The Support Neutral Win-Loss pitching report has been an integral part of Baseball Prospectus since our inception in the mid 1990s. In fact, Michael Wolverton invented SNWL way back in 1993, when it was presented in a paper published by SABR’s “By The Numbers” statistical newsletter. The concept behind Support Neutral pitching statistics is a simple one: determine what a pitcher’s W/L record “should” have been, if he had gotten average performances from his teammates, adjusted for park, and looking at each start individually.
However, even the most venerable tools occasionally need an update, and the Support Neutral pitching reports have now gotten a slight facelift. A few of the more significant changes and differences between the old and the new are detailed below:
- SN Pitching reports are now available from 1972 through 2004.
- Replacement level is now being computed based on the research presented and the formula introduced in Baseball Prospectus 2002, rather than assuming a .425 replacement-level winning percentage.
- Expected wins and losses for each start are now based on historical data for how often a pitcher got a decision with the same pitching line.
- A new lineup adjustment that corrects for the actual strength of opposing hitters faced by a pitcher. These lineup-adjusted stats are indicated by starting with “SNL” (with apologies to Lorne Michaels), rather than just “SN” in the new reports.
- Bullpen support stats are computed for all starting pitchers.
Because of the amount of new information to present, I opted to break all this down into three new statistical reports:
- Pitcher’s Expected Win-Loss Record (corresponds to SNW and SNL and includes how “lucky” a pitcher’s record is)
- Support Neutral Value Added (measures changes in expected team wins due to the pitcher’s performance)
- Starting Pitcher Bullpen Support (shows how often the pen stranded runners bequeathed by the starter versus an average pen)
The new column headers in each report are listed below:
Bullpen Support Statistics for Starting Pitchers:
BQ_RUNNERS – bequeathed runners by this starting pitcher
BQ_SCORED – bequeathed runners who scored (allowed by bullpen, charged to starter)
FAIR_RA – RA computed assuming that an average number of bequeathed runners score
BULLPEN – Bullpen support level; number of runs above average the pen allowed to score that were charged to the starter. Negative numbers mean the pen stranded more runners (good pen performance), positive means more runners were allowed to score (bad pen performance)
Support-Neutral Lineup-adjusted Value Added for pitchers:
SNVA – Support Neutral Value Added – wins above average added by the pitcher’s performance
SNLVA – Support Neutral Lineup-adjusted Value Added – like SNVA, but also adjusted for the MLVr of each batter the pitcher faced.
SNVAR & SNLVAR – like SNVA and SNLVA, but comparing to replacement level, rather than average. Replacement level is now being computed the same way in SNVA and in VORP (using the formulas from my BP 2002 article).
FLAKE – standard deviation of per-start SNVA for each pitcher. This was previously shown as the variance, and was used to compute the “flakiest” pitchers. Standard deviation is just the square root of the variance, so these are equivalent.
Expected Win-Loss and Team Records for Starting Pitchers
W, L – actual wins/losses credited to the pitcher
E(W), E(L) – expected win/loss record for the pitcher, based on how often pitchers with the same innings pitched and runs allowed earned a win or loss historically (this differs from how Michael was computing it, which was a more complicated, theoretical calculation).
LUCK – Luck, as measured by the number of extra wins, and short losses the pitcher actually got, versus his expected record. LUCK = (W-E(W))+(E(L)-L)
TEAM_W, TEAM_L – Team’s expected wins and losses in the games started by the pitcher. This will always add up to the pitcher’s total games started.