The Measure of Baseball Immortality
Image from: USA Today
The MLB All-Star game is just around the corner. It brings with it debates over which pitcher should have made the roster and which one was snubbed. The unspoken premise of any such argument rests on intangible qualities of what an all-star season “looks” like. It reignites timeless comparisons to long-lost memories of All-Stars of yesteryear.
Of course, modern pitchers have always carried the burden of their forebears. Living up to the ghost of legends will probably always be the price of admission for the rite of passage. In 1999, Pedro Martinez of the Red Sox fell short of fellow Hall of Famer Carl Hubbell in nearly striking out five consecutive batters in the midsummer classic. But that’s just baseball.
What if there were a simple and standardized way to compare pitchers from any season or era in baseball?
Many have tried. Official MLB historian John Thorn co-authored a tome with Pete Palmer in the 1980s that upset the innate wisdom of baseball elders. Another upstart-baseball-researcher-turned-famous, Bill James, provided a wealth of new metrics for baseball fans. Twenty years ago, James created a statistical concept known as win-shares. His metric relies on a nebulous comparison to a mythical “replacement” player. Essentially, should a star go down with a prolonged injury, his team would have to call up a replacement player from the minor leagues. As such, the replacement player would be expected to perform at a much lower level.
The claim isn’t entirely unfounded. At the same time, there isn’t a specific definition for a replacement player. Why are all replacement players considered to be equally subpar? Both George Brett and Nolan Ryan were sent back to the minors only to be later inducted into the Hall of Fame. They are hardly alone in taking an alternate route to the bigs. More than 800 Major Leaguers showcased their skills in semi-pro baseball.
There are so many replacement-type stats that Baseball-Reference maintains a page on its website to help readers keep them all straight. Each researcher painstakingly laid another brick upon the edifice of baseball lore by advancing our collective understanding of the game we love. Still, those measures fall short because they rest on a faulted premise — they evaluate position players and pitchers against a monolithic, fairy-tale replacement instead of evaluating baseball players on their individual performances. Moreover, those same measures claiming to reflect win-shares accurately cannot agree — or adequately infer — where runs come from. How can Clayton Kershaw be compared to Sandy Koufax when there isn’t consensus on what makes up the lifeblood of baseball?
Thinking back to which pitchers are worthy of All-Star Game appearances implies evaluating their effectiveness at their craft. In order to do so, a person must definitively understand how runs are generated or prevented. All events in a baseball game either contribute to run production or limit it. Before designing yet another metric that rests on faulted premises, I set out to understand where runs come from; exactly where they come from. It required deconstructing the events of a game — actually, all 229,513 games since 1876 — to explain where all runs came from. Long story short, my Predicted Outcomes Runs formula accurately describes an average of 99.96% of all runs ever scored. It means, on average, the formula is off by 0.1 runs-per-team-per-game while its closest direct competitor is off by 0.48 runs over the same period. Enough of that for now; it’s a story for another day.
A well-designed understanding of runs is the precursor to crafting a robust metric that measures a pitcher’s effectiveness. A person needs to know which statistics, and at what precise interpretive calculations, go into making such declarative judgments. It allows for audacity like claiming that Christy Mathewson is the definitively greatest pitcher of all time. In second place is current Dodgers pitcher Clayton Kershaw, should he maintain his current pace. It is sure to spark outrage for older fans of the storied club from Los Angeles and Brooklyn before then. Those salty fans will undoubtedly pound the table for the likes of Sandy Koufax, who ranks 25th, or Don Drysdale, who is still a respectable 51st overall.
These arguments fill conversations for grown men and boys alike. Still, they don’t describe what an all-star season “look” like. To offer defensible claims requires applying a new data-driven, run-focused metric — like my Pitcher Outcomes Calculator — to single seasons. Baseball-Reference and Fangraphs offer proprietary calculations regarding which pitcher they rank the highest. Pud Galvin’s 1884 season was worth over twenty wins above a replacement player, according to BR, while the very best of modern pitchers are lucky to get half that number. The BR list is dominated by long-dead pitchers nobody alive today could have seen play. These icons — well, some are still household names for avid fans like me — hold forty-four of the top fifty spots on what might politely be described as a questionable assortment of names. The most recent addition to the list came back in 2000 when Pedro Martinez — ranked 37th by BR — got his revenge for falling short of Carl Hubbell’s All-Star game record.
Reexamining Baseball Reference’s list using my Pitcher Outcomes Calculator shakes up the leaderboard considerably. Looking only at BR’s top fifty pitchers, Pedro Martinez shoots to the top of the list. Dwight Gooden’s 1985 season now finds itself ranked ninth. Digging deeper, each of the top ten pitchers in wins above replacement for each of the last five seasons was tossed into the blender with the original gang of ghosts to offer a sample of old and new for comparison. The mixed-tape playlist includes fourteen current pitchers hurling a combined twenty-three spectacular seasons over the last five years. A new — and improved — ranking places seven current players in the top ten. Newsflash; Kershaw edged out Pedro Martinez with the best season ever.
In truth, all 20,000-plus Major Leaguers, which now thankfully includes Negro League baseball, should be reviewed using these new metrics. It may ease hard feelings when a favorite player is snubbed from the All-Star team because current metrics are inadequate for measuring effectiveness and efficiency. At the very least, a dispassionate, data-driven approach quiets much of the hyperbole that fogs the mind with hazy memories of a bygone youth. It means the threshold for the rite of passage has been crossed. Current players are every bit as good, if not in some cases better, than legends they are held to.
All of this is part of the apparatus of baseball. In 1876, few statistics were kept. The body of available data grew over the years. In fact, it is still growing. By and large, it has less to do with simply having more factual information and more to do with attempting to compare one player’s performance accurately to another. The quest for universal accuracy may remain elusive. The pursuit of that quest — measuring baseball immortality — will stay at the heart of any such endeavor.
Oh, and to settle a previous debate between a grandfather and his grandson, Dizzy Dean is markedly more effective than Zack Greinke. Please don’t take my word for it. The data speaks for itself.
Postscript:
I have deep respect and admiration for the folks like me who wanted nothing more than to better understand and explain our beloved game. This includes the names of men I have listed here and many more names I haven’t mentioned. I don’t know any more or better than anyone else about baseball. I simply chose to use a completely objective approach to deconstructing what we though we knew. I didn’t care where the data took me. I just wanted to find answers. My metrics are not perfect, and I intend to improve these prototypes as I make better sense of what I learn along the way. If you would like MS Excel files of the calculators, please email me at chriswrites@schristophermichaels.com