June 23, 2015
The Lost Art of the Forfeit
The complete game, the seven-man bullpen, and the batted ball in play aren’t the only elements of baseball that are dying. It’s been twenty years since a baseball team last forfeited a game, the longest such streak in the game’s history. Forfeits became endangered a long time ago: according to Retrosheet, there have been 139 artificial endings in baseball history, and only ten came after the start of the Great Depression. After a brief revival during turbulent post-Watergate America, they appear to be as dead as the bullpen cart.
The decline isn’t difficult to understand. Forfeits have fallen into three major categories, two of which have been dulled by time and technology. The vast majority of them arose from delaying tactics, teams hoping to stretch games into the maws of rain or darkness. (Pacing is still a thriving tactic, but now it’s moved onto the micro-level until the pitch clocks come.) The other involuntary forfeits arise from the angry mob, fans pouring onto the field or making ranged attacks. Lights, domes, fences and rules have all but eliminated these possibilities; the last forfeit in 1995 occurred when fans threw souvenir baseballs into play, forcing the MLB to mandate that potential projectiles must be given out at the exit gates.
Voluntary forfeits were more common in the barbarous age of the game, thanks in no small part to the existence of John McGraw. Little Napoleon was a master of getting ejected and then refusing to leave the field, forcing a forfeit, always when his team was behind. Once he locked an umpire out of the stadium as thanks for the previous day’s performance; the umpire called the game a forfeit from the street. The most realistic example of the voluntary forfeit is one of the most recent: during a rainstorm and a 4-0 deficit, Orioles manager Earl Weaver complained about tarps covering the foul territory bullpen mounds in left field. When the umpire refused to remove them, Weaver pled the safety of his left fielder and refused to take the field.
It seems as though baseball has essentially solved the problem of the forfeit: it has criminalized fan interference, fined manager and player misbehavior, and disincentivized most forms of stalling. Barring a player storming the crowd, we may never see a called game again. Even Weaver’s example seems unlikely, since points of protest no longer need to be sent by train; feeds are live and deputy commissioners are a text message away.
This leads to a simple, dumb question: what would it take for a team to give up?
The biggest blowout in baseball history took place on August 22, 2007, when the Texas Rangers defeated the Baltimore Orioles by a score of 30-3. It’s a deceptive game, because after five innings the game was still only 5-3. Two outs later, the Rangers were up by nine and held a win probability of 100 percent, give or take some minor rounding. Miraculously, the Orioles only needed three relievers to go the rest of the way, despite the fact that they combined to give up 24 earned runs. Even giving up an MLB-record 30 runs didn’t create talk of throwing in the towel.
How many runs would it take?
It’s one of baseball’s virtues that there is no clock, that no matter the game state, the losing team always has a chance to come back and win. At some point, though, that chance becomes essentially meaningless. Despite the hours, despite the lack of suspense or entertainment value, no one thinks to stop. In terms of WPA, there’s no more reason to stop than there is to go on.
The problem is that there is a reason to stop: tomorrow.
On the Effectively Wild podcast, Sam Miller once noted that he would prefer seeing a running win probability on the television than the score. WPA isn’t a difficult concept for even the casual fan; after all, poker fans have subsisted on it for years. But every game actually has a second WPA hidden behind the first. Let’s call it tWPA: the winning percentage for tomorrow’s contest. Ordinarily this would be set at 50%, since every game is a blank slate. But this is not technically the case.
On the most basic level, every single pitch, every single ball in play has a slightly negative value, because it affords the opportunity of injury. It’s the NFL preseason philosophy. In a blowout, this effect is lessened by the presence of backups during garbage time, but in the age of eight-man bullpens there aren’t enough lifeboats to save all the position players. Some starter or another is going to run the very slight chance of hurting himself.
The ravages of time also play a role: ballplayers are people, and tend to be cranky (and creaky) when they get less sleep. This might force a manager to run with a suboptimal lineup, with rested reserves being more valuable than somnambulistic starters. This is especially true of catchers.
Finally, and perhaps most obviously, a blowout can ravage a bullpen. The Orioles escaped their shame with only four pitchers consumed; they were lucky, in that sense. The 1996 Orioles, who lost to the same Rangers 26-7, went through six arms, each of whom threw 20-plus pitches. When the following day’s starter, Jimmy Haynes, only managed four innings, the team was forced to throw in a replacement-level AAA pitcher, called up that morning, to go the rest of the way.
It’s easy to calculate a generic form of WPA: take the game state, look at the results of other games at that same point, and done. Unfortunately, tWPA is context-driven and thus far more difficult to quantify. You could calculate the injury risk of any given pitch for pitchers or dive for fielders with a simple formula: (number of injuries divided by number of pitches or plays). Player attrition is much more difficult to calculate: it’s theoretically possible to scrape the standings for the every game following one with over a certain number of pitches or plate appearances. Ultimately, tWPA becomes the worst kind of statistic, the unquantifiable kind, but it still exists. Even Weaver’s forfeit is based on the same internalized calculus: the chances of winning this game are less than the chances of losing another game based on a potential injury.
But there’s a barrier to forfeiture that goes beyond that simple balance sheet. The decision would come at additional cost, if only for the fans that paid to see their nine innings. A manager who quit a game for the sole purpose of adding a 1% chance of winning a different game would be reviled, mocked, and fined. Like any successful politician faced with a courageous decision, they’d have to look like they were forced into it.
Manny Alexander got the last two outs of that 1996 game for the Orioles, while also giving up five runs on four walks and a homer, in that order. One would ordinarily assume that a team could simply sacrifice position players on the mound’s altar and let the BABIP gods deliver the team to the end of the game. However, Jeff Francoeur’s eighth inning several days ago proved that every man on the mound, no matter how much gas they throw, only has so much of a tank. It’s embarrassing for the Phillies that they had to resort to semaphore to signal the reserve troops, but it’s also valuable for providing us with a benchmark for exactly how much we could expect out of a position player pitcher.
Between 1988 and 2015, the average position player pitcher has required 6.09 pitches per out, not a terrible ratio; they tend to be wild, but hitters in blowouts tend to swing freely. In extra-inning games, when the position player is pitching out of desperation rather than despair, that number rises to 7.16. But we’re not interested in the realistic. We’re creating a hypothetical situation where everything goes wrong. So we’ll look at the situations where the pitcher just didn’t have it, where they gave up at least three runs. In those cases, it took 12.24 pitches to earn each out. Thirty pitches are plenty for any reliever, let alone one who hasn’t pitched since high school.
Only three times in the past thirty years has a team used two position players on the mound in the same game; twice were last week. But even so, not all non-pitchers are equal: there’s massive selection bias in the figures we do see, as teams are likely to use a guy who pitched in college, or has the best throwing arm. Few rosters are likely to carry more than a couple of arms capable of throwing even occasional strikes, and those players are even more likely than their specialized colleagues to hurt themselves, like Jose Canseco, through exhaustion and poor mechanics.
Our flag-waving scenario is of an already-tired bullpen with multiple innings of time to fill, and a string of relievers and non-relievers who simply don’t have it. The starter lasts two innings and gives up nine. Four men in the bullpen aren’t available after throwing 25+ pitches the previous day. The other four throw an inning apiece, each getting shelled as though they’re jumping out of the trenches at the first battle of Amiens. They give up five runs apiece. Now it’s 29-6 after six innings, and the manager must make a choice: throw out potentially three position players? Warm up the starter on his throw day, and force the team to call up Shaun Marcum from the void to start? Let that last reliever throw until his arm falls off?
At this unlikely stage, from a game theory perspective, forfeit is highly preferable to gutting it out. The manager could even simply call it and walk away without getting fired, down 22, though he’d be burned in effigy by sports radio for months. Or, if he’s wise, he can channel his inner McGraw, and find something, anything, to pick a fight with the umpire about. Start a brawl. Declare the turf unsafe, or full of hidden magnets. Have the pitcher throw a beanball directly into the opposing team’s dugout. Anything to get away.
Forfeits, in general, are bad. A glance at the forfeitures of baseball’s adolescence, the silly things that caused games to halt and fans to ask for their money back, leave the impression of unprofessionalism. But while the game is often credited for its supernatural design and balance, the possible value of quitting—in terms of maximizing the following day’s win probability—is unfortunately real. And just like the other aesthetically displeasing aspects of our perfect game, the exponential, bacterial growth of relief pitchers, strikeouts and service time manipulation, voluntary forfeiture is a tactic that someone will think to use someday, when the ends appear to justify it. Gird yourself accordingly.