Pointing Fingers: Rollie Fingers and WAR

A week ago, here on Beyond the Boxscore, Adarowki, wrote an article called, "The Ongoing Quest to Define a Hall of Fame Reliever." He stated the following:

Rollie Fingers is a Hall of Famer because, for a while, he was the all time save leader. It's too bad somebody got in for leading in a flawed and meaningless statistic. Fingers was certainly a great pitcher-he's 11th all time in raw reliever WAR. But he doesn't even sniff the Top 20 as a rate stat, though. He actually ranks 34th, behind the likes of Greg Minton and Dan Plesac. I don't think Rollie Fingers is a bad pitcher. I just don't think he's any better than Kent Tekulve.

I have a problem with this statement. Of the top ten relievers in Adarowki's Study, only Rivera, Wagner, Sutter and Smith have better career FIP numbers than Fingers. The margin between Fingers (2.96), Sutter (2.94) and Smith (2.93) is so small that it cannot clearly distinguish either as a more dominant pitcher. Sutter is very close to Fingers in most peripherals and probably somewhat better due to his higher K/9 rate (7.43 for Sutter to 6.87 for Fingers). Lee struck out more but also walked more, negating much of his edge. If there is one difference we can see between these three men that should significantly impact their value it is innings pitched per season.

During the seasons where each player could be considered the primary relief ace, Smith averaged 77.25 innings, Sutter averaged 99.21 innings and Fingers averaged 106.03 innings. This understates the difference in their usage though. Fingers topped 120 innings five times and twice topped 130 innings over his 15 full seasons. Sutter reach 120 only twice and never pitched more than 122.2 innings. Smith only topped 100 innings once and 90 innings only three times. Despite being extremely similar in FIP and peripherals, Sutter and Smith look much better than Rollie Fingers in WAR/200 IP because that system rewards pitchers who are used in the mode of the "modern closer." Isn't it better to have a pitcher who throws 110 innings of 2.96 FIP ball than one who throws 75 innings of 2.93 FIP ball? Why is it that we are eager to reward players who are used in a way most of us in the sabermetric community at least question, if not wholly reject?

In the BJHBA, Bill James cites Sutter's 1977 and 1978 second half struggles for transitioning relievers from the Hoyt Wilhelm mode of heavy usage to the modern usage under which Wagner and Rivera have flourished. Rollie Fingers had one foot in both systems. Early in his career he split time between starting and closing games. Then in 1972 Fingers became the teams relief ace full time. He did make two starts again in 1973 though and this shows that reliever was not as of yet a fully developed role, but just congealing at this time. From 1972 to 1977, Fingers pitched an average of 127 innings with a 2.53 FIP. After 1977, Fingers never again topped 120 innings, sliding into a usage pattern almost identical to Sutter's and not too far from Lee's. He would post his best ever FIP in 1981 with the Brewers pitching 78 innings with a 2.07 FIP. That season he was 34 years old.

We generally expect relief pitchers who throw between 70 and 90 innings a season to post better FIP numbers on the average than starters pitching 200 innings a season. Shouldn't also expect the numbers to change if the reliever doubles his innings while still being in a relief role?

The limited innings, combined with the short length of each appearance help create the illusion of greater dominance for modern relievers and Runs Allowed (which is the basis for the Rally WAR numbers) is even more subject to such biases than FIP. Fingers career ERA (2.90) is almost indistinguishable from his career FIP (2.96) Gossage is very close as well (ERA 3.01, FIP 3.18) Rivera and Wagner both outperform their FIP by significant amounts- Rivera's ERA is .55 points lower than his FIP and Wagner's is .43 points lower. Both of these guys are great pitchers, but I don't think they are just outperforming their FIP so much as they are benefiting from each season's sample size being tightly limited.

Rollie Fingers is not as great a reliever as Mariano Rivera, no one is. However, simply looking at Rally's WAR data and concluding he is closer to Kent Tekulve (who has a 3.26 FIP in 265 fewer innings) is not much better than assuming he is great because he had a lot of saves for his time. Mr. Adarowki is a good analyst and his article does a great job at drawing some underappreciated relievers, like Wagner, Smith and Hiller into the Hall of Fame conversation. Letting one statistic, even a good advanced metric like Rally's WAR, tell the story is just a dangerous way to analyze players, especially when their eras and roles are distinctly different.