Looking Back on the 2019 Postseason

The 2019 postseason is history, in a number of senses. The Washington Nationals reached and then won their first ever World Series, half a century after the franchise was born as the Montreal Expos. The loss of the biggest star the Nationals ever had, Bryce Harper, in free agency was only the beginning of a daunting series of challenges the team needed to overcome to reach this pinnacle.

May 23 saw the team mired in a 19-31 hole, just a game and a half ahead of the hapless Miami Marlins. The bottom of the eighth in the NL Wild Card game saw the Nationals down 3-1 to the Milwaukee Brewers with six outs to live. They trailed the NLDS against the L.A. Dodgers 2-1, then trailed the deciding fifth game 3-0 through five and 3-1 through seven. They absorbed a three-game sweep at home to the Houston Astros, leaving them needing to win two in Houston to salvage the World Series. In the all-or-nothing seventh game, they had a 2-0 deficit to surmount after six innings, against a veteran pitcher who was one-hitting them. And they did it all.

I’ll be reviewing the just-past postseason today, as I have since 2014, from the perspective of gauging the excitement of the games rather than the narratives of teams and players. My method for doing this is the Win Percentage Sum (WPS) Index, a system I have used since 2012, when I was doing daily reports on postseason games. I have more recently added a wrinkle to this method, which I will explain soon.

The next several paragraphs summarize my WPS methodology, though you can instead refer to my original articles from August 2012 for a fuller explanation. If you remember this stuff from previous editions, feel free to skip ahead. As always, I will first recognize THT’s Editor Emeritus Dave Studeman for creation of the original game excitement metric. When I conceived my system, I did not know of his similar work, published in the 2007 Hardball Times Annual. Dave graciously gave me leave to continue using WPS, and I gladly give credit where it is due.

WPS is calculated with a base score that adds together the change in game Win Expectancy created by every play, meaning every continuous field action* that changes the base-out state of the game. If a player’s foul-out lowers his team’s chance of winning by five percentage points, or his single raises the chance by five percentage points, that’s a score of five either way.

* In 2017, when intentional walks were made automatic, I stopped adding their Win Expectancy numbers to the WPS score. If there is no action on the field, there is no excitement gained from game play.

To the base score I make two additions, meant to reflect fans’ subjective enjoyment of a game’s high points. I add the three plays with the highest Win Expectancy changes to the base, then add the value of the game’s final play. These reflect, respectively, the highlights of the game and how much the game was in doubt at the very end.

If a big walk-off play is one of the top three, that means it counts three times: in the base, in the highlights, and as the final play. This is intentional. Fans love walk-off wins, especially those swinging a game from defeat to victory, so the WPS system favors them. As an example: Kirk Gibson’s famed walk-off homer in Game One of the 1988 World Series had a Win Percentage Added of 87 percent. In the WPS system, it thus added 261 points to the score of the game, turning a merely good game into an all-time great baseball memory.

The median WPS score of a game is roughly 300, with wide variations running from below 100 to over 1,000. Under 200 denotes a snoozer, while 500 points is my criterion for a “great” game. (The above example, which would have ended in the low 400s if Gibson struck out, came in at 632.5.) The mean of WPS scores is somewhat higher than the median, because there is more numerical room for a game to be great than to be boring. (As you’ll see when we discuss this year’s NLDS.)

The system has blind spots. Team narratives, like the curse-breaking 2016 Cubs, have no impact on the numbers. Outstanding individual performances or milestones likewise count for nothing, except by how they affect the game’s competitiveness. No-hitters and perfect games are hardest-hit: since they usually lack see-saw movement between offense and defense, or between the competing teams, their scores are generally flat. Best-play and final-play bonuses can capture some subjective reactions, but others are beyond a system with any pretension to simplicity.

Another, more equivocal, feature of the system is that it believes more baseball is always better. WPS points can only accumulate, never diminish, with added plays. Thus, extra-inning games have a structural advantage, growing greater the longer the game goes.

Along with the WPS scores, I use a second method to judge playoff games: the CPS system, which I introduced two years ago. Championship Percentage Sum, or CPS, combined the WPS scores with Championship Leverage Index, the brainchild of Sky Andrecheck. Andrecheck’s CLI measures postseason games by how much they move their teams toward, or away from, a World Series victory.

As examples: Game Seven of the World Series is worth 1.0 Championships Added. Game Six of the Series is worth 0.5, the same as Game Seven of a League Championship Series. Game Five of a League Division Series is worth 0.25 ChampAdded, while Games One and Two are worth 0.09375 apiece. Here is the full breakdown of Championship Leverage Indices.

The CLI works very similarly to the in-game Leverage Index. The closer a series stays, the higher the leverage moves. A high WPS-scoring game late in a series will send the CPS soaring, the way a three-run homer breaking a tie in the eighth would score much higher than the same play in the third. This way, a blockbuster Game Seven is considered to make a bigger impression on fans than if it happened in Game Two. This is certainly true, though whether it occurs in the ratios that the CLI uses is debatable.

I believe this is an improved way of gauging the excitement of postseason games, but reasonable fans could disagree. If nothing else, consider it a second eye on the action, providing some more depth to our view of events.

The Games

The Brewers-Nationals showdown shot off some early fireworks with three home runs in the first three innings, but that didn’t produce strong WPS reactions. The five and a half scoreless innings that followed, with a 3-1 Milwaukee lead in place, steered the game toward Snoozer-ville, though scoring threats in the fifth perked things up slightly. The bottom of the eighth saved the game, and the Nationals’ season, with a three-run, 86.1 WPS point rally. The game posted a score of 365.3, pretty good if no instant classic. Wild Card games have been historically dull, though, so this game came in fourth out of the 16 ever played.

Juan Soto’s game-winning knock provides an interesting view of the WPS mechanism that I will touch on briefly. The 0.588 WPA swing the play produced could have been less, and more. Less because of Trent Grisham’s error that let the go-ahead run come home; more because Soto got pickled between second and third for the final out. There were both positive (three runs scoring) and negative (Soto put out) elements to the play that partly cancelled each other, lowering the WPA and the WPS score. This is sometimes a flaw of WPS, but in this case, with the anticlimactic end of the play, it felt like the bug in the program got it right.

The Rays and A’s played a much more typical Wild Card game, sadly meaning it was much more boring. It had a similar start, the visitors mashing two home runs in the first two innings to push ahead 3-0, and was actually more exciting in the early innings than the Brewers and Nats, as Rays starter Charlie Morton had to wriggle out of a few jams. Oakland could never assemble a real rally, though, and the game fizzled to the tune of a 203.8 WPS score. This is 11th-best out of the 16 Wild Card games, meaning a full third of the remainder have been less interesting than this one was.

The divisional round sets a higher standard for excitement, which this year’s edition managed to meet, if just adequately.

To date, 104 LDSes have been played, starting with the 1981 strike season then going 1995 to 2019. This year put a series in each quartile of the WPS results. The Cardinals and Braves played the 11th-best LDS by that measure, while Nationals-Dodgers scored 43rd, Astros-Rays 69th, and the Yankees-Twins sweep a mere 96th of 104, eight places from the cellar. Taking all four series together, their WPS ratings come in 16th of the 26 years of the LCS.

Using the CPS lens shifts things, particularly one thing. The Cardinals-Braves series falls eight places to 19th, the Astros and Rays rise nine slots to 60th, and the Yankees-Twins match-up steps down from 96th to 97th. The Nationals and Dodgers, however, benefit dramatically, their series rising from 43rd place to 14th, leaping past St. Louis/Atlanta to be the most exciting series of the round. This surge lifts the entire round from 16th of 26 up to 12th by CPS.

These mid-range scores came despite ample opportunity for excitement. The four series lasted a combined 18 games, two below the maximum. Only three LDS rounds have ever gone longer (all by a single game), and only four others lasted as long. By that first glance, this year’s edition should have finished higher. The games themselves simply did not measure up, with a few poorly timed clunkers making matters worse.

The WPS average of all the games was 281.4, significantly below the historical mean. Five games in the round suffered WPS scores below 200, including the last three of the Astros-Rays series. In contrast, only three games posted scores above 400.

All of those, though, were clustered in the Cardinals-Braves series, including a “great game” 538 in Game Four. The series also saw the most exciting half-inning and full inning of the postseason. Atlanta’s three-run top of the ninth in Game Three, overcoming a 1-0 deficit in the Braves’ last licks, scored a tremendous 120.9 WPS points, and pushed the full ninth to 144.2. With the lead-up of that inning and the great game that followed, it seemed likely that the series was headed toward a memorable, even historic, finish.

It was, but not the way many fans hoped. The Cardinals’ devastating 10-run first inning put the game away before Atlanta could bat. The game scored a sub-rock bottom 84.2 WPS points. That is not only the worst score ever for an LDS game, it is the worst for any postseason game, from the 1903 World Series to today. Whatever the theoretical minimum WPS score for a game is, this one got very close. Having this game as its decider, with the highest Championship Leverage, is what knocked the series down in the CPS listings.

Contrariwise, it was a merely good, not even great, extra-inning Game Five that lifted the Nationals-Dodgers series in the CPS standings. Just lasting long enough to produce a fifth game is a big potential boost in the CPS system, where the top 14 LDSes all went the limit. Nats-Dodgers was 14th out of the 35 LDSes that went five, despite dull-to-ordinary numbers for the first four games, because of that fine extra-inning Game Five.

Not even length could save the Astros-Rays set by CPS, though. Theirs was the worst five-game LDS in history. Having the one decent game come late in the series rather than early could have raised it another 10 spots or so overall, though that is only the difference between so-so and blah. In a round with two other all-the-way series, it still would have been overshadowed.

The worsts were, sadly, not over for the year.

Let’s clear away the deadwood first. Not only was the Nationals-Cardinals series the least exciting in LCS history—beating out the 2015 NLCS for that dishonor—it comes within a whisker of having been the least exciting postseason series ever. Its 805.2 WPS score for all four games just misses the mark set by the 1989 World Series, which in its four-game A’s sweep of the Giants posted an 805.1. By CPS, and adjusting for the higher scores that World Series games get, the 1989 and 1963 World Series come up as duller, leaving this year’s NLCS the third-worst playoff series of all time.

Ironically, it’s because the Washington-St. Louis games got progressively duller and more lopsided that the series managed not to tally up as the worst ever. The two semi-exciting games came early in the Nationals sweep, giving those games higher championship leverage. That nudged up the overall CPS score, which in a race to the bottom was a fatal handicap.

Still, this was an unfortunate reversion. The years 2015 to 2017 all had a turkey in the LCS round making things dull, before 2018 spared us with a merely mediocre ALCS and an exciting NLCS. This year was the Revenge of the Turkey. It would have taken a classic ALCS to overshadow how flat the Nats and Cards were, and we didn’t get it.

Not that the Astros and Yankees were bad. Their series came 29th out of the 68 best-of-seven LCSes that have been played, both by WPS and CPS. Much of its quality derives from lasting six games: it’s a mere 15th out of 24 LCS series that went that length. That middling finish is a mild surprise, because the three substantially below-average games the series produced were balanced by one great and one almost-great game, the latter coming with the highest Championship Leverage of the series.

The two games took different routes to excellence. Game Two never saw either team lead by more than one, avoided any really poor innings after the scoreless first, and stretched to two extra innings before the walk-off home run. Game Six had substantial stretches of Houston multi-run leads, endured several sub-par innings (three due to double plays killing nascent rallies), only to have a comeback in the ninth electrify the game before the subsequent walk-off home run. Carlos Correa’s and José Altuve’s game-winners were unsurprisingly WPS’s two favorite final plays of the postseason. The latter’s, by CPS, was worth just under one-ninth of a World Series ring all by itself.

Then came the World Series, which produced very different impressions depending on how one looked at it. WPS’s look was one of the most positive.

Overall, this was a persistently, almost aggressively average World Series. One game was good, two games were pretty bad, and the rest were mid-range or a smidgen above. It’s the persistence, into seven games, that raises the series historically. Of the 115 World Series, this one ranked 29th overall by WPS, and 26th by CPS. Compare it to all the seven-out-of-seven World Series*, and the averageness shines through again. It ranks 21th out of 39 such World Series by WPS, and slips to 25th by CPS.

* This excludes the 1920 World Series, which went seven in a best-of-nine, and the 1912 Series, which due to a game called for darkness took eight games for one team to win four.

There were commentators, including some in THT’s online family, who called out this World Series as dull midway through the proceedings. The linked article by Tony Wolfe was written after Game Five, which partly explains the impression he had. Games Four and Five were easily the poorest WPS scorers in the whole series; they followed a Game Three that, though moderately exciting, came with the Nationals leading two games to none, which lowered the Championship Leverage.

There’s another factor that was suppressing subjective impressions of excitement, one that received very much attention. Not one game in the entire World Series was won by the home team. This has a downward effect on WPS, since walk-off wins are a great way for games to produce big point totals. It also has an effect on the kind of fan’s subjective experience that WPS does not really cover: the engagement of the home-town spectators with the game.

It may well be that Wolfe and other observers had tapped into the disappointment of the hometown fans, and made it their own. This makes total sense: human beings do internalize such external cues, though I won’t bore you with talk of mirror neurons and other such physiological details. Better to explain by example.

A full and rollicking stadium is a strong sensorial experience, pulling spectators at home into the excitement it’s feeling and expressing. A full yet subdued stadium, its excitement muted by the reverses the home team has suffered, has the opposite effect. An ill-attended game with silent stands would be even worse in this regard, but that’s not something likely to affect a World Series game, or any postseason game.

These facts made me wonder whether WPS would better reflect our subjective reactions to ballgames if I weighted the scores in favor of events positive for the home team, and against those positive to the visitors. I did not wonder long, though, as there are just so many problems with that idea.

Leave aside choosing how much of a weighting I would give. Never mind figuring out whether I should index the weighting to the attendance at the game, reflecting the magnitude of the crowd’s effect on observers, or whether to throw in multipliers for postseason games. Instead, think about how one can possibly equate a World Series game in 1909, when you had to be in the stands to experience directly the excitement of the game*, to a World Series game in 2019, where you might be in the ballpark, or watching at home, or listening on the radio, or looking at the pitch-by-pitch Gameday feed**.

* There were “electric scoreboards” at the time, recreating the game from telegraph reports, but these would have taken reactions from the in-stadium crowd totally out of the equation.

** This is our modern equivalent to those delightfully quaint electric scoreboards, notably in how it separates us from the reactions of the fans at the game.

This is a perfect example of why, despite the WPS system’s weaknesses and blind spots, it needs to stay simple to be useful. Adding too many complications means adding too many assumptions, ones that may not hold up.

Analyzing the games of this World Series looks unpromising at first, given the lack of outliers positive or negative. One thing that does stand out, though, is how the games managed to have average excitement when the final scores were so lopsided so often. The mean result in the Series was a 7-2 final, exactly: 49 runs for the winning teams, 14 for the losing teams. That should have been reliably dull, but it wasn’t.

A large part of the explanation is that the games seldom turned into blowouts early. Game Two, which ended 12-3, was tied through six innings. The 8-1 Game Four was just 4-1 through six; the 7-1 Game Five was 4-1 through seven. Game Six was 3-2 after six, before Washington piled on four more runs, and Game Seven was 2-0 after six, before the trailing Nationals stormed ahead with six unanswered runs. These games had long stretches of decent leverage or better, letting WPS scores rise though not soar.

Game Three managed its above-average score due to five innings more active than the score indicated. From the second to the sixth, the Astros and Nationals had multiple baserunners in nine out of 10 opportunities, consistently raising the threat of scoring and thus the WPS figures. Houston converted four of its chances into single runs, while Washington, in a game where its batters went a dire zero-for-10 with runners in scoring position, managed only one run. The offenses, perhaps exhausted at that point, gave out for the rest of the game, leaving the WPS score firmly in the middle range.

Spectators, at least those not in their home teams’ stands, got pretty lucky. Several games in this last World Series were more than the sum of their parts. In game excitement as much as in team offense, sequencing matters, even if you can’t do too much to help it along.

On the Clock

My postseason retrospectives have habitually included an examination of the lengths of playoff games, as pace of play has been a mounting concern of mine. This year saw 2018’s remission in the lengthening of regular-season games unwind itself and more. Playoff games, in contrast, got noticeably shorter—but deceptively so, as the pace of the games barely budged.

How do we square the circle of playoff games getting shorter while time per inning dropped only about two seconds? The cause is something that helps explain why this year’s games seemed, and perhaps were, blander than in years past: a dearth of extra-inning contests. Just three of this year’s 37 postseason games went to extras, for a total of four innings. That is fewer than half the extra innings that we got in one World Series game last year.

Without that 18-inning Dodgers-Red Sox marathon bending the data, average postseason game times would be much flatter over the last three years. By time per inning, they are virtually flat. Baseball—or Commissioner Manfred, if you care to personalize it—has at least temporarily halted the trend of ever-longer playoff games. The gap between regular-season pace and postseason pace also narrowed, from 4.09 minutes last year to 3.19 minutes this year, a hair smaller than 2017’s margin.

These are, however, pyrrhic victories. Regular season games broke the 3:10 barrier in 2019, where just four years ago they had been averaging under 3:01. The gains made by restricting mound visits and tightening sponsor breaks between innings in 2018 dissolved in 2019. The game is longer than ever.

In postseason recaps past, I have made quite a fool of myself trying to predict how Commissioner Manfred would address the mounting crisis of game times. I confidently predicted the introduction of the pitch clock for this season, and for the season before. I should be glad to have been wrong—I oppose the pitch clock aesthetically; umpires could enforce pitchers’ pace without the technological intrusion—but the reforms he has introduced have not held back the tide.

Were I to offer another ill-advised prediction this year—not that I’m being so foolish—my guess would be that Manfred will do … nothing. He will have his attention diverted to baseball’s other crisis, that being the flubber ball. He will get something done to reduce the inter-continental carry of the baseball, while denying that any alterations have been made. Home run rates will inch downward. The average game time will inch upward, again.

Oh well. At least, with just three playoff games this year going into the 10th, we will be spared the temptation for Manfred to introduce the minor leagues’ “runner on second” rule for extra innings. And no, that’s not a prediction either. If it were, it would all but guarantee that he instituted the rule. See, I’ve learned my lesson. I’m not even going to mention the reforms Manfred could import from their tryouts in the Atlantic League, like robo-umpires. There was no clamor to see those during the postseason, right? We don’t have to worry about those for a while, right?

Oh.

So, with no special insight into what’s to come, I’ll just have to watch next year to see what happens. As will we all.

References and Resources

• FanGraphs for 2019 WPA data, and for ”Why Has the World Series Felt Boring?” by Tony Wolfe
• Baseball-Reference for game time and historical WPA data
• “The One About the 2013 Postseason,” Brad Johnson, The Hardball Times Baseball Annual 2014, for the postseason Championship Value table
• “Automated strike zone to debut in Minor League Baseball in 2020, Rob Manfred says” The Athletic, by Evan Drellich