The WPS Index (part two)

In yesterday’s installment, I introduced a statistical method for measuring the excitement of baseball games. The WPS (Win Percentage Sum) Index uses the change in Win Expectancy on each play of a game to total up a number representing how exciting that game was. This works well as a base, but there are a couple tweaks I can make, drawing from the same WE data.

The first tweak is the Best Plays modification. An exciting game has dramatic plays that stick in the memory after the field has emptied. These don’t line up perfectly with the highest Win Percentage Added (WPA) plays, though the really pivotal ones almost always do. For the Modified WPS, I will add the three highest WPA scores to the overall rating.

Baseball-Reference helps us in this matter, listing the five highest WPA plays of each game for which it has play-by-play data. (The game graphs at FanGraphs mark high-WPA plays, but they don’t always get the top set.) I nearly took all five for the index, but I did not want this element taking on excessive importance, at least compared to my next modification. Neither did I want to be seen as merely cribbing from B-R: I’m only mostly cribbing from them.

Second is the Last Play modification. Looking at memorable games in baseball history, a huge number of them are walk-off victories. Taking MLB Network’s recent list of the “20 Greatest Games” from 1961-2010 as an example (and you’ll soon see why I’m doing that), 14 of the 20 games are walk-offs, comprising all but one of the home wins on the list. (The 15th has the visitors’ tying run cut down at the plate for the final out.)

Fans love the adrenaline rush of walk-offs, and remember games that give them that thrill. A system that rates exciting games is well served by an adjustment that values such moments. I will therefore add the WPA value of the final play of each game to its modified WPS score. This will often be very low or even zero, but sometimes it will be quite high (c.f. Mr. Gibson). It also means that particularly climactic final plays will be counted three times in the system: once in the base count, once as a Best Play, once as the Last Play. I believe this fits fans’ perceptions, their thorough enjoyment of walk-off drama.

And that’s the Modified WPS Index: the total of all positive WPA numbers for each play in the game, plus the top three WPA scores, plus the score of the last play. You now have everything you need to play the WPS home game. But I’m still going to play it here.

Tangent

Max Marchi, as you’ll recall from yesterday, put a great deal of work into his own game-excitement system. At one point (midway down this page), he took a long list of variables that can help measure the excitement of a game, and calculated how well they correlated with three vital factors in that measurement: equilibrium, rally, and late-game importance. (1 meaning perfect correlation; 0 meaning no connection at all.) Some factors correlated well with one factor, or two, or all three, and Marchi chose those factors that produced the least overlap, covering as wide a base as possible with a limited number of factors.

It was fairly deep into my own WPS writing that I noticed one of the factors he had discarded: mean Win Expectancy swing. This is effectively my own base WPS system, only with the result divided by the number of plays in each game to produce an average value. And of all the variables he examined, this is the only one that correlates at least at a 0.5 level with all three factors that make up an exciting game. If you had to choose one variable from that list to measure the excitement of a game, this was the one.

This isn’t to say that Marchi made a mistake discarding it. He was producing a precise system, accepting a high level of complexity in order to get that much closer to whatever the objective truth might be. I’m going for something simpler, more transparent—and it’s gratifying to see that I hit upon something that produces a reasonably accurate answer with a fraction of the effort. Max Marchi wasn’t trying to show me I was on the right track, but he did anyway, and I’m glad.

Test drive: rating the classics

The time has come to put the system through its paces. The aforementioned “20 Greatest Games” list from MLB Network will serve as the test track. I’ll show you how the WPS Index rates each game, how those games show its strengths and weaknesses, and how WPS highlights facts about the games.

First, though, I will show you a baseline game. From May of the current season, this game scores about as poorly as any you will ever find, showing the floor of the WPS system. (The game graph is from FanGraphs, but the WPA numbers are from Baseball-Reference.)


Source: FanGraphs

Game       1   2   3   4   5   6   7   8   9    F
Blue Jays  0   0   0   0   0   1   0   2   0    3
Rangers    6   1   1   0   0   2   4   0   x   14

WPS        1   2   3   4   5   6   7   8   9 Ttl.
Blue Jays  5   4   2   0   2   0   0   0   0
Rangers   41   2   1   0   0   0   0   0   0   57
Best Plays: 23  Last Play: 0  Grand Total: 80

That’s rock bottom, or close. You’d have to work hard to find a game with a lower WPS Index, and I’m sure you can use your time better. Using FanGraphs numbers would rate the game a little higher, at 93.9. Their tenth-of-a-point precision in their WPA numbers lets tiny numbers add up, rather than get rounded down to zero as Baseball-Reference does.

That said, I’m going to use Baseball-Reference’s numbers for my ratings, for two reasons. One, as mentioned yesterday, is that FanGraphs doesn’t have WPA numbers for nearly as many games as B-R, hamstringing historical comparisons. Another is that inning-by-inning totals scan much easier with whole numbers than decimals, letting you see patterns at a glance. Ease of use is worth the loss of a little precision.

I list the 20 games in the inverse order that MLB Network rated them, saving their greatest for last. Keep in mind that their subjective ratings take the importance of the game into account. Only one mid-season game makes the list (and barely), with a pair of Game 163s squeezing in as well. The WPS Index just considers the games, not the championship implications.

I give the line score for each game, the WPS Index numbers below, its rating among the 20, and my observations afterward. And away we go.

#20. 5/17/1979, Phillies @ Cubs

Game       1   2   3   4   5   6   7   8   9  10    F
Phillies   7   0   8   2   4   0   1   0   0   1   23
Cubs       6   0   0   3   7   3   0   3   0   0   22

WPS        1   2   3   4   5   6   7   8   9  10 Ttl.
Phillies  64   9  43   0   0   2  18   3  38  53
Cubs      52  33   0   1   9  20  22  99  15  21  502
Best Plays: 92  Last Play: 5  Grand Total: 599 (12th)

Notes: The 116 score in the very first inning, before leverage has had a chance to rise, is huge. All those runs did set the tone for the game. As I said yesterday, you get some low ratings when Philly has a big lead, ignoring the strong wind blowing out of Wrigley that made a comeback so much likelier.

#19. 10/4/2003, Giants @ Marlins (NLDS Game 4)

Game       1   2   3   4   5   6   7   8   9    F
Giants     0   1   0   0   0   4   0   0   1    6
Marlins    0   1   2   2   0   0   0   2   x    7

WPS        1   2   3   4   5   6   7   8   9 Ttl.
Giants     9  15   6   6   4  62   9  16  61
Marlins   13  24  43  24   3  50   9  50   x  404
Best Plays: 70  Last Play: 14  Grand Total: 488 (18th)

Notes: WPS’s dislike of outs on hits injures this game’s score. Ivan Rodriguez‘s climactic tag-out of J.T. Snow at home to win the game rates just as a game-ending out, no better than a pop to second. A better accounting might have pushed this game from “really good” to “great,” or at least “really, really good.”

#18. 10/12/1980, Phillies @ Astros (NLCS Game 5)

Game       1   2   3   4   5   6   7   8   9  10    F
Phillies   0   2   0   0   0   0   0   5   0   1    8
Astros     1   0   0   0   0   1   3   2   0   0    7

WPS        1   2   3   4   5   6   7   8   9  10 Ttl.
Phillies   4  34  11   9   4   4  20 125  32  66
Astros    26  10  20  16  15  44  54  92  12  17  615
Best Plays: 113  Last Play: 4  Grand Total: 732 (3rd)

Notes: Observe the high value of prompt comebacks that I mentioned in Part One. Houston goes ahead by three in the seventh, but Philadelphia surges ahead in the eighth, only to get tied immediately after. The WPS goes sky-high, over 200 for the eighth inning, past 270 for the three half-innings. A textbook example of how a good game can become a tremendous one.

#17. 10/17/2004, Yankees @ Red Sox (ALCS Game 4)

Game       1   2   3   4   5   6   7   8   9  10  11  12    F
Yankees    0   0   2   0   0   2   0   0   0   0   0   0    4
Red Sox    0   0   0   0   3   0   0   0   1   0   0   2    6

WPS        1   2   3   4   5   6   7   8   9  10  11  12 Ttl.
Yankees    5  17  26   7  10  62  14  14  11  22  46  32
Red Sox   11   8   6   6  64  18  12  34 101  15  30  34  605
Best Plays: 75  Last Play: 27  Grand Total: 707 (6th)

Notes: Not too much room for complaint on the excitement level after the first four innings, but I can always find something. A zero-out rally to win takes the direct path to 100 percent, losing the value of the “sawtooth,” and the walk-off homer came with the winning run already on base, cutting into its value. Boston really needed two outs to get the best out of that frame. (Yankees fans would argue for three.)

#16. 10/6/2009, Tigers @ Twins (AL Central Tiebreaker)

Game       1   2   3   4   5   6   7   8   9  10  11  12    F
Tigers     0   0   3   0   0   0   0   1   0   1   0   0    5
Twins      0   0   1   0   0   1   2   0   0   1   0   1    6

WPS        1   2   3   4   5   6   7   8   9  10  11  12 Ttl.
Tigers     5  27  37  10   6   4  20  59  76  62  15  70
Twins      9   5  29   6   6  44  61  12  29 129  15  38  774
Best Plays: 116  Last Play: 29  Grand Total: 919 (1st)

Notes: Okay, my memory wasn’t tricking me. This really was an awesome game. Dull patches early on are more than outweighed by the ratcheting tension from the home sixth onward. A 105-point scoreless ninth is very good; a 191-point 10th as the Twins match a Tigers run is superb. (This meshes very well with Chris Jaffe’s system, and that’s no bad thing.) Detroit had two half-innings in the 70s without a run coming across, showing how a pitcher squeezing out of a late jam can, even in this offense-biased system, produce big excitement. This is the highest-rated game on the list, and it’s not close.

#15. 10/8/1995, Yankees @ Mariners (ALDS Game 5)

Game       1   2   3   4   5   6   7   8   9  10  11    F
Yankees    0   0   0   2   0   2   0   0   0   0   1    5
Mariners   0   0   1   1   0   0   0   2   0   0   2    6

WPS        1   2   3   4   5   6   7   8   9  10  11 Ttl.
Yankees    5   5  13  29  13  53   6   6  55  15  47
Mariners   5  33  17  29   7  26   9  79  32  40  77  601
Best Plays: 91  Last Play: 32  Grand Total: 724 (5th)

Notes: Deja vu all over again: a zero-out extra-inning rally against the Yankees ends up leaving WPS equity on the table. It’s simply not that exciting a climax if the hero doesn’t have to struggle at the end before triumphing. Wait, did all that reading in English classes just help me in real life? Stop the presses!

#14. 10/23/1993, Phillies @ Blue Jays (World Series Game 6)

Game       1   2   3   4   5   6   7   8   9    F
Phillies   0   0   0   1   0   0   5   0   0    6
Blue Jays  3   0   0   1   1   0   0   0   3    8

WPS        1   2   3   4   5   6   7   8   9 Ttl.
Phillies   7   4   6  16  22   4  81  13   4
Blue Jays 29   5   3   9   6   4  11  48 106  378
Best Plays: 102  Last Play: 66  Grand Total: 546 (14th)

Notes: Joe Carter gratuitously exploits the Last Play bonus (that has to be what he had in mind, right?). That helps a little, but as the Blue Jays held a three or four-run lead the majority of the game, WPS numbers were depressed much of the way. An exciting game, eventually, but as championship-winning walk-offs go it’s a touch on the dull side.

#13. 10/26/1997, Indians @ Marlins (World Series Game 7)

Game       1   2   3   4   5   6   7   8   9  10  11    F
Indians    0   0   2   0   0   0   0   0   0   0   0    2
Marlins    0   0   0   0   0   0   1   0   1   0   1    3

WPS        1   2   3   4   5   6   7   8   9  10  11 Ttl.
Indians    5   5  43   8  14   9   3   5  18  25  30
Marlins   16   5  16   6  10  14  35  14  67  39  87  474
Best Plays: 78  Last Play:  34  Grand Total: 586 (13th)

Notes: The ninth inning, and beyond, are so important to WPS. If the Marlins don’t tally, this game ends barely above average. But they do, and the last 2 1/2 innings rack up more points than the 8 1/2 that came before. Of course, without that comeback and the extra innings, we remember this game much less vividly. The WPS Index isn’t saying anything we didn’t already believe.

#12. 10/31/2001, D-Backs @ Yankees (World Series Game 4)

Game       1   2   3   4   5   6   7   8   9  10    F
D-Backs    0   0   0   1   0   0   0   2   0   0    0
Yankees    0   0   1   0   0   0   0   0   2   1    3

WPS        1   2   3   4   5   6   7   8   9  10 Ttl.
D-Backs   27   5  18  28  23   9  29  49   3  15
Yankees    5   5  16   6   7  25  36  10  88  57  461
Best Plays: 124  Last Play: 46  Grand Total: 631 (9th)

Notes: The added factors boost this game: fourth-highest Best Plays number, fourth-highest Last Play, and third-best combined figure. And the walk-off homer came with two outs, for maximum WPS effect. So the Yankees were smart to let Derek Jeter handle that.

#11. 10/2/1978, Yankees @ Red Sox (AL East Playoff)

Game       1   2   3   4   5   6   7   8   9    F
Yankees    0   0   0   0   0   0   4   1   0    5
Red Sox    0   1   0   0   0   1   0   2   0    4

WPS        1   2   3   4   5   6   7   8   9 Ttl.
Yankees   15   5  11  16  18   9  85   9   6
Red Sox    5  14  17   5   9  16  22  67  67  396
Best Plays: 82  Last Play: 21  Grand Total: 499 (17th)

Notes: Not as strong a score as the game’s reputation would suggest. There isn’t a full-inning score out of the 20s until the seventh and Bucky Dent. Things pick up steam then, but a little late.

#10. 10/15/1988, A’s @ Dodgers (World Series Game 1)

Game       1   2   3   4   5   6   7   8   9    F
A's        0   4   0   0   0   0   0   0   0    4
Dodgers    2   0   0   0   0   1   0   0   2    5

WPS        1   2   3   4   5   6   7   8   9 Ttl.
A's       22  68  17  16   3   3  10   4   4
Dodgers   28   7   7   6  19  43  20  13 110  400
Best Plays: 142  Last Play: 87  Grand Total: 629 (10th)

Notes: My modifications to the base WPS Index get their sternest test. Did Kirk Gibson‘s at-bat really provide more than 40 percent (87*3/629) of the excitement of this entire game?
{thinks about game}
{hears Vin Scully’s call in his head}
{gets chills all over again}
Yes, it probably did.

#9. 11/4/2001, Yankees @ D-Backs (World Series Game 7)

Game       1   2   3   4   5   6   7   8   9    F
Yankees    0   0   0   0   0   0   1   1   0    2
D-Backs    0   0   0   0   0   1   0   0   2    3

WPS        1   2   3   4   5   6   7   8   9 Ttl.
Yankees    5   5   6   6   7   9  49  35   5
D-Backs    9  18  16  13  10  16  17  23 118  367
Best Plays: 92  Last Play: 16  Grand Total: 475 (19th)

Notes: Akin to the Bucky Dent game above, this contest took until the seventh to have an exciting inning. Worse, as Curt Schilling didn’t have to pitch with a man on base until the seventh, the game had a case of Don Larsen Syndrome (without actually being a no-hitter: Paul O’Neill was thrown out stretching a double). Combined, those factors produced an even weaker game than the 1978 playoff. Still exciting; just not competitive in this fast company.

#8. 10/12/1986, Red Sox @ Angels (World Series Game 5)

Game       1   2   3   4   5   6   7   8   9  10  11    F
Red Sox    0   2   0   0   0   0   0   0   4   0   1    7
Angels     0   0   1   0   0   2   2   0   1   0   0    6

WPS        1   2   3   4   5   6   7   8   9  10  11 Ttl.
Red Sox    5  32   4   5  14  14  14  13  96  66  42
Angels     7  18  25  10  15  46  18   0 106  20  21  579
Best Plays: 142  Last Play: 5  Grand Total: 726 (4th)

Notes: Not too much missing from this game. It’s tight most of the way, and while California’s late 5-2 lead deflated the eighth, a titanic ninth left that dip long forgotten. The Angels going down in order in the 11th was an anticlimax; a couple baserunners would have beefed up the index nicely. But I’m not sure the 1986 postseason could have contained that much more excitement.

#7. 10/14/2003, Marlins @ Cubs (NLCS Game 6)

Game       1   2   3   4   5   6   7   8   9    F
Marlins    0   0   0   0   0   0   0   8   0    8
Cubs       1   0   0   0   0   1   1   0   0    3

WPS        1   2   3   4   5   6   7   8   9 Ttl.
Marlins   16  10  20  10  15   9   7  95   0
Cubs      24   9   5  13   5  28  12   1   1  280
Best Plays: 66  Last Play: 0  Grand Total: 346 (20th)

Notes: MLB Network dropped the ball, much like a certain Cubs fan I could mention. This game is judged “great” not because of the comeback, but because of the notorious event that preceded (and debatably triggered) the comeback. Without that foul ball—which doesn’t rate at all on the WPS Index—this is a moderately exciting game played at a crucial time. To be fair, my system and their list are measuring different things, but even by their standards, I wouldn’t call this game great. Historic, maybe. Not great.

Okay, grumble ended.

#6. 10/16/2003, Red Sox @ Yankees (ALCS Game 7)

Game       1   2   3   4   5   6   7   8   9  10  11    F
Red Sox    0   3   0   1   0   0   0   1   0   0   0    5
Yankees    0   0   0   0   1   0   1   3   0   0   1    6

WPS        1   2   3   4   5   6   7   8   9  10  11 Ttl.
Red Sox    9  35   3  18   7   3   3   9  27  29  15
Yankees   13   5   5   7  13   6  26  88  15  15  36  387
Best Plays: 83  Last Play: 36  Grand Total: 506 (15th)

Notes: A lead of at least three runs from the second to the seventh takes a lot of starch out of this game. And once again, a no-out walk-off gives us an ending less thrilling than it might have been. At least there was nobody on, which helps a little. So Aaron Boone got that much right, but he really should have listened to Jeter more closely.

#5. 10/15/1986, Mets @ Astros (NLCS Game 6)

Game       1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16    F
Mets       0   0   0   0   0   0   0   0   3   0   0   0   0   1   0   3    7
Astros     3   0   0   0   0   0   0   0   0   0   0   0   0   1   0   2    6

WPS        1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16 Ttl.
Mets       4   4  13   4   4   4   4   8  77  13  25  13  13  75  19  48
Astros    44   3   2   2   4   4   0   0  12  12  12  22  12  62  12  51  582
Best Plays: 102  Last Play: 16  Grand Total: 700 (7th)

Notes: This classic’s high score is due primarily to its great length. The second through eighth innings, with a three-run lead and almost no rallies, are a WPS desert, just the kind of game uncommitted viewers might give up on before the real excitement begins. If you tuned in at the top of the ninth, you saw a 596 game instead of a 700, and in half the time.

#4. 10/14/1992, Pirates @ Braves (NLCS Game 7)

Game       1   2   3   4   5   6   7   8   9    F
Pirates    1   0   0   0   0   1   0   0   0    2
Braves     0   0   0   0   0   0   0   0   3    3

WPS        1   2   3   4   5   6   7   8   9 Ttl.
Pirates   21   4   4   5  14  21  11   7   2
Braves     5   5  21   6   7  63  31   9 177  413
Best Plays: 129  Last Play: 74  Grand Total: 616 (11th)

Notes: This is a pretty middling game until the bottom of the ninth—but what a bottom of the ninth! A clear example of how one half-inning can make a game memorable, if it’s the final half-inning. And if it’s an elimination game deep in October.

#3. 10/25/1986, Red Sox @ Mets (World Series Game 6)

Game       1   2   3   4   5   6   7   8   9  10    F
Red Sox    1   1   0   0   0   0   1   0   0   2    5
Mets       0   0   0   0   2   0   0   1   0   3    6

WPS        1   2   3   4   5   6   7   8   9  10 Ttl.
Red Sox   29  21   4   8  11  14  28  16  25  49
Mets       5  10   6   6  41  34  10  42  49 106  514
Best Plays: 116  Last Play: 40  Grand Total: 670 (8th)

Notes: The intra-inning action raises the index here. There’s only a single 1-2-3 inning after the fourth, and those little threats do keep the pot bubbling even when nobody scores. I can’t add much to what everyone knows about the ending, except that a line drive lashed over first base to score Ray Knight would have gotten no more WPS points than Mookie Wilson‘s croquet shot through Bill Buckner‘s legs. Make of that what you will.

#2. 10/27/1991, Braves @ Twins (World Series Game 7)

Game       1   2   3   4   5   6   7   8   9  10    F
Braves     0   0   0   0   0   0   0   0   0   0    0
Twins      0   0   0   0   0   0   0   0   0   1    1

WPS        1   2   3   4   5   6   7   8   9  10 Ttl.
Braves     5  13  20  12  29   7   9  64  15  15
Twins      5  11  15  14  10  18   9  55  51  36  413
Best Plays: 75  Last Play: 16  Grand Total: 504 (16th)

Notes: I told you WPS wasn’t enamored of pitcher’s duels. It sure does like pitcher’s jams, though. Check out that eighth inning: 119 points without a run scoring. There aren’t too many 1-2-3 innings to suppress the numbers, but the system just isn’t going to like this game as much as the (non-Braves) fans did.

#1. 10/21/1975, Reds @ Red Sox (World Series Game 6)

Game       1   2   3   4   5   6   7   8   9  10  11  12    F
Reds       0   0   0   0   3   0   2   1   0   0   0   0    6
Red Sox    3   0   0   0   0   0   0   3   0   0   0   1    7

WPS        1   2   3   4   5   6   7   8   9  10  11  12 Ttl.
Reds       9   4   6  13  53  17  62   9  15  38  30  43
Red Sox    32  3  10  11  16  15   9  75  72  15  15  36  608
Best Plays: 112  Last Play: 36  Grand Total: 756 (2nd)

Notes: Second place on the WPS scale for the fabled Greatest Game Ever? What does this lack that Tigers-Twins 2009 had? Turns out, an extra-inning comeback to re-tie the game: factor that out, and the two are all but even. Once you count the greater consequence of a World Series Game Six versus a divisional tiebreaker, though, you can justify returning this game to the top of the heap.

At least, from 1961 to 2010. As you may recall from last October, there’s a new contender for the all-time title. And the WPS Index says …

10/27/2011, Rangers at Cardinals (World Series Game 6)

Game       1   2   3   4   5   6   7   8   9  10  11    F
Rangers    1   1   0   1   1   0   3   0   0   2   0    9
Cardinals  2   0   0   1   0   1   0   1   2   2   1   10

WPS        1   2   3   4   5   6   7   8   9  10  11 Ttl.
Rangers   38  37  11  31  39   7  47   1   4  54  24
Cardinals 32   5   5  23   7  70   4  38  95 103  38  713
Best Plays: 144  Last Play: 38  Grand Total: 895

Clearly, the WPS Index says “Wow!”

I’ve mentioned that WPS likes high-scoring games, but largely for what they make possible: frequent lead changes. This game is a glowing example. All eight runs in the first six innings change the answer to the question “Who’s winning?”. Texas grabs a bigger lead in the seventh, but St. Louis chips away, then ties it in the most WPS-friendly manner, with two outs in the bottom of the ninth. Then they pull another do-or-die multi-run comeback in the 10th. Then David Freese caps it with the walk-off longball.

Yes, he was supposed to do it with two outs for maximum effect, but people get impatient in these games. (A passing fact: his walk-off was only the fourth-highest WPA score in the contest. That’s okay. His triple in the ninth was #1.)

By the WPS Index, 2011′s Game Six leaves 1975′s Game Six 139 points behind. That’s about the same gap as the Fisk classic boasts over the 1992 Pirates-Braves Sid Bream game. The system doesn’t boast micrometer precision, but that’s pretty strong support for those who believe the 2011 version was more exciting.

(I had no particular dog in this hunt to bias the ratings. I respect baseball history, of which 1975′s Game Six is a huge part. But I was too young to see, much less fully appreciate, that game in its time. I got to see the 2011 version, beginning to end, as it happened, and I’ve taken to thinking of it as my Game Six. I could have gone either way on which one’s better, and I was expecting a closer finish.)

No, this doesn’t finish any arguments. The WPS Index isn’t the final word. It’s more like one judge at a boxing match, and hopefully not the kind we usually read about who seems to have been watching a different bout. But rather like the Rangers-Cardinals game, I’m thinking of it as my judge now, and even given its noted flaws, I rather like it.

The road ahead

What could be done with the WPS Index to further improve its usefulness? One modification is obvious: we could factor in the importance of the games. Tigers-Twins 2009 maintains a WPS edge over Rangers-Cardinals 2011, but the sheer import of a World Series Game Six above a divisional tiebreaker means that viewers’ subjective excitement is keyed to a higher pitch. A bonus or multiplier for championship importance would reflect that, and as this system is trying to match our personal reactions, that’s all to the good.

Something akin to Sky Andrecheck’s Championship Leverage Index could be added to the WPS to produce this modified measurement. I may do that myself, especially as October nears, but anybody who wants to mess around with it themselves is fully welcome. I might start feeling proprietary over the system later, but not right now. For today, I’ll be glad if people actually use the system.

Along those hopeful lines, it would not be impossible for FanGraphs or Baseball-Reference to start counting up and displaying the WPS Index themselves. The coding ought to be pretty simple, another potential advantage of a streamlined system. It might even give FanGraphs the impetus to get those older playoff games tracked and displayed in all their graphical, tabular glory. It may be presumptuous even to suggest this, but I’m allowed to dream big.

And of course, there are so many more games left to run through the WPS machine. Maybe there are some games that should be crowbarred into MLB’s top 20 list (like the 2007 Padres-Rockies tiebreaker, or the 2005 Braves-Astros NLDS game that went 18 innings). And we’ve got playoffs coming up in a couple months, games ready to be sifted and judged. But all those games can wait for another day.

Postscript one

Well, except for one.

Before coming up with his excitement-rating system, Chris Jaffe gave his own personal answer to the question of what the greatest game ever was. He chose the July 4-5, 1985 contest between the Mets and Braves, a 19-inning ironman triathlon of a game. It had no immediate post-season implications, which suited Jaffe fine: he was judging it in isolation, purely as one game.

Just the way the WPS Index does. So what does my new toy think of Jaffe’s choice? I thought you’d never ask. (Line scores are compressed because there wouldn’t be room on the page otherwise.)


Source: FanGraphs

Game       1   2   3   4   5   6   7   8   9 10 11 12 13 14 15 16 17 18 19    F
Mets       1   0   0   4   0   1   0   1   1  0  0  0  2  0  0  0  0  1  5   16
Braves     1   0   2   0   1   0   0   4   0  0  0  0  2  0  0  0  0  1  2   13

WPS        1   2   3   4   5   6   7   8   9 10 11 12 13 14 15 16 17 18 19 Ttl.
Mets      30   5  18  60  10  38   5   9  78 25 13 25 62 33 25 13 29 38 53
Braves    31   7  32  13  24  14   8 108  13 19 39 13 74 13 13 29 19 73 10 1121
Best Plays: 163  Last Play: 4  Grand Total: 1288

That outstrips any other game I’ve measured, by miles. There is an element of pure quantity involved—extra innings do pack on the points—but this was a 500-point game through nine, and the extras did more than just skate along. Other super-long games could post similar scores, but for the moment, the WPS Index likes Chris Jaffe’s choice best. I hope that’s compensation for treading on his turf.

Oh, and remember when I said I nearly called this system the Richter Scale? As much like a hyperactive seismograph as that game graph is, this contest is one more reason why I broomed that idea. If ballgames were earthquakes, this thing would have cracked the planet in two. And I didn’t want to give Michael Bay or Roland Emmerich ideas for their next movie.

Postscript two

If you’ve read all this way (my congratulations and sympathies), by now you have gotten some feel for how the WPS Index draws itself around the contours of a game, giving it shape. You might even think you could recognize a given game by its inning-by-inning WPS numbers. For extra credit, I give you this problem.

WPS (B-R)  1   2   3   4   5   6   7   8   9 Ttl.
Visitors   7   6  15  69   8  27   0   0   0
Home      24   5  26  16  14   0   0   0   0  217
Best Plays: 50  Last Play: 0  Grand Total: 267

Despite that middling score, this is a pretty well-known regular-season game from the last 10 years. Without consulting game graphs, or Win Percentage Added or Win Expectancy numbers (the line score and regular play-by-play are okay), can you identify the game? The winner gets nothing more than bragging rights, but geek-cred is better than no cred. Have fun!

References & Resources
FanGraphs and Baseball-Reference remain indispensable.

Print Friendly
« Previous: Currently HIstoric: More TTO fun
Next: The daily grind: 8-16 »

Comments

  1. Paul G. said...

    Hmm.  The last three innings are all 0s, which means the score must have been lopsided at that point.  Three innings must mean something… could it be the infamous Wes Littleton game, the one where he earned a 3-inning save in a blowout?

    One addition that could be added to the system for greater accuracy, but makes the number more difficult to compute, would be to break up the compound plays into their component parts.  For example, runner on first, single to right field, runner gets thrown out trying to go to third.  As a play that is more or less a WPA wash.  But if it was broken down into two chunks – Part I: single, runners on first and second; Part II: runner thrown out at third – then I suspect that has value, especially in a close game.

  2. Asa W. said...

    2004 ALCS Game 5 scores a 770. Probably due to the close score throughout and the fact the Yankees left EIGHTEEN men on base. They went 1-2-3 twice in fourteen innings.

    And of course, WPS fails to take into account the intangibles such as the historical narrative of Red Sox futility, the desperation of being down three games to one, and the fatigue of both bullpens after Games 3 and 4. In my opinion, this is the most exciting game in history.

  3. Shane Tourtellotte said...

    Paul G., is this ever your day.  You get name-dropped in Jason Linden’s weekly article, and you win the hypothetical kewpie doll.  It is indeed the Wes Littleton 30-3 game. (Yes, fellow readers, Paul is a friend and one-time co-author with me, but he had no insider knowledge on the pop quiz.  He nailed that one on the square.)

    Asa:  You may have lowballed your case.  I count up a base score of 780, modified to 914.  That’s Twins tiebreaker/David Freese territory.  And of course statistical systems can’t encompass the intangibles.  At some point, the subjective experience really is more important.

    You’ve thrown a great contender into the ring.  Who’s next?

  4. David said...

    Great article, great piece of work!  An additional intangible would be who made the big plays?  For example, the Yankees-Diamondbacks game 7 was a comeback against arguably the greatest reliever ever.  Or how about the Gibson-Eckersley match-up?  Certainly the key players in the spotlight for the biggest plays adds additional excitement.

  5. Shane Tourtellotte said...

    Yikes.  How did I ever put that down on the screen?  Sorry, my brain went to the Bahamas for a moment or something.

  6. Sam said...

    While it’s most notable for Art Shamsky’s performance(3 HR in 3 AB as pinch hitter), which was good for 1.5 WPA, Pirates/Reds on 8/12/66 was pretty exceptional anyway.

    http://www.baseball-reference.com/boxes/CIN/CIN196608120.shtml

    I just crudely added up the WPA+ and WPA- from the boxscore, then added best plays and last play, and I came up with 1161. A lot of that was because there were five plays with over 40 WPA.

  7. studes said...

    Shane, why give extra weight to certain plays?  Those plays already have extra weight.  End-of-game plays get extra weight if they’re high leverage. As do the top three plays.  I don’t understand why you added these tweaks.

  8. Shane Tourtellotte said...

    Studes:  I added them to try to match fans’ reactions to the games, especially for the final play.  Take two games with, say, a base 500 WPS index.  Fans will remember the game with the walk-off homer much more fondly than the one where the final rally fell short, or never happened at all. (I mean this for fans without rooting interests.  Yankees fans are thrilled whenever Mariano shuts the door, and won’t think much of Game 7 in 2001, whatever the numbers say.)  Lists of great games are stuffed full of walk-offs.  My tweak may cut across the analytical grain (a “great” game), but it meshes with fan psychology (an “exciting” game).

    The Best Plays tweak is similar, as it taps into the “remember when?” aspect of our memory of games.  It is less precise, and I did include it partly to get a little more separation in the numbers between middling, good, and great games.

  9. studes said...

    Thanks.  Seems to me the three plays thing is completely arbitrary.  If you want more separation between games, you could just multiply each change in WPS by two.

    The thing about the two games with identical scores of 500 is that the game with the big last play was, by definition, less exciting before it.  The big last play brought it up to the other game in the base score.  You can argue that it’s already been given its due weight by WPA.

    Just my two cents.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current day month ye@r *