcher.
2. Subtract 0.35 from the tRA, placing it on an ERA scale.
3. Subtract that from replacement level (5.75 - #2) for starters, (4.75-#2) for relievers.
4. Divide innings by nine.
5. Multiply that number by the number found in #3.
---
For example:
1. James Shields: 215 IP, 3.82 tRA.
2. 3.47
3. 2.28
4. 23.88
5. 54.4464 NLRSAR
---
The results:
Pitcher | IP | tRA | NLRSAR |
Shields | 215 | 3.82 | 54 |
Sonny | 193.3 | 3.84 | 49 |
Jackson | 183.3 | 5.68 | 9 |
Garza | 184.7 | 4.32 | 37 |
Kazmir | 152.3 | 4.71 | 24 |
Percival | 45.7 | 5.59 | -2 |
Wheeler | 66.3 | 4.72 | 3 |
Howell | 89.3 | 3.68 | 14 |
Balfour | 58.3 | 2.08 | 20 |
Bradford | 59.3 | 4.3 | 5 |
Add 'em all up and you get around 211 (more here, because I rounded) or about 21.1 WAR. Assume some regression from Kazmir, Balfour, Howell, and addition of a new LOOGY and a long man, and that number might remain static, depending on how much you wish to regress. I used tRA* and league average totals for Price, LOOGY, and Long-Man just to see:
Pitcher | IP | tRA | NLRSAR |
Shields | 210 | 4.09 | 47 |
Sonny | 190 | 4.31 | 38 |
Price | 150 | 6.1 | 0 |
Garza | 175 | 4.58 | 30 |
Kazmir | 170 | 4.71 | 27 |
Percival | 50 | 5.43 | -2 |
Wheeler | 65 | 4.75 | 3 |
Howell | 65 | 3.92 | 9 |
Balfour | 55 | 3.17 | 12 |
Bradford | 55 | 4.04 | 6 |
Long Man | 50 | 5.1 | 0 |
LOOGY | 50 | 5.1 | 0 |
For innings I either estimated (new three) or just took five off each high total for arguments sake while adding a few to the injured. We're missing about 200, I would guess 50-60 come from our "sixth starter" and the rest will be distributed amongst the relievers. Far from perfect, but it is what it is, a general idea.
All together that puts us at about 42-45ish WAR. Replacement level is a team that wins 29% of it's games, or 47. So right now, we're between a 89 and 92 win team, with quite some room to grow, and in some cases, possibly some more regression to take place.