Game Five Was as Weird as It Felt

As illustrated by basically all the win-probability graphs featured here on any given night of the season, the average baseball game tends to be composed mostly of plays that, individually, have little influence on the outcome but which, when taken collectively, push the game towards one conclusion or the other.

This graph from the Brewers-Cardinals game on the last day of the season illustrates the point:

You can see Brett Phillips‘ home run annotated here. That one play shifted the probability of victory about 20% in Milwaukee’s direction. Other than that one event, however, the game is defined mostly by a series of small ups and downs before it flatlines in favor of the Brewers in the ninth.

This is the how these win-probability graphs typically work. There aren’t often moments where, based on one play, you tell yourself, This game is over — or, alternatively, Wait a second, this changes everything. There’s usually some build-up, an accumulation of smaller moments leading to a bigger one.

This World Series, however, has abandoned the slow burn. It continues to produce big moments at an unprecedented rate.

If we define a “big moment” as the sort of play the produces a win-probability change of 25% or greater — that is, a play that brings a club back from precipice of defeat or, alternatively, appears to render a tight battle over — we find that 96 of the 113 World Series played since 1903 have featured four or fewer “big moments” over the course of an entire series. This year’s World Series matchup between the Los Angeles Dodgers and the Houston Astros has beaten that number this series… in a single game… twice.

*Setting the bar at .25 is admittedly somewhat arbitrary, but going too much higher increases the rarity of something that doesn’t occur all that often. Even using the threshold of .33, for example, would reduce the number of instances by half. Similarly, lowering the bar to .18 would double the amount of plays. Although slightly haphazard, .25 seems to set a decent balance.

When the Dodgers and Astros combined for five game-changing plays in Game 2 of this series, it was unprecedented, and I wrote about it at length. To briefly recap: prior to Game 2, there were just three World Series games with four plays featuring a WPA of at least .25. Only 11 other World Series games had even had three such moments. Game 2 had five of those big, single-plays that change a game.Sunday’s night’s Game 5 matched that number and produced an even wilder result than we saw just last week.

As I did after Game 2 to provide some context, here are the biggest plays of this year’s postseason by WPA, with Sunday night’s game now included.

Of the 11 biggest plays in this year’s playoffs, eight have come during the World Series. The fact that it’s the World Series doesn’t make these plays more likely. Any individual game can produce dramatic swings — only one in three actually do have plays with a WPA of at least .25 — but this matchup has produced more dramatic moments than we’ve seen throughout the rest of the playoffs. Individual plays aren’t the end-all-be-all of excitement. In the bottom of the seventh inning on Sunday, for example, the Astros increased their win expectancy from 35% at the beginning of the inning to 93% by the end due to multiple plays of significance, but no single play did as much damage as the plays in the chart above.

When we consider the sheer number of plays we’ve seen in this World Series that have changed win probability by 25%, it blows away every other series we’ve seen, as the graph below shows.

The World Series of 1912, which went eight games due to a tie in Game 2, featured eight big moments. Fred Merkle, who is known mostly for his “boner” in the 1908 series, very well could have been the hero in Game 8 after recording an RBI single off of Smoky Joe Wood in the top of the 10th that put the New York Giants ahead 2-1. He didn’t receive the distinction, however, as Tris Speaker produced an RBI single off of Christy Mathewson to tie the game and then a bases loaded sacrifice fly ended the series.

The 1975 World Series, perhaps most famous for Carlton Fisk’s Game 6 homer, also had eight big moments, including an RBI single by Joe Morgan in the top of the ninth of Game 7 that broke a 3-3 tie with two outs. This year, the Dodgers and Astros tied the record in Game 5 when Jose Altuve’s three-run homer tied the game at 7, and the two teams broke it three times before the game ended. Here are some of those same numbers from the graph above, but only showing the leaders.

All the years of the World Series omitted from the chart above — and there are number of them — featured fewer game-changing moments over the entirety of the series than occurred in Game 2 and Game 5 of this year’s Dodgers-Astros matchup alone. More than one-third of all series had either one or zero moments of that magnitude. They weren’t necessarily boring, but they might have lacked for some particularly big moments.

We’ve had eras off good and bad pitching and eras with a lot of homers, but we’ve never had a World Series quite like this one. Maybe the ball is juiced and slippery and maybe the teams and the bullpens are worn down because of the long season and increased usage of late, and maybe everybody is trying to hit it out of the park, but we aren’t just talking about homers here. Three of the five biggest plays on Sunday night were on batted balls that stayed in the park. Two were singles. Because we haven’t looked at it yet, here’s the win-expectancy graph from Sunday.

When looking at Game 2, I couldn’t help but compare it to probably the craziest World Series game anyone has ever seen: Game 6 in 2011. While Game 2 this year had five “big moments,” the moments back in 2011 were bigger at the top, and there was a lot more depth in terms of tension and moments. Game 5 comes a lot closer to matching up with Game 6, which, as an elimination game, carried a bit more weight in terms seasonal leverage.

Here are the top-20 plays in terms of WPA for each game:

Ultimately, even the games from the current World Series — objectively crazy as they’ve been — can’t compete with Game 6 from 2011, which featured three win-probability changes of at least 40%. Where Sunday night’s contest can match up with 2011 is the significance of the plays after the top three. If we simply average the top-10 plays, it comes to 28% for 2011 and 26% for Sunday night. If we go down to the top-20 plays, the averages are 19% and 17%, respectively.

While we might not remember it with all the hits, Sunday’s game also had its share of big outs. In Game 2, there were just two plays where the pitcher’s team increased its chances of winning by at least 10%; on Sunday, there were four. That doesn’t quite measure up to the seven from Game 6 in 2011, but it isn’t too far off. Game 6 also has the 11 plays with a leverage index greater than 4.0, compared to just one in Game 2 and three in Game 5.

Game 2 was crazy. Game 5 was even crazier. We aren’t guaranteed more of the same in the next contest (or two), and odds are actually against it. It felt surreal watching Sunday’s game go back and forth. The numbers support that feeling. Removing emotion from the equation reveals just as crazy an outcome as it felt watching — which for me, enhances my already great appreciation for the game.