Ye Olde Phillips Milk of Magnesia Relief Award?

How do you spell relief?
M O R D E C A I  B R O W N
At least in the Deadball era, Frank Chance did. As has been chronicled elsewhere, Brown was both the Cubs’ ace and their closer. He led the league in saves four times, all consecutive. In 1910 he became baseball’s all-time saves leader with 1910 with 39. He extended that total to 49 before retiring and held the record until 1926 when Firpo Marberry broke it. He also held the single-season record (13, later tied by Chief Bender) from 1911 through 1923 until it was also broken by Marberry. He finished nearly every game he relieved in.

He wasn’t alone.

On the South Side of town, Big Ed Walsh led his league fives times in saves. Joe McGinnity led three times. Chief Bender and Bob Shawkey twice. Imagine Justin Verlander leading the league in saves.

And time went on. John McGraw, Bucky Harris, Joe McCarthy, and other managers began experiments that eventually led to the emergence of relief aces such as the Yankees’ Johnny Murphy. But Connie Mack continued using aces like Lefty Grove and as bullpen stoppers well in the 1930s, and other managers even past World War Two. Grove was the active saves leader from 1937–1939. Hal Newhouser racked up 18 saves from 1942–1950. Bob Feller picked up 10 saves from 1946–1948. Virgil Trucks notched 13 saves from 1949–1956 while also finding time to start 194 times. In fact, when you think about the teams who dominated baseball after the War—the Yanks, the Dodgers, the Indians—they were mostly those who were on the leading edge of the reliever evolution.

Thanks to play-by-play data that currently ranges back to the mid-1940s, we have a very strong sense of how post-War managers leveraged their bullpens. There’s a descriptive stat called Leverage Index (LI) that measures the relative urgency each pitcher entered the game with. A starter always begins at 1.0, exactly average. Most closers today are used with an LI of around 2.0, meaning that their point of entry is twice as urgent as the first pitch of the game.

So what about Brown and Grove? Or Big Ed Walsh, Christy Mathewson, or Walter Johnson, all of whom were used in the hybrid starter/reliever ace way? Because we don’t have play-by-play information for them, we can’t say for sure what sort of leverage they pitched in. Which also means that value calculations aren’t capturing their full contributions.

Here’s what we can know. BB-REF has starter/reliever splits for most pitchers dating back to around 1916. I looked at starter/reliever splits for 42 pre-War pitchers for whom we currently have extensive splits. I chose only guys reasonably thought of as among the top 200 or so pitchers of all time. From Pete Alexander to Tom Zachary. They averaged about two and one-third innings per relief appearance. In other words, they usually entered in the sixth or seventh and pitched through the eighth or ninth (depending on whether their team came back or stayed ahead).

About 9 percent of their total innings came in relief. They finished three-quarters of the games they entered in relief and earned a decision or a save in about half of those appearances. Among the games they finished, they got a decision or a save 69% of the time. Their managers saved them for relatively high leverage situations.

I compared the information on these pre-War pitchers to a handful of well known relievers ranging from the 1950s through today. Let’s chart it:

You can see all the strong historical trends at work here that lead to more specialization in relief pitching and our modern closer-centric bullpen. Fewer innings, higher leverage, more saves, more games finished. Not surprisingly, the relief appearances of pre-War aces look the most like Satchel Paige who came right after the War. Perhaps more surprisingly, they look a bit like Mike Marshall from the 1970s.

So what’s a nerd like me to do to give the pre-War guys something like approximate credit for their relief work? WAR has an LI component that’s used as a multiplier for any relief appearances starters get. For Play-by-Play era guys, it uses that calculation on runs given up in relief. Before that it defaults to 1.0 for all appearances.

That’s not real helpful.

For me, I make a rough estimate of a pre-play-by-play guy’s relief and starter innings. Then I allot his runs allowed to each role by the ratio of his estimated relief innings to his estimated starter innings. Then following BB-REF’s calculations, I use a conservative 1.5 leverage index for each relief appearance. If/when the heroic members of Retrosheet turn out play-by-play info for before the 1940s, we’ll update to the observed values.

To be honest, it doesn’t add up to a ton of value. On the other hand, for Mordecai Brown, even that little bit of value helps his borderline case. And for 1921 candidate Ed Walsh, it’s a little more icing on the top of his peak.