For those who are unfamiliar with metrics, the gap between pitching WAR at Baseball-Reference.com and Fangraphs.com can be difficult to fathom.
By Baseball-Reference metrics, Jose Quintana has been the most effective starting pitcher in Chicago this season (through Sunday) with a 1.4 bWAR, 1.2 with the White Sox and 0.2 with the Cubs. Kyle Hendricks is next at 1.3, with fellow Cubs starters Jake Arrieta at 1.1, Jon Lester at 0.9 and John Lackey at 0.3.
At Fangraphs, however, Lester is No. 1 at 2.8, followed by Quintana at 2.3, Arrieta at 2.1, Hendricks at 1.1 and Lackey at -0.1.
The difference is in fielding-dependent wins. Baseball-Reference’s bWAR includes them, and Fangraphs’ fWAR doesn’t.
That’s because of what the metrics are designed to do. The Baseball-Reference version reflects what has happened while a pitcher was on the mound. Fangraphs attempts to balance that with a predictive element.
Hendricks (3.70 ERA, 3.81 runs allowed per nine innings) has allowed fewer runs than Lester (3.99, 4.73), so Hendricks has a better bWAR. That’s influenced by defense, as seen in part by 12 unearned runs against Lester and only one against Hendricks.
Fangraphs factors out defense by basing its fWAR on fielding-independent pitching — walks, strikeouts and home runs allowed. Lester, who matches Hendricks with 1.2 homers allowed and 2.8 walks per nine innings, strikes out 9.5 per nine innings against 7.5 for Hendricks. So Lester, with a 3.69 FIP to Hendricks’ 4.21, has a better fWAR.
That brings us back to fielding-dependent wins. Fangraphs breaks the total into balls-in-play wins and left-on-base wins.
Balls-in-play wins calculates how many more or fewer runs normally would score given balls in play against a pitcher. The formula is complex, includes park factors and a runs-to-wins conversion factor and uses linear weights that change from year to year: ((((H-HR)*(w1B * p1B + w2B * pxBH)) / (TBF – HR – BB – HBP – SO)) – lgwBABiP) * (TBF – HR – BB – HBP – SO) / PF / RtW * -1.
That boils down to high BIP-Wins if a lower-than-average portion of balls in play drop in for hits. Left-on-base wins will be high if a higher-than-usual percentage of runners are left on base.
Unusually high BIP-Wins and LOB-Wins usually don’t hold up from year to year, so Fangraphs omits them from its WAR calculation.
But if you add those two factors to fWAR, you get runs allowed WAR, which is very close to bWAR.
In Lester’s case, BIP-Wins (-0.2) and LOB-Wins (-1.2) are both negative, so his runs allowed-based WAR will be lower than the FIP-based fWAR. Hendricks is on the positive side with 0.4 BIP-Wins and 0.2 LOB-Wins, so fWAR will be lower than runs-allowed WAR or bWAR.
Both runs-based and FIP-based WAR give us valuable information. BWAR reflects what has happened so far, while fWAR is a strong indicator of near-future performance. In a Lester-Hendricks comparison, your choice depends on what you want from the metric.
Follow me on Twitter @GrochowskiJ.