Looking Back at the 2015 Postseason

It was a special postseason for the Royals and Salvador Perez. (via Arturo Pardavila III)

It was a special postseason for the Royals and World Series MVP Salvador Perez. (via Arturo Pardavila III)

The 2015 baseball season, and postseason, is part of history now. The Kansas City Royals completed their unfinished business, leaving the New York Mets business of their own to finish. Cubs fans, hoping Back to the Future II was prophetic, instead got a trolling 26 years in the making as their team was eliminated on the day they were “supposed” to win it all. And Matt Harvey’s status as The Dark Knight contended with Jose Bautista’s Bat-Man stylings.

It was a memorable postseason on several levels, but today I will be examining the games themselves for the excitement they provided. I will be doing it with a creation of mine that long-standing THT readers probably know pretty well by now.

Back in 2012 and 2013, I did daily reports on postseason games backed by a metric I call the Win Percentage Sum, or WPS. A change in THT’s format in early 2014 made daily updates impossible, so I switched to a post-postseason wrap-up to keep my WPS work alive, and I am doing it again this year.

Newer readers can scan my original rollout of the WPS system here and here, though I’ll summarize it presently. Original credit for the system to measure game excitement belongs to THT’s honcho emeritus Dave Studeman, who published it in the 2007 Hardball Times Annual. I was unaware of his work when I produced my own similar system, and it is thanks to his forbearance that I continue to use WPS, while acknowledging who beat me to the concept.

WPS is calculated with a base score that adds together the change in game Win Expectancy created by every play, meaning every continuous action that changes the base-out state of the game. For example, if a player’s popout lowers his team’s chance of winning by four percentage points, that’s a score of four. If Carlton Fisk leads off the bottom of the 12th with a home run that takes his team from a 64 percent chance of winning to actually winning, that’s 36 points added to the WPS score.

To the base score I make two additions, meant to reflect fans’ subjective enjoyment of the high points of a game. 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 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. As walk-off wins, especially those going from defeat to victory in one play, are generally remembered as the most exciting, the WPS system favors them.

The average WPS score of a game is roughly 300, with wide variations running from below 100 to over 1,000. Under 200 means a snoozer, while 500 points is my criterion for a “great” game.

The system, inevitably, is not perfect. It cannot measure milestones (3,000 hits, 715 home runs) or outstanding individual performances (four homers, no-hitter) except by how they affect the competitiveness of the game. Indeed, no-hitters and perfect games will generally score lousy because they lack the back-and-forth between offense and defense, or between the competing teams, that builds up good scores. WPS handles the usual excitement of baseball better than the unusual events.

Another blind spot, or if you will an assumption, of the system is that it believes more baseball is always better. WPS points can only accumulate, not be subtracted. This means that extra-inning games have a structural advantage, that grows the longer the game goes.

Examples from the 2015 regular season bear this out. The top five games rated by WPS lasted 19, 18, 17, 14, and 16 innings. The highest-rated game that went nine innings scored 738.4 points, well shy of the 887.9 tallied by the number five game overall.

It might instead be fairer to measure game excitement by rate of WPS point accumulation. Divide WPS score by the number of innings played in a game, and you get the “density” of excitement provided by the game. One can also do it by game time, but although I have toyed with this before, I will hold off for the time being.

I ranked 2015’s regular-season games by this method, to see whether some regulation games got in among the marathons at the top of the standings. In fact, nine-inning games completely took over the top five.

1. 9/23, Orioles @ Rays, 738.4 pts., 82.0 per inning
2. 9/20, Giants @ Padres, 701.0 pts., 77.9 per inning
3. 7/30, Yankees @ Rangers, 658.4 pts., 73.2 per inning
4. 8/31, Astros @ A’s, 658.0 pts., 73.1 per inning
5. 10/3, Tigers @ White Sox, 653.7 pts., 72.6 per inning

A Hardball Times Update
Goodbye for now.

The closest an extra-inning game got to this list was a 12-inning affair between Arizona and San Francisco on April 16, scoring 846.2 WPS points or 70.5 per inning. All of the top five, by the way, were walk-off wins for the home team.

Now, while a long extra-inning game during the regular season can tax even a fan’s patience, the rules do change once the calendar turns to October. The WPS rate stat loses a bit of its rationale when the games are so meaningful that we’ll commit to watching them even into the wee hours of the night. So while WPS per inning has its uses, and I’ll note it at apropos times, pure WPS is still my standard for postseason play.

Before going into this postseason’s WPS numbers, one quick notice. I used FanGraphs’ WPA numbers to work out WPS scores for the games, but for historical games (which for me means before 2012) I use data provided at Baseball-Reference. This leads to small differences in the scores, but certainly nothing that will turn an above-average game below average or vice versa.

Postseason, By the Numbers

The 2015 playoffs began rather weakly, which has unfortunately become a habit with the Wild Card rounds. The Astros’ 3-0 whitewashing of the Yankees scored a drab 201.8 points, while the Cubs’ 4-0 blanking of the Pirates managed just 184.0. Of the eight Wild Card games played over the last four years, only one, last year’s classic Royals-A’s 12-inning showdown, was any better than dead average. Still, these weren’t the worst games the postseason would produce.

The divisional round thankfully provided more excitement, if not a great bounty.

WPS Scores in the 2015 Division Series Round
Game 1 302.3 244.5 201.3 239.6
Game 2 628.9 382.2 190.1 333.4
Game 3 216.4 297.4 332.0 273.3
Game 4 136.5 437.5 363.9 217.7
Game 5 391.4 230.2 327.2
Total 1675.5 1591.8 1087.3 1391.2

Just one of the 19 LDS games passed the threshold of greatness: the Rangers’ 14-inning win over Toronto in Game Two. Going by the WPS per inning measure, it falls short (by 44.9 points/inning to 48.6) of Game Four between the Astros and Royals. That game featured a five-run eighth by Kansas City to push ahead 7-6, right after Houston had scored three to build its lead from one run to four. Big swings like that are a great way to pile up WPS value.

That K.C. eighth worked out as the highest-scoring half-inning of the LDS round, at 105.3 points. Houston’s one-two-three reply meant, though, that it fell short of the top full inning. That went to the second of Game Three between the Dodgers and Mets. Los Angeles scored three, and the Mets replied with four of their own, for 123.5 points. This was a remarkable result for such an early, medium-leverage inning, and spiced as it was by being the first game after the Chase Utley/Ruben Tejada incident, it made for compelling baseball.

Just behind that for one-inning excitement was the seventh inning of Game Five between the Rangers and Blue Jays, at 116.9 points. This was the inning of Russell Martin’s throw off Shin-Soo Choo’s bat, and the Toronto comeback punctuated by Jose Bautista’s tie-smashing home run and ultimate bat-flip. WPS cannot measure the peculiar drama of those moments, so if you want to set the Mets’ and Dodgers’ claim aside in favor of this inning, I cannot say you are wrong.

The set of four series scored fairly well compared to all LDS ever played. Out of 88, they placed 15th, 27th, 41st, and 60th. This good result is mainly due to length, the four coming in just one game under the maximum. Separate them by series length, and the results are more middling. The three five-game series finished seventh, 15th, and 23rd out of 29 five-game LDSes, right smack in the center. The lone four-game series, Chicago and St. Louis, was 23rd out of 29.

This ends up paralleling my caution about extra-inning games. The LDS round this year was pretty exciting, but got there by the length of the series rather than the concentrated excitement within those series. Still, quantity produced a quality all its own.

That did not last.

WPS Scores in the 2015 Championship Series Round
Game 1 201.1 283.7
Game 2 366.0 126.8
Game 3 205.4 321.4
Game 4 159.8 154.2
Game 5 157.4
Game 6 499.8
Total 1589.5 886.1

Well, at least one of them wasn’t the lousiest LCS ever played.

The NLCS had two games that were roughly average, and two that were put to bed early and never woke up again. The series WPS total of 886.1 points makes this, far and away, the least exciting best-of-seven League Championship Series ever held. Second place belongs to the 1990 A’s sweep of the Red Sox at 1,128.2, and third to the 2006 Tigers’ blanking of the A’s at 1,287.4.

Should we include the best-of-five LCSes held from 1969 to 1984, the 2015 NLCS matchup ends up fifth-worst. This looks less terrible, but consider: nine out of 13 three-game sweeps provided more excitement than the Mets’ four-game sweep of the Cubs did.

Toronto and Kansas City gave us something better, including a concluding game that hangs right on the cusp of greatness. Do you give the game the benefit of the doubt on account of Lorenzo Cain’s Mad Dash Home, or penalize it for the rain that postponed the final two half-innings for an hour? That’s a matter of personal tastes, not to mention rooting interests. WPS scores can be a supplement to our understanding, but it doesn’t replace watching the games themselves.

The top of the ninth in that Game Six, scoring 109.0 WPS points, was the highest-rated half-inning in the LCS round. It was also the highest-rated full inning, despite there not being a bottom of the ninth. It edged out, naturally, the eighth inning of the same game, which came in at 104.8 points. Truly an inning and a half to remember.

Still, that was one borderline great game and one good game set against four clunkers. Of the 60 best-of-seven LCSes, this one finished 43rd in WPS. Narrow it to six-game series, and the Blue Jays and Royals finished 19th out of 22. It’s informative to recall that last year’s ALCS finished just 25 WPS points behind this one, despite being a four-game sweep.

Things woke back up in a hurry for the World Series, though perhaps it stayed in too much of a hurry.

WPS Scores in the 2015 World Series
Teams Gm. 1 Gm. 2 Gm. 3 Gm. 4 Gm. 5 Total
NYM/KC 729.0 250.5 253.4 384.6 391.3 2008.8

Before covering the full Series, Game One deserves some special attention. There have now been 646 World Series games played, and the WPS index of this year’s Game One ranks as the eighth best of all time. It beats the Bill Buckner game. It beats the Kirk Gibson game. It beats the Bill Mazeroski game. It also means four of the highest WPS scorers have come in the century of the 2000s*: the others are in 2011, Game 3 in 2005, and Game 1 in 2000.

* It’s not the 21st century, which began in 2001. To quote The X-Files: “Nobody likes a math geek, Scully”—but if that were true, we wouldn’t have half our content.

There is a caveat: Game One went 14 innings. If we go by WPS per inning, where this game scores 52.1 WPS per frame, this contest slides a long way. The first game on the raw WPS list that comes up with a lower WPS/inning rate is the 21st-place Game Two in 1916. I didn’t dig up this game’s exact per-inning ranking, partly because even Scully would find that too math-geeky, and partly because the point is already made without the exact numbers.

While the game was a walk-off win for the Royals, it was the weakest kind of walk-off one could have: bases loaded, nobody out, and the winning run coming across on a sacrifice fly. The Win Expectancy change on that play was a mere 6.6 percent. (This isn’t because of the sac fly: a grand slam would have had the same WPA.) The 2015 postseason would be just the second to end with no walk-off hits since 1989, six years before the wild cards and divisional rounds.

In WPS terms, the whole Series was a reversal of last year’s. The 2014 World Series went seven games, but those games were just slightly above average at best and spectacularly uncompetitive at worst, with three games below 175. This year’s series was over in five, but its floor was just mildly dull, with a very high ceiling and two good games filling out the rest. (Game Five was merely good because Matt Harvey kept suspense fairly suppressed for his first eight innings, and because Kansas City made a total blowout of the last one.)

The 2015 Series ranks 39th in WPS out of the 111 that have been played, moderately above average. Not only does it cruise past last year’s 1,659.8, it has the highest WPS score for a World Series since 2011, which tallied 2,840.1. There’s not too much shame in falling short of the sixth-best Series, by WPS, ever played.

Among five-game Series, it finishes a much more impressive second out of 25, trailing just the 2000 edition. (There have actually been 27 World Series that went five, but two were sweeps with one tie game thrown in.) If the World Series left us wanting more—and outside western Missouri and eastern Kansas, that’s likely the prevailing sentiment—it’s because it was good baseball, as long as it lasted.

The World Series provided us with the highest WPA play of the 2015 postseason. That came in Game One, when KC’s Alex Gordon tied the game in the bottom of the ninth with a solo home run off Mets closer Jeurys Familia. The WPA of the play was 0.472, which translates to 47.2 WPS points—or really 94.4, as it counts both for itself and as one of the game’s top three plays. That’s also the highest of the playoffs, making this, by my method, the most exciting play of the postseason.

The highest-scoring final play of any game can be counted two ways. One method puts it at the end of the ALCS. Toronto was a run down to Kansas City with two gone in the ninth, but had runners at second and third, putting their Win Expectancy at 19.4 percent. Josh Donaldson’s groundout to third changed it to zero. The WPS score ends up a total of 38.8 points: 19.4 for the play, and 19.4 for being the final play. It wasn’t one of the game’s top three plays, though, so it doesn’t get points for that.

This gives an opening to the last play of Game Four in the World Series. Lucas Duda’s one-out soft liner was caught by Mike Moustakas, and caught runner Yoenis Cespedes halfway off first for a game-ending TOOTBLAN* double play. The 16.9 percent WPA of this play gets counted three times in the WPS system, as it was the third-highest scoring play of the game, for a total of 50.7 WPS points. Which of the two plays you consider more exciting depends on what you think of the quirks of the system, not to mention how much you enjoy TOOTBLANs.

* TOOTBLAN: Thrown Out On The Basepaths Like a Nincompoop.

The least exciting game was Game Two of the NLCS, where the Mets scored three runs in the bottom of the first, then put both the Cubs and the game in a sleeper hold. Just once did the Cubs bring more than four batters to the plate in an inning, this happening in the sixth when they were four runs down. Nine of the 17 half-innings in the game went one-two-three. You will seldom find a duller 4-1 contest.

The most exciting game, as said above, was the opener of the World Series. Go by rate of excitement, however, and the aforementioned Game Six of the ALCS wins the prize, by 58.8 to 52.1 WPS points per inning. Alex Gordon’s heroics versus Lorenzo Cain’s: that’s the right note on which to close this section.

Good Timing

In closing last year’s edition of this retrospective, I spent significant time decrying the slowing pace of play in baseball, and how much worse this became when October rolled around. I had even more to say about it in a January article, part of a series by which THT greeted Rob Manfred’s rise to the Commissioner’s office. Manfred acted on the problem (though surely not on my advice), making several reforms to speed up the game going into the 2015 season.

Would this carry over into the postseason? There was some reason to think it wouldn’t. MLB apparently relaxed some of its pace-of-play restrictions for the playoffs, notably giving batters freer rein to step out of the batter’s box during an at-bat. This may have been the continuation of a relaxation that crept in during the season.

Another factor threatening longer postseason games was an expansion of managers’ challenge privileges. Instead of having one challenge available for the first six innings of a game (umpires can initiate a review on their own initiative, or a manager’s, after that), they received two challenges. This became significant, as three times managers made a second challenge they otherwise could not have directed be made. (One other second challenge came after the first brought a reversal, in which case the manager retained his challenge.)

What ended up making more difference is that managers, and umpires, were much likelier to call for reviews in a situation where there was so much more to gain or lose. In the regular season, 2,429 games produced 265 reviews. In the postseason, 36 games had 27 reviews, a per-game rate almost seven times greater. The time expended was not huge—47:33 total for the 27 reviews—but little things do add up.

This postseason, however, it didn’t add up excessively. Following the trend set in the regular season, the playoffs saw a notable and welcome quickening of pace. Rather than recapitulate numbers and comparisons I gave last year and add this season’s tallies, I’ll distill things into (hopefully) handy tabular form.

Average Pace of Play from 2013 to 2015
Year Reg. Time Post. Time Reg. Min./Inn. Post. Min./Inn.
2013 3:04:14 3:23:32 20.52 23.06
2014 3:07:48 3:37:53 20.93 23.44
2015 3:00:48 3:21:58 20.23 22.17

The average time of a postseason game dropped almost 16 minutes from ’14 to ’15. This was partly due to fewer extra-inning games being played this year, three as opposed to six last season, including an 18-inning Giants/Nationals marathon. The minutes per inning accounts for that, and it shows the pace quickening by over a minute and a quarter per frame.

Despite some relaxed rules, and despite much more frequent reviews, postseason play this year was clearly quicker, not just than last year but than 2013 as well. This parallels the regular season figures, where games were seven minutes shorter than 2014 and three and a half shorter than 2013, with similar minutes per inning numbers.

Even before I had the postseason numbers to tabulate, though, I could tell they were going to be good. The difference was palpable. I was still enduring the long nights I inveighed against last year, but not as bad and not as often. It felt like a reasonable price to pay for playoff baseball, rather than an imposition.

Commissioner Manfred accomplished something good for baseball this year. If I have even the slightest hesitation in lauding him for it, it is because I don’t want him relaxing and letting things slide. Keep up the good work, make moving the game along a habit, and he will have done major league baseball a lasting service.

Now if he could only do something about the Viagra and Cialis commercials every other inning in the playoffs, sending parents lunging for the remote to avoid awkward questions from their kids. And I thought games lasting until midnight would drive children away from postseason baseball.

Though it is pleasant to reflect that we seldom had to worry this year about a game lasting longer than four hours.

References and Resources

FanGraphs for Win Expectancy tracks for this year’s playoffs
Baseball-Reference for otehr 2015 game data and historic WE numbers
Retrosheet for replay information

A writer for The Hardball Times, Shane has been writing about baseball and science fiction since 1997. His stories have been translated into French, Russian and Japanese, and he was nominated for the 2002 Hugo Award.
Newest Most Voted
Inline Feedbacks
View all comments
Rays 7th Infielder
6 years ago

On September 23rd, the Rays played the Red Sox, not the Orioles.

6 years ago

Looking at games based on win probability change favors high-scoring games too much. Couldn’t you do something with leverage indices instead and give pitching match-ups more credit?

Shane Tourtellotte
6 years ago
Reply to  David

David, both Baseball-Reference and FanGraphs list Average Leverage Index (aLI) for games. This does bump up pitchers’ duels, but not as much as one might think. Don Larsen’s perfect game, which was a pretty tight 2-0, comes out to just 0.85 aLI for the Dodgers (1 being average) and 0.69 for the Yankees. Still a substantially below-average game. Look over the aLI numbers: if you like them, absolutely use them.

As I have mentioned in the past, WPS is better at measuring excitement than tension. High-scoring games lean this way (as long as they stay reasonably close), though run environment has an impact. One run in a high-run environment registers less with WE and WPS than it would in a lower-run setting. Low-run games can score well, too, if a good number of zeroes on the board represent threats to score that the pitchers snuff out. Dancing between the raindrops is fun; everybody going down 1-2-3 less so.

6 years ago

Ok, thanks for the reply. Another thought is that who (a star player or not) comes up in key situations can multiply the excitement. Also, the type of play counts, not just change in win probability (a strikeout > routine fly out, etc.) I don’t mean to knock your method, I’m just saying there should be allowance for something on top of the raw numbers. (Kind of an issue I have with SABRmetrics in total, but I don’t want to get off on a rant here.)

I had noticed how many classic games we have been privileged to witness in just the past 20 years or so, and I think your research here seems to back that up. Keep up the good work!

6 years ago

When will the 2016 hardball times annual be released?

Paul G.
6 years ago

TOOTBLAN has a onomatopoeia feel to it, kinda like the sound of a fan getting excited and then dropping his beer in disgust. I’ll have to start dropping it in conversations and wait for reactions.

As for male enhancement pills, just be glad its not Rafael Palmeiro. Or for that matter, be very, very, very glad that it was never Jose Canseco.

6 years ago

I enjoy TOOTBLAN’s a great deal and likewise enjoyed this article. TOOTBLAN awareness is vital for the discerning baseball fan.

6 years ago

Wasn’t it Hosmer’s mad dash home?

6 years ago

Shane – how did you get the 2015 data to compile your top WPS games? I wanted to analyze this data with some different methods/metrics but I obviously don’t want to have to copy/paste all of the game log info for all of 2015. I have emailed Fangraphs a couple times in the last two weeks but have not heard back from them. Thanks in advance.

Shane Tourtellotte
6 years ago

Anthony, for compiling the top regular-season games I went to Baseball-Reference rather than FanGraphs. I ran a Play Index search for games where one team (or both) had a very high Leverage Index, to gather an initial pool somewhere around 175 games. (As I said to David above, it definitely has its uses.) From there, I tallied up the game WPA+ and WPA- figures for both teams, given in the box scores on each game page: these are the positive and negative plate appearance WPAs. Top three plays come from the top five plays that the game page lists; final play is obvious.

This does not give exactly the same total that FanGraphs would, but aside from some rounding it isn’t any less valid than FG’s numbers are. B-R is, game for game, quicker, though it still requires a high threshold of tolerance for grinding through searches and lots of numbers. Fortunately I have a pretty high tolerance.

6 years ago

Shane – So my theory that insiders such as yourself can get Lahman-esque like data dumps from places such FG or BR is now proven false. Too bad, because FG has already done the hard work – to me making a .CSV file of all the Play Logs should be simple enough for the FG folks. Oh well, thanks for your insight.