Fangraphs recently introduced a set of new win value pitching stats, which might be characterized as alternatives to the standard Fangraphs' WAR. I know some of you may be thinking, "I need new stats like I need a hole in the head." But after examining them, I think they can be helpful in deconstructing the ways that pitchers have succeeded or failed.
Fangraphs' WAR uses Fielding Independent Pitching (FIP) to measure pitching performance; and FIP, in turn, is based only on the events that the pitcher can control the most: strike outs, walks, and HRs. This is based on DIPS theory---pitchers have limited control over BABIP, batting average on balls in play. In the previous "talking sabermetrics" article, I pointed out that Fangraphs' pitching WAR is superior to Baseball-Reference's pitching WAR (which doesn't rely upon FIP) if the objective is to evaluate the pitcher's potential future pitching ability.
The new stats start with a WAR calculated on the basis of runs allowed per 9 innings (RA9-Wins). This is the most raw form of actual pitching results. This attributes everything that happens to the pitcher. Errors, fielding quality, etc., are not taken into account. The other two new stats are based on the difference between the standard fangraphs' WAR and RA9-Wins, identifying the two components that comprise the differential. In my view, the main value of the new stats is that it gives us more information on why a pitchers' results are better or worse than FIP.
The new stat "BIP-Wins" quantifies the difference between FIP and RA which is due to the impact of BABIP. Conventional sabermetrics attributes excessive or very low BABIP mostly to luck. However, the quality of the defense behind the pitcher also affects the BABIP level. (Before advanced metrics, the inverse of BABIP, called "defensive efficiency ratio" was used to compare team defenses.) The pitcher's ability to produce mostly flyballs or groundballs can affect BABIP too. Some pitchers have trick pitches (like a knuckleball) or other unusual skills that makes them BABIP outliers over the long term. But, in general, pitchers' ability to sustain BABIPs far above or below league average is unlikely. BIP-Wins is not based solely on batting average, because it considers the additional impact of extra base hits on balls in play.
The new stat "LOB-Wins" quantifies the difference between FIP and RA which is due to the stranding of base runners. In effect, LOB-Wins is a clutch-like measure, in that the new stat reflects the timing of run prevention actions (whether hits, Ks, BBs, pick offs, etc.). Measures of stranding runners can reflect both skill and luck. Extremely high or low left on base numbers probably are unsustainable, and may indicate a lucky or unlucky pitcher. (For example, a pitcher, who walks the bases loaded without any runs scoring, can't expect to to do that consistently.) However, we know that pitchers can control a good portion of their skills with runners on base: some pitchers are worse pitching from the stretch; pitchers can increase their velocity or use different sequences of pitches with runners on base; some pitchers are better at inducing double plays; some pitchers hold base runners better, etc.
So, the math for the new stats is shown below:
FDP = RA9-Wins minus WAR
FDP = BIP-Wins + LOB-Wins
For example, the Astros currently are ranked as the worst team in RA9-Wins at -0.7, but are 28th in pitching WAR at 4.2. This means that the fielding dependent pitching (FDP) component for the Astros is the difference, or -6.2 (ranked 28th). -3.6 wins is due to BIP-Wins (25th ranking). -2.6 wins is due to LOB-Wins (ranked 28th).
The Astros starting pitching has been more productive than the bullpen pitching. No surprise there. RA9-Wins stands at 1.7 (27th) and WAR is 5.2 (25th). The difference is mostly LOB rather than BABIP. The Astros' BIP-Wins equals -0.5 (18th ranked) and the LOB-Wins stands at -3.0 (29th ranked).
What we see, with the new stats, is that the Astros' starting pitchers tend to break down when runners are on base. Some of the results may be bad luck, either in terms of fielding issues with runners on base or bad luck with sequences of hits. However, to some extent, it reflects how the starting pitchers are responding to runners on base. I suspect that the inexperience of the Astros' starting pitchers affects their ability to pitch with runners on base. And, if I were advising the pitching coach, I would suggest re-doubling efforts at improving the starting pitchers' delivery from the stretch.
Among all ML starting pitchers (at least 100 IP), Jordan Lyles has the worst LOB-Wins (-1.4). In game threads, people frequently mention that Lyles has a frustrating tendency to encounter the big inning in the 5th or 6th. Lucas Harrell is also among the worst (13th worst) ML starters in LOB-Wins. Harrell's groundball tendency is a two edged sword with runners on base: groundballs produce more hits, but also can erase runners with a GIDP. My impression is that both Harrell and Lyles have a tendency to overthrow with runners on base.
Bud Norris and J.A. Happ (both -0.6) were among the ML worst qualified starters in BIP-Wins. Norris' BABIP of .322 is about 30 points higher than league average. Undoubtedly both Norris and Happ suffered some damage from the Astros' weak overall defense. Happ should benefit from a better defense in Toronto. However, Norris appears to have suffered bad luck on batted balls too. His BABIP is above the "normal" range caused by defense or long term random variation. I think there is a decent chance that we will see some positive regression from Norris next year.
What do you think about the use of these pitching stats? Too much? Helpful?