Every year when the postseason rolls around, we enjoyers of baseball try our best to make sure we’re properly appreciating the history unfolding on the field before us. We want to acknowledge when we’ve just watched a game so magical that it will be spoken of in tones of awe and disbelief for years to come. Downstream of that, we like to evaluate whether a game, a series, or even an entire postseason was a good one, mentally sorting them into tiers with other postseasons we’ve watched. Some measures of “good” are subjective, coming down to our personal preferences for certain strategies, styles of play, narratives, teams, or players. Other measures are more universally agreed upon and objectively quantifiable. In particular, most neutral observers value a close, exciting game, one that features both tension and action to keep observers engaged.
Win Probability Added (WPA) provides a reasonable proxy for measuring both tension and excitement. At the plate appearance level, it uses the score, inning, and base-out state (i.e. runner on second, two outs) to calculate a team’s win expectancy based on historical outcomes. The difference in a team’s win expectancy after a plate appearance relative to what it was before it represents the WPA during the plate appearance in question. WPA will be negative for the team whose odds of winning decreased while being positive for their opponent, but in this context, we’re going to focus on the magnitude of the change in win expectancy. Without a rooting interest, it’s less about which team wins and more about seeing big plays that impact the outcome of the game. Games with a large quantity of WPA have a lot of high-impact plays and lead changes that allow teams to pass win probability back and forth between one another.
Using WPA, we can evaluate the quality of the action in a given game by both looking at the average WPA per plate appearance and by adding up the game’s total WPA. Both methods provide useful insight. Average WPA per plate appearance controls for the variable number of plate appearances in a game, since games with more plate appearances have more opportunities to accumulate WPA. Sometimes that accumulation constitutes empty calories; other times it’s more substantial. Ultimately, we want the games that top the charts from both perspectives.
When comparing games in this manner, extra innings games have the scales tilted in their favor. More innings means more plate appearances, so total WPA gets an obvious bump. But average WPA gets inflated as well. WPA does consider the inning, so any innings beyond the ninth must be handled as if they are the last inning of the game, making individual plate appearances more impactful due to the limited opportunities for an opponent to respond in the event of any scoring. This isn’t necessarily a problem — many extra inning games are more exciting than regulation games — but some extra inning affairs are a real slog. We don’t want to automatically crown Game 3 of the 2018 World Series between the Red Sox and Dodgers as the most exciting postseason game just because it rambled on for 18 innings. Thus, we’ll consider extra inning games in their own category when necessary.
I was able to pull WPA for postseason games going back to 2002. The typical postseason game from 2002 through 2024 averaged 3.3% WPA per plate appearance and 263.5% total WPA. That number tops 100% because we’re adding up the magnitude of the change in win expectancy and ignoring whether it’s a positive or negative change, so a team may exchange large chunks of win expectancy with its opponent several times over the course of a game. At the low end, average WPA checks in at 0.7% per plate appearance, while total WPA bottoms out at 60.0%. Meanwhile, the highest observed values are 7.5% WPA per plate appearance and 791.3% total WPA.
But before we get to the best and worst individual games, part of why we want to evaluate whether an individual postseason game is good is to feed into the larger narrative of whether that postseason as a whole is any good. Compiling the average change in WPA per plate appearance across the entire postseason instead of just a single game gives us a way to rank recent postseasons from Banger to Snoozer (given the changing playoff format, comparing total WPA doesn’t really work as desired).
At the far end of the Banger side of the spectrum is 2009, with a per-plate appearance average WPA of 3.7%. This was famously the last time the Yankees won the World Series, but WPA is just numbers, so we can’t blame it for having landed the top spot on east coast media bias. The individual play with the largest impact on a team’s WPA that year came from Game 4 of the NLCS between the Phillies and the Dodgers. With Jonathan Broxton pitching for Los Angeles, Jimmy Rollins hit a walk-off RBI double to score Eric Bruntlett (the pinch-runner for Matt Stairs) and Carlos Ruiz (82.7% WPA) and give the Phillies a 3-1 series lead:
But the individual game from 2009 with the biggest swings in WPA was a matchup between Philadelphia and Colorado (teenagers may not realize this, but the Rockies did once play exciting games in the playoffs). This was the fourth and decisive game of an NLDS matchup that ended in a 5-4 Phillies win. The WPA highlights included Jason Giambi hitting an RBI single off Ryan Madson to tie the game with two outs in the bottom of the eighth (28.1% WPA) and Yorvit Torrealba following that with a double to extend Colorado’s lead to 4-2 (33.0% WPA).
But in the top of the ninth, Philadelphia scored three runs of their own off Rockies reliever Huston Street on a two-RBI double from Ryan Howard (37.3% WPA), who came around to score one batter later on a single from Jayson Werth (33.5% WPA). In the bottom of the ninth, with Todd Helton and Carlos González already on base, Brad Lidge got Troy Tulowitzki to strike out waving at a low-and-away slider to end the game (18.8% WPA):
The postseason that most masqueraded as melatonin according to WPA came in 2023. The per-plate appearance WPA average in this one was 3.0%. That might not sound much lower than 2009’s 3.7%, but it’s two full standard deviations below average compared to other years. To refresh your memory, the 2023 postseason ended with the Rangers triumphing over the Diamondbacks, a matchup casual fans and TV execs may not have cared for, but again, WPA doesn’t know who is playing and it certainly doesn’t care about TV ratings.
The single most WPA-shifting play happened in the bottom of the eighth inning of Game 2 of the NLDS series between Atlanta and Philadelphia, when Austin Riley sent a Jeff Hoffman slider into the seats for a game-winning, two-run homer, good for a 55.2% change in WPA, compared to an 82.7% change in WPA for the top play in 2009:
Across 41 games in 2023, there were 145 plays that sparked a shift in WPA of 10% or more. In 2009, a postseason spanning just 30 games, 165 plays resulted in a 10% or larger shift in win expectancy, 20 more than 2023 in 11 fewer games.
But 2023 wasn’t without exciting individual games. The best example was Game 5 of the ALCS between the Astros and the Rangers, with an average WPA per plate appearance of 5.3%. The matchup featured the Astros coming from behind, scoring three runs in the top of the ninth to take a 5-4 lead that would hold up and give Houston a 3-2 lead in the series. All three runs scored on a Jose Altuve home run off José Leclerc, who came in to get the final out of the eighth and stayed in to pitch the ninth, yielding a single to Yainer Diaz (7.7% WPA) and a walk to Jon Singleton (12.2% WPA) before Altuve took him deep on a low-and-inside changeup (55.1% WPA). Mitch Garver and Jonah Heim both singled to start the bottom of the ninth (13.5% and 18.4% WPA, respectively), but Ryan Pressly retired Marcus Semien (18.6% WPA), Corey Seager (16.8% WPA), and Evan Carter (17.8%) to seal the game for the ‘Stros:
The graph below shows how average WPA per plate appearnace has changed from one year to the next and evolved over time. Large spikes and dips are common, but it does seem that the average has trended down in recent years:
As we zoom in on individual games, let’s start in the recent past and identify the most and least exciting games from the 2024 postseason. The Snoozefest 2024 award goes to the Detroit Tigers facing off against the Cleveland Guardians on October 5. The total WPA in this one was 94.4%. A number that close to 100% without going over basically indicates that each subsequent plate appearance was simply one small step closer to the eventual outcome. On a per plate appearance basis, this game averaged 1.3% WPA. No wild swings, no barrel-rolls, definitely no attempts at a Biles II.
The hardest hit ball was a groundout off the bat of Brayan Rocchio at 108.7 mph; the next three hardest hit balls were also groundouts. The fifth-most well-struck ball was a three-run home run off the bat of Lane Thomas (8.8% WPA) that put the Guardians up 5-0. In the first inning. Cleveland tacked on two additional runs in the sixth to win 7-0. The Thomas homer was the second-largest WPA event of the game, only outdone by a double from José Ramírez earlier in the inning (13.0% WPA). The Tigers managed just four hits and two walks against a parade of five pitchers that began with Tanner Bibee throwing 4 2/3 innings and ended with Emmanuel Clase tossing a blanket over everyone who fell asleep on the couch before turning out the lights on his way to bed:
On the exciting end of things, two games made strong cases for staying up past one’s bedtime. In terms of total WPA, the Yankees taking on the Guardians in Game 3 of the ALCS in Cleveland was the leader with 506.7%, a total helped by the game’s jaunt into extra innings. Second place on the leaderboard with 498.2% WPA was also an extra innings game — Game 1 of the World Series in Los Angeles. But Yankees/Dodgers also led the way in terms of WPA per plate appearance, with 5.7%, while Yankees/Guardians clocked in with an average WPA of 5.3% per plate appearance. Awarding a close call to the World Series game feels like a safe rule of thumb. What’s more, it was the first game of the series compared to the third game of a series where the Yankees entered with a 2-0 lead, making their loss in Cleveland less troubling than their loss in Los Angeles. Plus, the game against the Dodgers had 12 plays with swings in WPA of 10% or higher, while the game against the Guardians logged just 10 such plays. Fortunately, no one is actually making us choose between these games. We get to enjoy both.
In the game in Cleveland, the excitement started with a two-run home run from Kyle Manzardo in the third (22.0% WPA). After that, things stayed pretty quiet until the eighth, when Aaron Judge and Giancarlo Stanton hit back-to-back homers (32.9% and 29.6% WPA, respectively) in the top half of the inning. In the bottom half, Luke Weaver came in to strike out David Fry for the final out (12.0% WPA). Weaver stayed in the game to pitch the ninth and got a crucial double play (16.2% WPA), only to get lit up like a Christmas tree two batters later by Jhonkensy Noel (48.8% WPA). And then, the final swing of the night: Fry turned a sinker from Clay Holmes into a walk off blast (37.5% WPA):
In the game in Los Angeles, several of the larger shifts in WPA occurred when the pitching team wriggled its way out of a jam. With Tommy Edman on third in the sixth, Gerrit Cole got a grounder from Mookie Betts for the second out of the inning (10.7% WPA). In the seventh, Holmes got Will Smith to pop out with runners on second and third (16.8% WPA), then Tommy Kahnle came in to face the next batter and got Gavin Lux to ground out to end the inning (13.1% WPA). In the ninth, Blake Treinen got the third out of the inning on a Judge popup with Gleyber Torres on second (11.1%).
On offense, the big swings in WPA came from Stanton’s two-RBI homer in the sixth to give the Yankees a one-run lead (30.7% WPA), Shohei Ohtani’s double in the eighth that turned into a triple after a Juan Soto throwing error (21.0% WPA), and an RBI fielder’s choice from Anthony Volpe in the top of the 10th (13.7%) to break the tie. Of course, the real heavyweight carrying this game’s WPA was the walk-off grand slam from Freddie Freeman in the bottom of the 10th (73.3% WPA):
In considering games whose numbers weren’t juiced by extra innings, the top WPA-getter was Game 1 of the ALDS between the Yankees and Royals, with an average WPA of 5.5% per plate appearance and 441.6% WPA total. The WPA movers in this one were dueling two-run, lead-changing home runs in consecutive half innings from Torres (19.5% WPA, bottom of the third) and MJ Melendez (22.8% WPA, top of the fourth), a pair of bases loaded walks from the Royals’ bullpen in the fifth (11.3% and 13.7% WPA), and some sloppy defense from the Yankees that allowed Yuli Gurriel to advance to third on a throwing error by Volpe (14.0% WPA) and then score on a Garrett Hampson single (18.6% WPA) to give the Royals a 5-4 lead in the sixth. But the Yankees got the last WPA laugh by hitting RBI singles in both the sixth (19.7% WPA by Austin Wells) and the seventh (21.8% WPA by Alex Verdugo) to secure a 6-5 win:
Digging deeper into the past, the WPA-approved, most exciting game since 2002 was Game 1 of the Wild Card matchup between the Reds and the Braves in 2020. This game was tied 0-0 all the way through to the bottom of the 13th, meaning if anyone so much as sneezed on the basepaths, the WPA needle was moving. The average per-plate appearance WPA in this game sat at 7.5% and the total WPA came in at 791.3%, ranking first in both categories.
As one might expect from a low-scoring game, most of the largest WPA movers in this one were defensive plays, such as Lucas Sims striking out Riley with a runner on second to end the 10th (28.2% WPA), Tyler Matzek getting Mike Moustakas to strike out with the bases loaded for the final out in the 11th (34.9%), and then Matzek replicating that performance against Freddy Galvis in the 12th to log the third out with runners on second and third (32.4% WPA). In the bottom of the 12th, Michael Lorenzen ended the inning by striking out Dansby Swanson with a runner on second (28.2% WPA). Finally, in the 13th, A.J. Minter ended the Reds’ scoring threat by getting Jose Barrero to ground out with the bases loaded (34.9% WPA); the Braves walked it off in the next half inning on a Freeman single (17.1% WPA) that scored Cristian Pache, who pinch ran for Nick Markakis following his single to open the frame (9.8% WPA), then advanced on a single from Riley (10.2% WPA) and a fielder’s choice off the bat of Ronald Acuña Jr. (0.5% WPA):
But since extra inning games provide more opportunities to add to the WPA total and skew the average with the inflated WPA numbers inherent to innings with potential to be the last, it’s worth shouting out the most exciting game that wrapped it up in regulation. The greatest nine-inning Banger also happened in 2020. Game 4 of the World Series, in which the Rays defeated the Dodgers by a score of 8-7 to tie the series 2-2, averaged 7.2% WPA per plate appearance and totaled 615.5% WPA. This game featured seven plays that either tied the game or put the batting team ahead, all of which happened in the bottom of the sixth or later.
A summary of the largest fluctuations in win probability starts with a three-run home run by Brandon Lowe (40.6% WPA) that scored Randy Arozarena and Ji Man Choi to give the Rays their first lead of the game in the sixth. But Joc Pederson snatched the lead back for the Dodgers in the top of the seventh with an RBI-single (29.4% WPA) that scored Seager and Justin Turner. However, the biggest bang of the night came in the ninth, when Brett Phillips (who entered the game as a pinch-runner for Choi in the eighth) hit a single off closer Kenley Jansen that, combined with poor defense from the Dodgers, scored Kevin Kiermaier and Arozarena to earn the Rays a walk-off win (83.1% WPA):
And lastly, the historical Snore Fest of Note Award goes to Game 5 of the 2019 NLDS matchup between the Cardinals and the Braves, which averaged 0.7% WPA per plate appearance and totaled 59.9% WPA. This one was a blowout in which the Cardinals scored 10 runs in the first inning on four walks (with one issued to starting pitcher Jack Flaherty, who had yet to throw a pitch), two singles, three doubles, and a smattering of other mishaps, but startlingly, no home runs. The Cardinals amassed 49.9% of WPA across 14 batters and didn’t see their win probability drop below 98% for the remainder of the game. At least everyone had the option to get to bed early, especially since this one started at 5:00 p.m. Eastern:
Having now seen which games WPA found the most and least exciting, I’m sure many of you have specific games in mind that you found more memorable, more detrimental to your blood pressure, more pleasing to the dopamine receptors in your brain. Maybe the game you’re thinking of factored more heavily into the overall outcome of a series, or played a bigger role in determining the eventual champion. There’s certainly a version of this exercise that incorporates championship WPA and/or any number of other factors. This is just one way to go about evaluating the quality of a playoff game or a postseason as a whole. It acts as a proxy for the thing that actually makes a baseball game or a movie or a book or a song good: when it makes us feel stuff and allows us to connect with other human beings.
