If you’re tired of hearing and rehashing baseball’s unresolvable MVP arguments — best vs. most valuable, context-independent vs. context-sensitive, Bryce Harper vs. Yoenis Cespedes — you can probably blame the St. Louis Browns.
In its current, BBWAA-bestowed incarnation, the MLB Most Valuable Player Award dates back to 1931. Even in Year 1, the preference for winners was firmly in place: The American League award went to Lefty Grove of the A’s, who won the AL pennant by 13.5 games, and the National League award went to Frankie Frisch of the Cardinals, who won the NL pennant by 13 games. Grove was a good pick by any measure, but the selection of Frisch — a deserving Hall of Famer1 whose best days were behind him — came down to his team and his reputed intangibles, which the voters valued more highly than the production of leading statistical candidates like Giants slugger Bill Terry.
And maker of many undeserving Hall of Famers.
But if we can trace the 21st-century MVP voter’s still-strong impulse to look at the top of the leaderboard and say, “No, not him” to Frisch, then we can trace the Terry snub to an even earlier lineage. Although the BBWAA award was the one that stuck, it was preceded by two proto-MVP awards, the first of which was established by Hugh Chalmers in order to sell some cars. The Chalmers Award, which debuted in 1910, was intended to go to the big leaguer with the highest batting average, a one-dimensional stat that might as well have been WAR for the contemporary reverence it received. Naturally, there’s a book about this — in baseball, there’s always a book — but the abridged version is that the competition between the Tigers’ Ty Cobb and the Indians’ Nap Lajoie came down to the last two days of the season. Cobb benched himself, believing his lead was secure, but Lajoie went 8-for-9 in a season-ending doubleheader, bunting for several hits with the help of Browns manager Jack O’Connor, who reportedly told third baseman Red Corriden to play back. The country believed that Lajoie had won by one point — until the AL announced that Cobb had finished ahead, .385 to .384. Chalmers foreshadowed Oprah and gave a car to each candidate.
Record-keeping being what it was at the time, this wasn’t easy for third parties to fact-check: Lajoie actually finished on top, but the truth wasn’t known for more than 70 years, when Pete Palmer discovered that the league had double-counted a Cobb 2-for-3. Statistics clearly couldn’t be trusted, so the following April, Chalmers changed his award to instead go to the person who “should prove himself as the most important and useful player to his club and to the league at large in point of deportment and value of services rendered” — more or less the MVP we love and hate today.
The Chalmers Award was discontinued after 1914, but by then, the Hall of Fame’s Craig Muder writes, it had “established a template for the Most Valuable Player Award” that was a friend to soft factors. If not for O’Connor and Corriden, the idea of a numbers-based standard might have gained traction and colored the way we think of awards. Or maybe we would have ended up with “most valuable” whatever we did, because the concept is too compelling not to have wormed its way into our discourse: A leaderboard brooks no argument, but “value” is amorphous enough to support any opinion; facts can be refuted, but feelings can’t be killed. And so we build our MVP arguments like a grade-schooler begins a book report: by citing the dictionary definition of a word we saw in the title.
Fortunately, this won’t be one of the worst years for award-season acrimony. Cespedes has hit .243/.299/.500 in 19 games since Fox Sports’s Jon Morosi baited our clicks with his MVP argument, while Harper has raised his game, batting .323/.444/.708 with seven homers and one closer suspended. Compared to his closest competitors, most of whom are equally out of the playoffs, Harper’s season has been so transcendent — with a strong finish, he could join Barry Bonds, Alex Rodriguez, and Mike Trout as the fourth player to top 10.0 WAR (via FanGraphs) in the wild-card era — that the odds are against a plurality coalescing around anyone else. And in the AL, Trout and Josh Donaldson are so evenly matched that there’s no obviously right or wrong answer.
The Cy Young race is similar: It’s hard to muster much outrage over any entrant in the NL’s royal rumble between Clayton Kershaw, Zack Greinke, and Jake Arrieta, or the AL’s head-to-head duel between Dallas Keuchel and David Price. MVP anger thrives on polarizing players whose cases are either slam dunks or duds, depending on your perspective. Regardless of your baseball belief system, all of this year’s candidates are great.
So let’s say you’re on Twitter, engaging in a little light trolling, and you come across a Cespedes supporter (or at least a non-Harper adherent). And yes, they are out there. Where should your snark setting be?
I’d vote for Harper, but as long as the BBWAA uses the V-word, it’s OK to consider context. As little blame as Harper bears for the Nationals’ disappointing season — aside from the countless wins he’s cost them by not sprinting out popups — it’s reasonable to suggest his performance would have been more valuable to his team if it had culminated in a playoff appearance. To paraphrase the old Branch Rickey/Ralph Kiner line, the Nats would’ve missed the playoffs with or without Harper, so a player on another team who did make the difference between qualifying and falling short could be a better MVP pick, from a certain point of view.
The good news is, there’s a way we can quantify that: Championship Probability Added. You may be familiar with Win Probability Added, a context-sensitive value stat that considers not only how well a player produced, but when he produced. WAR always assigns the same value to a home run, whether it’s a walk-off shot against a division rival or a meaningless score-extender up 12 runs against a last-place team. But WPA assesses how much a player’s contributions actually helped his team win, by calculating the difference between his team’s win expectancy before and after his actions. Clutchness generally isn’t repeatable, so WPA isn’t the best indicator of true talent or forecaster of future performance. But it’s still a useful way to quantify value in retrospect: As Dave Studeman of The Hardball Times describes it, WPA is a “story stat,” one that captures the excitement we experience in the moment. And as long as voters care about context, the MVP will be a story award.
Although WPA takes into account the sequencing of a player’s actions within games, it doesn’t consider their sequencing within seasons. And if we really care about context, we want to know that too. Just as a walk-off shot is worth more than a tack-on homer, a walk-off shot in a pivotal game against an important opponent should be worth more than one that comes after the campaign’s outcome is already assured.
Baseball Analysts author Sky Andrecheck developed Championship Win Probability Added — which owed a debt to Studeman’s Postseason Probability Added — in 2009 in an attempt to answer the question, “What portion of a title did a player contribute?” Andrecheck is now the Indians’ senior baseball analyst, and Studeman hasn’t updated his code to account for the 10-team playoff format, so I teamed up with Dan Hirsch, creator of historical stats and analysis site The Baseball Gauge, to re-create (and perhaps even improve upon) Andrecheck’s model.2
And by “teamed up,” I mean “I asked Dan to do it, and he did.”
Dan calculated the division-title and wild-card expectancy for every team on every day of the season by simulating the rest of the schedule 100,000 times, factoring in home-field advantage. He then ran another 20,000 sims to calculate the potential change in playoff expectancy that a team could expect from a win in every given game, and compared that to the typical change in playoff expectancy between games. The change in each game divided by that baseline change in the typical contest is the game’s Championship Leverage Index, a measure of how much it affects the team’s playoff outlook. The higher the stakes in a given game, as determined by the standings, the setting, and the opponent, the higher its CLI.
This is easy to see on a division level. Here’s a graph of the game-by-game CLIs in the NL East, which got less momentous as the Mets opened up an insurmountable lead.
You can see the spikes in the Nationals’ black CLI line at Games 101–103 (July 31–August 2) and Games 137–139 (September 7–9), which represent series against the Mets. The Nats lost all six of those high-leverage games, which put the division away.
Compare that to the still-unsettled AL West, where the stakes for the remaining wild-card and division contenders have never been higher:
Once you have the team’s CLI and the player’s WPA (obtained, in our case, from Baseball-Reference), you can multiply them together to get the player’s Championship WPA, or cWPA: the percentage by which he increased his team’s odds of winning the World Series. The more important the game, the more the player’s WPA in that game will count toward his full-season cWPA total. Once a team is eliminated, its CLI falls to zero, and any WPA its players accrue doesn’t count (since there’s nothing the player can do to budge his team’s nonexistent odds of making the playoffs).
Measured by cWPA, the clutchest game of the season so far — who knows what’s in store for this weekend — belongs to Houston’s Jed Lowrie, and it took him only one plate appearance. On September 13, Lowrie pinch hit against Angels closer Huston Street in the top of the ninth with the Astros losing 3-2 and down to their last out. On a 2-1 count, he hit a three-run homer just beyond Angels right fielder Kole Calhoun’s outstretched glove, completing a five-run comeback and putting the Astros on top to stay. The hit was worth .79 WPA, or 79 percent of a win, improving the Astros’ chances from 87 percent likely to lose to 92 percent likely to win. And it came in a game with a lofty 3.0 CLI. That hit increased the Astros’ championship odds by 1.5 percentage points — a small number but a big impact, considering it was a single swing in an extremely long season. In fairness to Street, cWPA doesn’t know that the closer was stricken with symptoms severe enough that I’m comfortable skipping past “flu-like” and calling them “flu.”
[mlbvideo id=”477891283″ width=”530″ height=”297″ /]
The least-clutch outing came four games later, courtesy of another, even more adorable Astro, Jose Altuve. On September 17 in Texas, the highest-stakes matchup this season (CLI of 4.2), Altuve flied out to lead off the game, popped out with a runner on first in the third inning, grounded out in the fifth, and grounded into a double play with two on and no outs in the seventh and Houston down by one run. All told, his outs increased the Rangers’ odds of winning the game by 29 percent and decreased the Astros’ championship odds by 0.78 percentage points.
So which player does cWPA consider the clutchest of them all? Unless you ascribe much of the Mets’ second-half success to Cespedes, giving him credit for not only his own performance, but that of the other players who’ve hit well since his arrival — which, of course, we can’t conclusively disprove — then Cespedes is the wrong player for the context-sensitive crowd to pick as its MVP. There’s a better candidate, one who’s hit well at huge moments for another NL contender, and who’s spent the whole season in the senior circuit.
|National League||American League|
|Anthony Rizzo||Cubs||4.5%||Josh Donaldson||Blue Jays||4.1%|
|Bryce Harper||Nationals||3.4%||Mitch Moreland||Rangers||3.9%|
|Adrian Gonzalez||Dodgers||3.1%||Mike Trout||Angels||3.8%|
|Kris Bryant||Cubs||3.1%||Jose Bautista||Blue Jays||2.8%|
|Matt Carpenter||Cardinals||2.9%||Lorenzo Cain||Royals||2.5%|
|Andrew McCutchen||Pirates||2.8%||Miguel Cabrera||Tigers||2.4%|
|Paul Goldschmidt||Diamondbacks||2.7%||Edwin Encarnacion||Blue Jays||2.2%|
|Curtis Granderson||Mets||2.7%||Nelson Cruz||Mariners||2.0%|
|Buster Posey||Giants||2.2%||Miguel Sano||Twins||2.0%|
|Andre Ethier||Dodgers||2.0%||Mark Teixeira||Yankees||2.0%|
Meet the NL’s cWPA MVP, Cubs first baseman Anthony Rizzo. By WAR, Rizzo just barely cracks the NL’s top 10, at 5.1, according to FanGraphs. But the timing of his hits puts him well ahead of Harper, who ranks second in cWPA. Rizzo has increased the Cubs’ championship odds by 4.5 percentage points, relative to an average hitter.
[mlbvideo id=”485780483″ width=”530″ height=”297″ /]
Rizzo has slashed .278/.388/.511 overall, a little more than 40 percent better than the average NL hitter after adjusting for the friendly confines. But he’s excelled in crucial spots, batting .313/.438/.553 with men on, .296/.417/.578 with runners in scoring position, and .283/.479/.547 with two outs and runners in scoring position. And because he plays for a postseason team, Rizzo’s clutchness counts. Contrast Rizzo with Joey Votto, whom cWPA hurts. Votto, who’s been by far the best hitter in baseball since the All-Star break, has the third-highest WPA in the majors. But because he plays for the cellar-dwelling Reds, his average CLI is only 0.3 and his cWPA sinks to 28th. Including his time with the Tigers, Cespedes sits at 1.7 percent, one spot below Mark Teixeira.
On the AL side, where Donaldson led Trout by a fairly slim margin entering Wednesday (when Trout likely narrowed or erased the gap by going 3-for-3 with a homer and two walks), the surprise is Rangers first baseman Mitch Moreland, who slotted in between the two MVP favorites. Moreland has hit well with men on (.303/.358/.514) and in late-and-close situations (.343/.377/.529); the closer the game, the more damage he’s done.
|Within One Run||254||.959|
|Within Two Runs||345||.917|
|Within Three Runs||397||.887|
|Within Four Runs||435||.844|
|More Than Four Runs||70||.674|
Moreland’s biggest hit of the season by cWPA was his two-run, go-ahead oppo shot on September 14 against, you guessed it, the Astros. The roar of the crowd agrees with cWPA’s assessment.
[mlbvideo id=”479367083″ width=”530″ height=”297″ /]
OK, enough positivity. The table below lists the five hitters in each league who’ve done the most to hold their clubs back.
|National League||American League|
|Jimmy Rollins||Dodgers||-2.0%||Didi Gregorius||Yankees||-2.3%|
|Casey McGehee||Giants/Marlins||-1.6%||Stephen Drew||Yankees||-2.0%|
|Chris Owings||Diamondbacks||-1.6%||Erick Aybar||Angels||-1.9%|
|Jordy Mercer||Pirates||-1.5%||Rene Rivera||Rays||-1.8%|
|Sean Rodriguez||Pirates||-1.3%||Luis Valbuena||Astros||-1.5%|
We hope this table will help Rays catcher Rene Rivera come to terms with why he’s been benched. Kudos to the Yankees and Pirates for winning wild cards despite each being saddled by two infielders who ranked among the five least-valuable players in their respective leagues; Gregorius has improved in the second half, but he’s been a reverse Moreland, batting .326/.394/.511 in 104 PA with margins of more than four runs and only .239/.270/.288 in a larger sample when the Yankees were within one run of their opponent.
Meanwhile, the Rookie of the Year Award isn’t generally regarded as so sensitive to team context — it’s missing that magic word “value” — but cWPA’s picks in each league, Kris Bryant and Miguel Sano, appeared in the big batters table above. And here’s what cWPA says about the most valuable pitchers in each league. Keep in mind that this method gives a leg up to late-inning relievers, who almost always pitch in high-leverage spots.
|National League||American League|
|Zack Greinke||Dodgers||5.0%||Dellin Betances||Yankees||3.4%|
|Clayton Kershaw||Dodgers||4.3%||Dallas Keuchel||Astros||3.0%|
|Jeurys Familia||Mets||3.0%||David Price||Blue Jays||2.9%|
|Jake Arrieta||Cubs||2.9%||Shawn Tolleson||Rangers||2.9%|
|Mark Melancon||Pirates||2.8%||Andrew Miller||Yankees||2.8%|
Championship Probability Added pegs Greinke as this season’s most valuable player, period, contributing 5 percent of a title to L.A. And the Yankees and Pirate on this list help explain how those two teams succeeded despite the Yankees and Pirates in the previous table.
One more list: the least valuable pitchers by cWPA, in both leagues.
After a solid first five months, Pat Neshek has come up short on several occasions in September, bearing as much responsibility as any Astro for the team’s swift decline in the standings. Speaking of relievers who’ve hurt the Astros, if Bud Norris is going to earn American dollars playing baseball, he should respect the game by not being so bad at it. And don’t overlook how hard it was for Kyle Kendrick to place so high on this list, given the Rockies’ lack of meaningful games. The righty really had to prove he belonged, which he did by allowing the NL’s most earned runs and the majors’ most homers in only 136.1 innings.
WPA makes some assumptions that can skew these results: Most notably, pitchers get full credit for every out, so position-player defense doesn’t count. Theoretically, one could apportion credit for every event to the appropriate player, but it would be such an intensive task that one might need to borrow a mystery team’s supercomputer to complete it. Even so, cWPA helps explain some of the disconnect between “best” and “most valuable.” The “most valuable” believers aren’t always misinterpreting or ignoring the numbers; sometimes, they’re just applying a different standard for what the word means. We’ll never agree on what pieces fit into the MVP puzzle, from park factors, to quality of competition, to umpires, to weather, to catcher receiving skills, and even to off-the-field factors like salary, promotional value, and, to borrow a criterion from Chalmers, deportment. But cWPA is a start.
On Monday, Rizzo became the second player to record 30 home runs and 30 hit by pitches in the same season. The first was Don Baylor, who hit 31 homers and was hit 35 times in 1986. Baylor’s 1979 MVP award looks like a big miss: He led the league in RBIs, but he barely ranked in the top 30 in AL WAR. We don’t have cWPA for 1979, but Baylor hit well in the clutch for a playoff team, so the system would value him higher than WAR does, if not highly enough to close a 5-WAR gap between him and the AL leaders. If Rizzo won in 2015, the WARriors wouldn’t approve, but by one of those dictionary definitions, the selection would make sense.
I’ll grudgingly accept an answer other than Harper for NL MVP. But only if that answer is Rizzo.
The cWPA stats in this article reflect action through Tuesday’s games; all other stats are current. Thanks to Dan Hirsch, Dave Studeman, and Hans Van Slooten of Baseball Reference for research assistance.
This post has been updated to correct a few instances where percentage was incorrectly used instead of percentage points.