Cy Young Award 2017: The BBWAA is getting smarter

Cy Young Award 2017: The Difference Between WARs

To determine how much progress has been made by the BBWAA electorate, I compared the history of Cy Young winners to the two versions of WAR mentioned above, Fangraphs and Baseball-Reference. The framework for both versions of WAR are the same, but there are differences in the inputs each uses.

Both versions of WAR care about how many runs a pitcher allows and how many innings he pitches as a starting point, with league, ballpark, and other adjustments included. Essentially, fWAR uses Fielding Independent Pitching (FIP) as the baseline, while bWAR uses a pitcher’s actual runs allowed (both earned and unearned).

To understand the difference, consider Pedro Martinez when he was at his absolute best during the 1999 and 2000 seasons. The raw totals:

• 23-4, 2.07 ERA, 213.3 IP, 313 K, 37 BB, 9 HR—1999
• 18-6, 1.74 ERA, 217.0 IP, 284 K, 32 BB, 17 HR—2000

Both seasons were amazing, especially when you consider the level of offense at the time. In 1999, Mark McGwire hit 65 homers and Sammy Sosa blasted 63. Jeff Bagwell scored 143 runs and Manny Ramirez drove in 165. Larry Walker hit .379/.458/.710.

In 2000, Bagwell scored 152 runs and Todd Helton drove in 147 while hitting .372/.463/.698. In this era of incredible hitting, Pedro Martinez had a 1.90 ERA in 430.3 innings across two seasons. It was just unreal and he demonstrated his brilliance in the 1999 All-Star Game.

When it comes to comparing fWAR and bWAR, ask yourself which incredible Pedro Martinez season was better? In 1999, he had a higher ERA but struck out more hitters and allowed fewer home runs. In 2000, he had a lower ERA, but struck out fewer batters and allowed more home runs (almost double the amount).

Fangraphs WAR uses FIP, which is only interested in strikeouts, walks, hit by pitches, and home runs allowed. These are outcomes that don’t involve the defense. The idea is to isolate what the pitcher is most responsible for.

Baseball-Reference uses the pitcher’s actual runs allowed as one of the initial inputs for their version of WAR. It should be noted that the Fangraphs version of pitcher WAR adds infield flies in with strikeouts because they are also automatic outs, for the most part.

Fangraphs has Pedro’s 1999 season worth 11.6 WAR and his 2000 season worth 9.4 WAR, which makes sense when you look at his strikeouts, walks, and home runs allowed. Baseball-Reference has the seasons flipped, with his 1999 season being worth 9.7 WAR and his 2000 season being worth 11.7 WAR. If you average the two WARs together, then the seasons are essentially equal (roughly 10.5 WAR).

One of the big differences between the two seasons is Pedro’s Batting Average On Balls In Play (BABIP). Despite being so dominant in 1999, with 313 strikeouts in 213.3 innings, Pedro allowed a BABIP of .323.

This was the worst BABIP allowed of his career in any season in which he pitched 50 or more innings. The very next season, Pedro allowed a .236 BABIP. His career mark is .279.

What about the Red Sox defense in those two seasons? Were they much worse in 1999, when Pedro allowed a .323 BABIP? Actually, no. In 1999, all Red Sox pitchers combined to allow a .287 BABIP. In 2000, they allowed a .288 BABIP.

The Red Sox defense ranked third in baseball with 44 runs saved based on Total Zone in 1999. The next year, they were sixth in baseball with 39 runs saved. Defense doesn’t seem to be the reason for the difference in Pedro’s BABIP.

Does it make sense that Pedro was worse at allowing hits on balls in play in such a dominant season? Or could it have been some bad luck? The Fangraphs version of WAR doesn’t concern itself with BABIP because it’s not a factor in FIP. Baseball-Reference values Pedro according to his runs allowed, which is influenced by his BABIP. Thus, they value those two seasons differently.