Monday, August 29, 2011

Statistics and Stories

Questions: Do statistics take the wonder out of our sports? Do statistics drain the humanity out of them? Do statistics pull our eyes away from the fields and diamonds and courts and toward the ledger book? Do statistics make us less appreciative of the most important things in life, qualities that we intuitively understand are important but are not easy to quantify such as leadership and guts and the ability to handle pressure and the willingness to be a good teammate? Do statistics make us turn away from the myth and joy and mystery that make sports fun in the first place?

Most of all: Do statistics ruin good stories?


* * *

You may have caught the recent iDiscourse between my friends Chris Jones and Jonah Keri. They discussed the timing of Barry Zito's collapse as a pitcher. Jones argued that Zito's collapse after he signed the big contract with San Francisco was largely caused by the big contract itself and the pressures that came with it. Keri argued that the numbers clearly show that Zito was well on his way to collapse before the contract came along and that what followed was pretty much just the natural course of events. Jones counter-argued that he has never slept with a virgin. Keri returned with the charge that Anchor Steam is a great beer. Jones double-countered that there are signs of divinity in the song "Montana" by Youth Lagoon.

You know, I failed philosophy back in school so I got lost somewhere in there.

More than anything, though, I think they argued about what so many have argued lately -- the human record versus the human heart. Chris, I know, is not a devout non-believer in the power of numbers. And Keri, I know, does not live in his mother's basement. Still, lines were drawn, pistols chosen, and it seems to me they were not really arguing about Zito -- after all they are probably both at least somewhat right about him. Zito pretty clearly was on his descent before the absurd deal was written up; it is obvious in his numbers the last three years in Oakland. Baseball observers pretty unanimously knew it was a New Coke deal the day it was signed. And still, it is at least as obvious that Zito was messed up by the expectations that come with seven-year, $126 million deals* -- as people around him will tell you, the guy has become a basket case; he has lost his confidence as well as his fastball.

*This is the worst number in baseball, by the way: 126 million. If you are a general manager or hope to become one someday, avoid this number. Run from it. In baseball history, three men have received seven year, $126 million deals.

1. Barry Zito: 2007-2013
2. Vernon Wells: 2008-2014
3. Jayson Werth: 2011-2017

ZIto's contract is a disaster. That's obvious. He is 43-61 with a 4.52 ERA over the length of the contract, which is bad enough. Worse he was left off the postseason roster when the Giants won the World Series, and he's now such a wreck that rumors emerge daily of ways that the Giants can simply get rid of him -- shoot him into space, leave him on a porch, or, craziest of all, trade him straight up for Carlos Zambrano, one of those rare trades that would hurt both teams.

Wells contract is so bad that it was considered a modern miracle -- something in the part-the-Red-Sea variety of miracle -- when the Blue Jays were able to deal him to the Angels. Wells, unlike Zito, has had a couple of good years under this contract; those were the even years. And Jose Bautista gives him a lot of credit for his own turnaround, which has been worth a whole lot to Toronto. But in the odd years, he was among the worst everyday players in baseball. This odd year is the worst yet -- he's hitting .215/.249/.395, though those numbers are up over the last couple of weeks

The Jayson Werth deal might have been the most puzzling of all. The $126 million deal was obviously cursed already. He was turning 32 years old. The Nationals did not seem close enough to contention for this move to help them all that much even if Werth did maintain his 2010 form. He has not so far -- he's hitting .230/.328/.389 at this writing -- and because of his age and various other factors there seems little likelihood of him being an $18 million player any year over the next six. But, I guess you never know.

Either way, you might want to offer $125 million or $127 million or something like that in the future.


Getting back to the point: I don't want to make the Jones-Keri argument too stark. But I think there is a stark argument about statistics, especially in baseball, that is especially interesting to me as a writer. The prosecuting side of the argument goes like this: Statistics, advanced metrics specifically, make the game less fun. They turn a quintessentially human activity, played by men of flesh and blood, into a cascade of numbers like in The Matrix. They mock the unmeasurable contributions of gutsy men who get their jerseys dirty and lead by example and have a knack of making the big play just when the team needs it most. They muddy up the story lines that have spoken to baseball fans for a century -- the hitter who comes through in the clutch, the pitcher who will somehow win, the fielder who doesn't make mistakes and, like DiMaggio, always throws to the right base. Well, you've heard it all before.

But here, I don't want to argue AGAINST those points. I think that's one of the things that bothers me about such arguments -- you have traditionalists raging against the numbers, and the numbers people raging against tradition. I want to tell you one reason why I love baseball numbers. I love them because I believe advanced numbers can help us tell better stories.

Listen to this: In 1997, Mariano Rivera became a closer. He is almost certainly the greatest closer since the concept was invented -- I don't think anyone is particularly close. His story has been written again and again and again. His calm. His faith. His cutter. His postseason performance. His calm again. Back to his cutter. Maybe I should mention his faith. Rivera himself doesn't add much to the narrative; he is in interviews both agreeable and unrevealing … his most famous quote is probably the one he said during a kids show: "I get the ball, I throw the ball, and I take a shower."

Most athletes are like this. They either cannot or will not let us inside. David Foster Wallace wrote the best words about this in a marvelous review he wrote about Tracy Austin's biography:

"It may well be that we spectators, who are not divinely gifted as athletes, are the only ones able to truly see, articulate and animate the experience of the gift we are denied. And that those who receive and act out the gift of athletic genius must, perforce, be blind and dumb about it -- and not because blindness and dumbness are the price of the gift, but because they are its essence."

I might take it even another step further. Maybe athletes are not entirely blind or dumb to what's happening inside -- they just know that it's not very interesting. I remember once quarterback Jeff Blake completed a long pass in overtime to set up a game-winning field goal and break an eight-game losing streak, and a TV guy rushed over to him as soon as the game ended and shouted something like: "Great pass. Big win. Wow!" And then he put the microphone in front of Blake, who seemed thoroughly unsure how to respond to the passionate non-question. The PLAY was exciting to watch. The RESULT of the play was thrilling for a team and a city that needed a win. But the moment itself was undoubtedly a collage of mind-numbing practice and technical details and the body reacting as it was trained to react.

So it goes. I suspect that what Mariano Rivera FEELS in the moment is pretty limiting. It has been written, and even written well, but not often well. And I'm not sure how true those stories can be anyway. How much do you think Mariano Rivera can express what is going through his mind? I'm a writer and I don't think I can express what was going through my mind when, say, I wrote the last paragraph.

But how Rivera makes US feel -- as Yankees fans, as opposing fans, as former players, as current players, as managers -- yes, now, that can be something great. We can express those emotions, I think, because watching Rivera pitch takes us closer to the surface, brings us into that heightened sense of awareness. That's what's fun about it.

And one way to tell how Rivera makes us feel is by the numbers. For instance, here's one: Since 1997, the New York Yankees have won 97.2% of the games they led going into the ninth inning. The record, if you want it to the moment, is 1,256-36. Obviously, Rivera did not finish all those games. But he finished 859 games over those years. He is largely responsible for that amazing record. Over the years, the Yankees have led going into the ninth inning 85 or so times a year. With Rivera closing, they have lost those leads once or twice a year.

Now, if you just look at that number as a number, it might not tell a story. But think about it this way: Every time a New York baseball team loses a ninth-inning lead, America's largest city convulses. The back pages spread the news in bold letters. The columnists and talk radio hosts talk doomsday. Everything is magnified 20x in New York, as we saw with Hurricane Irene. We also saw it in the recent exchange between manager Joe Girardi and pitcher A.J. Burnett. Girardi pulled a thoroughly battered Burnett, and Burnett unleashed a profane rant that was watched and discussed on television. The postgame interview with Girardi was vaudeville comedy as he did a Jerry Lewis quality double-take when asked about Burnett's little tirade. Pitchers have cussed out managers who pulled them since the beginning of time, and A.J. Burnett has been something approaching unpitchable for two seasons now, but in New York this exchange was basically turned into the meeting at Yalta. That sort of thing happens to every team in sports, but it's louder in New York.

And so, the point is that Rivera, being Rivera, has saved the New York Yankees a lot of headaches and a lot of angry back page headlines the morning after.

Here's the beautiful thing about interesting statistics. They lead to other statistics … and other stories. For instance: What does a 97.2% success rate with ninth-inning leads even mean? It is, over Rivera's time, the best percentage in baseball. So that tells you one thing. But then, with a little more digging, I find that between 1951 and 1964 -- the last time the Yankees thoroughly dominated baseball -- the Yankees won 97.3% of the games they led going into the ninth inning. They did that without a closer. The save was not invented yet, but figuring out the saves retroactively the Yankees actually had FIFTY FIVE different pitchers over those 13 years with at least one save, from Luis Arroyo to Ryne Duren, from the young Al Downing to the ancient Sal Maglie, from (Whitey) Ford to (Virgil) Trucks.

And still, those Yankees won those ninth inning games at a slightly higher percentage than the later Yankees would with the greatest closer who ever lived. How many stories are there to be told about baseball and the closer and our power to believe that progress must make things better?

* * *

So, speaking of statistics and stories: Brilliant Reader John had this idea. Why not go back and look at the highest WPA games for hitters over the last 50 years.

WPA -- Win Probability Added -- is one of the most interesting statistics out there, and to be honest I do not see why it has not become more popular among mainstream baseball fans. Maybe it needs a better presentation, a better name, a public relations person because WPA, it seems to me, speaks so clearly to what so many baseball fans love about baseball: The winning plays.

The idea is this: WPA assigns a winning percentage to every single play. Let's say that the Indians are playing at the Royals. When the game starts each team has exactly a 50 percent chance of winning. That's pretty easy to get. Well, from that point on, each event has some impact on that winning chance. For instance, if Cleveland's Duane Kuiper leads off with a single, the Indians winning chances goes up ever so slightly. If he makes an out, the Indians winning chances goes down ever so slightly. Those fluctuations are WPA.

The Kuiper examples are tiny fluctuations. A single in the first inning would only improve the Indians chances of winning by the tiniest percentage. But there can be enormous fluctuations too. For instance, let's say there are two outs in the the bottom of the ninth, and the Indians lead by three runs … they will almost certainly win. But then the Royals load the bases, and Frank White hits a grand slam to win the game. Well, now you have an enormous fluctuation -- from an almost certain loss to a certain win. The final WPA would be credited to Frank White (and a negative WPA to the pitcher who allowed the home run, probably Sid Monge).

It's a fun statistic … it's not necessarily predictive of what will happen later, but it's a great stat for telling you who has had a huge impact on impact on the game.

Here, thanks to the marvelous Baseball Reference, are the 10 biggest WPA games over the last 50. These, by their nature, are 10 of the most thrilling games, and performances, of the last half century. I knew almost nothing about them, and there is no way I could have known about them without the magic of WPA and Baseball Reference:

10. Dwight Evans (June 23, 1990): This was a June game between the Red Sox and Orioles, and Baltimore led 2-1 in the eighth when Dewey homered to tie the game. Baltimore took the lead in the top of the 10th, and Evans hit a two-run walk off homer in the bottom of the 10th to win it. Not bad.

9. Hank Aaron (Sept. 10, 1971): You will notice that these games have a similar ring to them. This was Atlanta vs. San Francisco, and the Giants led 3-1 in the eighth. Aaron drove him one run with a double, and scored the tying run. In the 10th, the Giants led again by a run, and Aaron doubled and scored the game tying run. In the 11th, with the Braves down by a run, Aaron hit a walk-off three-run homer.

8. Jim Hickman (May 28, 1970): Hickman's Cubs trailed Pittsburgh by two in the seventh when he hit a two run homer to tie the game. In the ninth, with the Cubs trailing again, he hit another home run, this one a three-run walk-off home run.

7. Carlos May (Sept. 3, 1973): White Sox-Rangers. May walked in the second and came around to score and tie the game 1-1. With the White Sox trailing by three in the seventh, he hit a three-run home run to tie the game. In the ninth, with the White Sox down one, he hit an RBI single to tie it once more. And in the 12th, with the score tied, his double scored that former Marine, Tony Muser, and the White Sox won it. I'm actually a bit surprised, considering everything, that this game wasn't a bit higher on the list.

6. Willie Mays (May 26, 1962): What an awesome game. Giants-Mets -- so we're talking here about the famed '62 Mets. In the third, Mays tripled and scored to tie the game 3-3. In the eighth, the Mets led by a run and Mays, undoubtedly perturbed by this revolting development, homered off Jay Hook to tie it again. And in the 10th, the Mets AGAIN led by a run, and Mays came up with the tying run on second and undoubtedly thought "All right, enough of this." He homered to win it.

5. Mel Hall (June 27, 1984): I do remember this game … I wonder how many of these games you remember. The Indians were terrible, the Twins were mediocre. In the bottom of the eighth, the Twins led by two runs, and Mel Hall doubled in Brett Butler and Pat Tabler to tie the game. It stayed tied until the 10th, when the Twins scored. Hall won it with two outs in the bottom of the 10th with a three-run home run.

4. Bobby Grich (July 15, 1979): Angels-Yankees. The Yankees took a 4-0 lead in the second inning. Grich singled in a run in the third. In the seventh, he hit a two-run double to make the score 4-3. And in the ninth, with two outs, he hit a two-run homer to win the game 5-4. I love games like that, where one guy drives in all the runs.

3. Brian Daubach (August 21, 2000): And … I love that the top three WPA games of the last 50 years were not achieved by all-time great players. With two outs in the bottom of the ninth, and Daubach's Red Sox down two runs to the Angels, he homered to tie the game. And in the bottom of the 11th, with the Red Sox down one and two outs, Daubach hit a two-run single to win it.

2. Jim Pagliaroni (Sept. 21, 1965): One more time, the Mets got beat by the Herculean efforts of one man. The Mets led 4-3 in the seventh, when Pagliaroni singled in a run to tie it. And then, with two outs in the bottom of the ninth, Pagliaroni hit a two-run walk-off homer to win it.

1. Art Shamsky (August 12, 1966): I feel like I should have heard a lot about this game … but honestly, I don't remember ever hearing anything about it. This was Pittsburgh vs. Cincinnati, and it was a wild game before Shamsky even entered. Three all-time greats -- Pete Rose, Roberto Clemente and Willie Stargell -- homered in they game. A couple more Hall of Famers (Tony Perez and Bill Mazeroski) played. The Reds led by two, the Pirates led by two, the Pirates led by one, the Reds led by one, back and forth. Someone could have written a book about this game.

Shamsky was put into the game in the eighth inning as part of a double-switch. At the time, the Pirates led by a run. Shamsky came up in the bottom of the eighth and whacked a two-run homer off Al McBean to give the Reds a lead. The Pirates tied it in the top of the ninth, and the game stumbled into extra innings.

In the 10th, Willie Stargell homered to give the Pirates a lead. In the bottom of the inning, yes, Art Shamsky came up again. And he homered to tied it up.

Then in the 11th, Bob Bailey -- who had already hit two home runs -- hit a two-run double to give the Pirates the lead. It's fun to note here that one of the runners who scored: Jim Pagliaroni.

And, yep, you probably know what happened in the bottom of the 11th. With two outs and a man on base, Art Shamsky came up to the plate. And, yep, he hit another home run to tie up the game.

There are two great bits from this game. One, the bat that Shamsky used is somewhere in the Hall of Fame. It's not there because this was the highest WPA game -- it's there because two days later Shamsky was sent in as a pinch-hitter with the Reds down a run. And, he hit a two run homer, meaning he had homered in four straight plate appearances.

The second bit? Well, Art Shamsky had the greatest WPA ever for a single game. His performance is the very peak of what man can do to win a baseball game. He homered to give his team the lead in the eighth, homered to tie it in the 10th, homered to tie it again in the 11th, there is not much more a baseball player can do. And so what happened? Art Shamsky's Reds lost to the Pirates that day.




41 comments:

  1. Fabulous article as usual, Joe. I, too, have long wondered why WPA hasn't caught on, since it speaks to some of those intangibles (like "clutchiness", to use a Colbert-ism) so frequently harped upon. I have been thinking about it more with pitchers, however. What do you think about changing the criteria for assigning a pitcher the win from whatever the rubrics are now to the highest WPA for the winning team of any pitcher? Losses could operate the same way.

    ReplyDelete
  2. WPA is my favorite stat. So much so that if I had an MVP ballot I would simply list the ten league leaders in WPA, case closed. I don't think it's ever really going to catch on. The concept is just too slippery for most people. For me it's the stat that comes closest to boiling down the entire game to one number.

    ReplyDelete
  3. Cool post. It's interesting that the guy with the greatest single game WPA didn't even play the entire game.

    I think this has implications for the everyday player vs. pitcher for MVP discussion. Al Shamsky played only 6 of 13 innings and batted 3 times in that game. Yet as you wrote, "His performance is the very peak of what man can do to win a baseball game."

    So why is it so hard to believe that a starting pitcher who averages 6 or 7 innings per start could be the best in his league at contributing wins to his team?

    ReplyDelete
  4. Hey Joe, did Jim Pagliaroni play for the Pirates in 1965, for his high WPA game?

    ReplyDelete
  5. It probably doesn't get remembered because his name sounds like a Polish nickname for a fraud.

    ReplyDelete
  6. I probably just don't understand WPA enough, but perhaps a couple BRs could enlighten me:

    Are all outs treated the same? (Does a pitcher get the same WPA "credit" for striking a guy out as he does for giving up a screaming line drive hit right at a fielder?)

    Has anyone addressed the role opportunity plays in WPA? (Batter X who comes up with RISP 50 more times than batter Y?)

    ReplyDelete
  7. Marco, WPA is more of a fun descriptive stat than a measure of true talent. All outs are credited to the pitcher, defenders get no credit. While you can add up WPA over a season, it's best used when describing individual games, so differences in opportunity, such as RISP chances, are not accounted for in any way.

    Fangraphs uses a WPA graph in their game stories which is always fun to look at.

    ReplyDelete
  8. WPA would give us Jose Bautista and Joey Votto as MVP, and both by a WIDE, WIDE margin, BTW. Oh, and Verlander trails Weaver and they both trail Bautista by a bunch with Weaver 3rd (behind Cabrera) and Verlander 5th (behind Hamilton).

    It should be noted that while it's a great idea, WPA does not include defensive plays. Once fangraphs figures out a way to include defense in the equation it will be a great stat.

    ReplyDelete
  9. @Mark Daniel - Sure, a position player had a huge impact on that one particular game. But that game was a huge outlier - he hit 3 home runs in 3 plate appearances. Obviously a hitter who hits three big flies, or goes 5 for 5, or whatever, can have a bigger impact on one game than the starting pitcher.

    However, in any given game, most hitters get 4-5 plate appearances, more than half of which end in outs. Even a really good hitter gets on base maybe twice per game. A really good starting pitcher, on the other hand, can give the team 7+ innings of quality pitching. So although an individual hitter may be able to outperform the starting pitcher in a given game, the starting pitcher probably outperforms most (if not all) of the position players in any given team win.

    Now, obviously that's offset by the fact that starters pitch only one game in five, while position players play in just about every game. Still, there's an argument to be made that a starting pitcher contributes more to his team's success than an everyday player.

    ReplyDelete
  10. Thanks ry.live & golfman - that helps

    So really, WPA doesn't really tell us much other than who had big moments in the game - which now that I think of it could just as easily be gleaned from the recap.

    (I think we've uncovered the reason it hasn't caught on)

    ReplyDelete
  11. How do you find WPA for past games? I would think Bobby Murcer’s game of Aug. 6, 1979 should rank quite high. The Yankees were trailing the Orioles 4-0 going into the bottom of the seventh, and as I recall, the Orioles hadn’t lost a game all season when they were ahead in the seventh. In the bottom of the inning Murcer hit a three-run homer off Dennis Martinez, his first since returning to the Yankees after his exile in San Francisco and Chicago. Then in the bottom of the ninth he hit a two-run walk-off single off Tippy Martinez, giving him all five RBIs in their 5-4 win. It was made more poignant because earlier that day he had delivered the eulogy at the funeral of his close friend, Thurman Munson.

    ReplyDelete
  12. It seems worth mentioning another story that masks the story. Historically, people who cover baseball are sports journalists who lack training in, say, mathematics. Thus, unable to speak in terns of mathematical models, they must revert to speaking in terms of metaphors, parables, analogies, etc.

    So it's not so much that certain stories have spoken to fans for generations; they've just been the only stories the people who cover the story are equipped to tell.

    If baseball reporting started out the province of scientists, probably it would be the "story crowd" that lives in the proverbial mothers basement.

    ReplyDelete
  13. "Aaron drove him one run" should be "drove in".

    "Three all-time greats -- Pete Rose, Roberto Clemente and Willie Stargell -- homered in they game" should be "the game".

    I love reading what you write, Joe.

    It seems to me that, as with WAR, WPA undervalues pitchers. Koufax's perfect game, for example, came with at most a 1-0 lead. Every at bat had substantial risk of losing the game. How could he have improved on it? Maybe by driving in the game's only run. But it doesn't make the top ten, nor does any other pitching effort.

    ReplyDelete
  14. So what exactly do you expect readers to take from an article like this? Without narrative and stories, reading about sports wouldn't nearly as entertaining or enjoyable. For someone like yourself who has made a living and raised a family on the profitability of sports writing, its interesting to see you turn your back on it for some math equations and spreadsheets. I really enjoy a lot of content on this website but I don't understand the need to bash people over the head with stats and numbers. Its a tired act.

    ReplyDelete
  15. Art Shamsky and Harvey Haddix. I love great performances that end in a loss.

    ReplyDelete
  16. I think it will just take a long time. But the stories have always had numbers in them, wins and hitting streaks and RBI. Those statistics have added to the storytelling for generations. The storytellers grew up with those numbers. And so they could be seamlessly built into the story. Right now, the numbers are new, and so to be a great storyteller with WAR and WPA and OPS+, you need to be both great with numbers and great with words. Not many swim in the deep end of each pool. But twenty years from now, the storytellers will be as comfortable with the new numbers as the old timers have been with ERA and RBI.

    ReplyDelete
  17. You underestimate the value of Rivera. Comparing him to a different era doesn't make sense. You also forget he has done throughout the steroid era. Just an example of what Rivera has meant to the Yankees:

    Name WPA -WPA +WPA IP WPA/IP
    Roger Clemens 76.15 -308.88 385.03 4916.667 1.549%
    Randy Johnson 53.85 -273.57 327.42 4135.333 1.302%
    Greg Maddux 53.69 -320.07 373.76 5008.333 1.072%
    Pedro Martinez 51.84 -164.57 216.41 2827.333 1.834%
    Mariano Rivera 51.09 -115.25 166.33 1150.000 4.443%
    Mike Mussina 39.76 -225.21 264.98 3562.667 1.116%
    John Smoltz 37.82 -247.27 285.09 3473.000 1.089%
    Curt Schilling 36.54 -208.14 244.68 3261.000 1.121%
    Roy Halladay 34.65 -141.92 176.57 2297.333 1.508%
    Trevor Hoffman 32.98 -123.03 156.01 1089.333 3.028%
    Rich Gossage 31.4 -172.15 203.55 1679.333 1.870%
    Dennis Eckersley 30.97 -255.17 286.15 3285.667 0.943%
    Kevin Brown 30.55 -220.1 250.65 3256.000 0.938%
    Tom Glavine 30.21 -303.25 333.45 4413.333 0.685%
    Johan Santana 26.62 -110.15 136.78 1908.667 1.395%
    Tim Hudson 25.79 -147.41 173.2 2288.333 1.127%
    Bret Saberhagen 25.62 -164.43 190.05 2562.667 1.000%
    Roy Oswalt 24.36 -125.81 150.17 2015.000 1.209%
    Lee Smith 23.97 -155.97 179.95 1289.333 1.859%
    CC Sabathia 23.66 -141.27 164.93 2127.000 1.112%

    ReplyDelete
  18. Thanks for reminding me that the Angels traded for Wells. Three years and $63 million left on his contract, and the best we can hope for is a 120 OPS+ in two of those seasons? *groan*

    As for statistics versus stories, I think that any of us who played at least sandlot ball as kids understand the magic of baseball on a level that the numbers can never tell. I can appreciate baseball by the numbers just as much as I can appreciate baseball from the heart.

    ReplyDelete
  19. Great post Joe,
    I'm wondering what the WPA would be for this scenario:

    Team X trails Team Y 7-5 in the eighth.

    Player Z hits 2-run shot in bottom of the eighth to tie it up.

    Game tied 7-7 in bottom of the ninth, Player Z hits two out grand slam to win the game.

    I know it wouldn't be anything like Shamsky or any of the top ten for that matter, but it happened recently at a game I went to and the grand slam incidentally won my wife a brand new F-150.

    You can read my post about the experience here:
    http://thegrandslam.wikispaces.com/

    Craziest thing to ever happen to me. Also, does anybody know how to find out the probability of a player hitting a grand slam in the bottom of the ninth when the inning began tied? I've been looking all over and can't find a thing. Thanks

    ReplyDelete
  20. WPA is indeed great. It has some drawbacks (the baseline is average not replacement, it doesn't deal with fielding or baserunning)...but it does what it does beautifully. It's for writers, it's not for GMs.

    Best 5 WPA players since about 1950: Bonds, Mays, Aaron, Mantle, McCovey. Best Hall of Fame reject: Dick Allen.

    Worst 5 WPA players since about 1950: Alfredo Griffin, Royce Clayton, Larry Bowa, Neifi Perez, Ed Brinkman.

    Best WPA postseason game ever: with one swing of the bat, it's Kirk Gibson in Game 1 of the 1988 World Series.

    ReplyDelete
  21. My single biggest wish in regards to those covering sports is not that they lean toward numbers or narrative. What I would like is that they don't use narratives when the numbers can actually show they're false. For example, does veteran leadership make a difference? I don't know, but I suspect we could find out. I am pretty sure I'm right that a number of WS rings have been won by rookie managers. I don't have the answer on how much experience matters, but it could actually be determined. If it then turns out not to matter much, that's the time the broadcasters and writers need to give up on it as a false narrative.

    ReplyDelete
  22. "It seems to me that, as with WAR, WPA undervalues pitchers."

    You can't search for both pitching and hitting results with one query on B-R.

    ReplyDelete
  23. The greatest start by WPA in that time period was Gaylord Perry's 16 (sixteen!)-inning shutout in 1967, in which he ended up with a 1.601 WPA.

    It starts to get stretched out the more innings pitched for back in the insane era, obviously.

    ReplyDelete
  24. Sandy's perfect game was .674, and only his third-best start of the year. His second-best was 10 shutout innings; that's why that's higher. But, his best was nine shutout innings. Reasons why that's higher than the perfect game: the Dodgers scored earlier in the perfect game, meaning that he had more innings in the lead, so there was less that he could add.

    ReplyDelete
  25. Dinky, as you suggest, WPA is not perfect for analyzing pitchers. The perfect game you mentioned was a brilliant performance giving the Dodgers a 1-0 victory. A week or so later, Koufax pitched another complete game shutout leading to a 1-0 victory, however he allowed 4 hits and 1 walk. He also struck out 6 batters, compared to 14 in the perfect game.

    But the 4-hit shoutout had a higher WPA (0.810) compared to the perfect game (0.674). It was because Koufax allowed runners to reach base in that game, which made the subsequent outs more important from a WPA perspective. In fact, Koufax's two biggest WPA events in that game were the last 2 outs of the game, both coming with a man on 2nd.
    In the perfect game, all outs came with nobody on base, thus they weren't as "important".

    ReplyDelete
  26. **OFFTOPIC - INDIANS TRIVIA**

    Came across this chart about franchise leaders in games played (noting Jeter's recent climb past the Mick)- http://espn.go.com/blog/sweetspot - and noticed that Cleveland all-time leader in games played has an extremely low number, Terry Turner with 1617. Turner was a Dead Ball Era shortstop.

    The only franchises lagging that number are expansion franchises: Arizona, Florida, Tampa Bay & Toronto. Ten other expansion franchises have a leader with more games, some by a lot: COlorado, Houston, KC, Angels, Milwaukee, Mets, San Diego, Seattle, Texas/Washington (Michael Young so this one is still running), Washington/Montreal.

    As a Clevelnad guy, thought you might find this interesting. . . .

    ReplyDelete
  27. You can have both, but of course i prefer the stats because they cut through the BS the media likes to run with. Like "Manny couldn't play anymore" and "Juan Pierre filled in as good as him during the suspension" the stats would tell you the real story. When he supposedly "quit" on Boston he had the 3rd highest OPS+ on the team at about 130. Stats tell us the TRUTH! I leave the narratives to Vin Scully and leave it at that.

    Did Joe just say he's a Yankees fan? I thought he liked the Royals?

    I dry my legs and feet first since i'm about to step out of the shower...

    ReplyDelete
  28. I'm a little disappointed the top 10 WPA games are so obscure. It would have been cooler if someone like Rick Camp, July 4-5, 1985 had made the list. (If only he had homered again in the 19th...) Anyway, I think Pete Ridges is onto something when he mentions Kirk Gibson as the #1 postseason WPA game. The postseason top 10 WPA games would bring back more memories for more people, and be more meaningful I think.

    ReplyDelete
  29. If a pitcher pitches a complete scoreless inning, his cumulative WPA for that inning will be the same whether he pitches 1-2-3 or loads the bases. The bases loaded inning will show a large "spike" or "dip" on the chart while the 1-2-3 inning will be a gentle slope. But the total effect is the same.

    The exception is for baserunning plays like SB, CS, WP and PB. Those plays are not assessed to the pitcher's WPA at all (even wild pitches!).

    The reason Koufax's non-perfect game has a higher WPA then the perfecto is for three reasons:

    1. He pitched on the road in the non-perfect game, at home in the perfect one. Pitching in the bottom half of the inning with a 1 run lead is more leveraged than pitching in the top half with a 1 run lead.

    2. Park factor. That year, protecting a 1-run lead in high-scoring Busch Stadium was statistically tougher than protecting a 1-run lead at Dodger Stadium.

    3. Maury Wills stole a base for the Cards in the non-perfect game. Koufax gets none of the blame for the SB, then gets a little "extra" credit for stranding him.

    The 10-inning shutout falls in the middle because of a combination of all factors above. Koufax's WPA was hurt here because he was always pitching with the score tied. Somewhat counter-intuitively, with all other things being equal, WPA is higher when protecting a 1-run lead than when pitching in a tie game.

    ReplyDelete
  30. Slight clarification: My last sentence above is true assuming the pitcher does not give up a run!

    ReplyDelete
  31. That Shamsky game comes up every week or so in the comments section of the blog at baseball-reference.com. I have read about it over and over again on that site, but I'm still amazed by it.

    ReplyDelete
  32. And also, I meant Lou Brock, not Maury Wills. Damn.

    ReplyDelete
  33. rokirovka -

    Got me curious:
    - GIbson 1988 .870 WPA - tough to rack up huge WPA numbers when you pinch-hit once. I suppose your team would have to be behind by more runs to score higher than this in a 1 PA pinch-hit game.
    - Morris, Game 7 1991 - .845 WPA
    - Francisco Cabrera, 1992 - .737 WPA - the difference between his and Gibson's is tying run on 3rd and winning run on 2nd
    - Puckett Game 6 1991 - .593 WPA, game-winner yes, but tie game at the time.

    Well, occurred to me that you can set b-ref's play index for post-season. Gibson's HR is #1 for hitters at .870. Steve Garvey is #2 for Game 4, 1984 NLCS - go-ahead single in the 3rd, tying single in the 5th, go-ahead single in the 7th and a game-winning HR in the 9th. Cabrera's game is #9 BTW

    As expected, pitchers score higher with the possibility of extra inning starts. There are 4 pitching performances ahead of Gibson's .870 led by. . . . .Babe Ruth's .975 WPA, CG, 14 IP, 1 R performance in Game 2 of the 1916 WS. Top 9-inning game is Art Nehf's shutout in a 1-0 win in Game 8 (yes, 8) of the 1921 WS, .912 WPA. (Nehf is also #7 on the list for his 9 inning shutout in a 1-0 win in Game 3 of the 1923 WS.) Morris is #6. The highest game since WWII is Dave McNally's 11 inning 1-0 win in Game 2 of the 1969 ALCS, .915 WPA.

    ReplyDelete
  34. Great piece, as normal. WPA sounds very interesting, and gets to the heart of things in terms of contributions to wins, but to someone with little knowledge, what does the percentage mean and what is an average nubmer?

    Also, I know Rivera is an excellent pitcher, but the post season adulation always baffles me a little bit. He lost two of the most crucial post season games over the past 10 yers - WS game 7 against the D-Backs and the playoff game against Boston that allowed them to get up off the floor and come back from the 3 to 0 deficit. Don't those games diminish Rivera's otherwise excellent post season pitching?

    ReplyDelete
  35. The percentage itself is "how probable it is that a team in this specific scenario will win".

    There's not really such a thing as an an average number; it fluctuates heavily. And that's just seasonal totals. Single game, holy crap, there's no such thing as an average.

    Last year's seasonal champs were Miggy Cabrera (7.5 wins added), and Joey Votto (a shade under 7 wins).

    The Doodoo Bat winners were Ryan Theriot (-3.5), and Jason Kendall (-2.8). They were both bad at the whole hitting thing, yeah, but obviously offensive positions are what are rewarded, since it's devoid of positional context, and that's why it's basically the same trap of "GIVE THE MVP TO GEORGE BELL!" as using RBIs to figure out the best players, even if it's a more telling stat; they say that you need to be "strong up the middle" for a reason.

    ReplyDelete
  36. Also: because this is partially about closers, am I the only one who would vote for a closer to win the Cy Young even though I wouldn't dream of it for MVP? It has NOTHING to do with pitchers being less deserving because they have their own award; in fact, I'm certain that there's been a more valuable starter who wasn't as exceptionally good as another starter because the former threw more innings, but the innings themselves weren't as good. Wilbur Wood/Vida Blue in 1971 might be an example; Wilbur was almost certainly the most valuable player in the AL, but he probably wasn't as good as Vida, because the latter racked up so many Ks.

    But on closers vs. starters. Eckersley's 1990, with the 0.61 ERA/610 ERA+ (like, holy crap in a hat). 3.2 WAR in 72 IP. 18.25 strikeouts for every walk. Like, that year had one of Clemens's best seasons ever. But if he had gotten hurt in June, and the best year was Chuck Finley's, a really great year but nothing historic, could you make the argument that Eck was the BEST pitcher? I think that you could. That's what the Cy Young is about; not value.

    ReplyDelete
  37. What happens when you combine a pitcher's pitching and his hitting? July 3, 1966: Tony Cloninger hits 2 grand slams, drives in a 9th run with a single. Complete game, 3 ER, 5 Ks, Braves win 17-3

    ReplyDelete
  38. WPA is a fun stat, really good for watching the ebb and flow of a game, but it is too based on situations to be used as an individual stat, in a similar manner to RBI's.

    Team 1's #3 hitter hits 4 solo home runs, accounting for all the runs in the game. Their closer blows up culminating in allowing a grand slam home run to put his team down 5-4. With 2 outs in the ninth, the ninth place hitter, who is 0 for 3 with 3 K's, hits a game winning two run homer.

    The #3 hitter had 4 consecutive home runs. His team certainly would not have won without him. He is a distant 3rd in WPA for the game, but easily had the game of his life. He is "penalized" because he did not come up with a chance to win the game in the ninth.

    A good player on a team that plays a lot of close games will have more WPA than a similar player on a dominant team, simply because there are more high leverage situations. If your team is already beating up on someone when you come up in the latter part of the game, you don't get the credit given someone else whose team always seems to be in a close game.

    Likewise a great pitcher on a dominant offensive team would not have as much WPA as a similar pitcher on a team that has to scratch out runs and plays a lot of low scoring games.

    This is just off the top of my head, and to me, it's already enough to show that it is too much about context, and luck to be taken seriously as a way to determine the best player.

    ReplyDelete
  39. Rusty Traveler, thanks for clarifying WPA. I didn't realize that WPA was park adjusted for one, and for some reason I thought a pitcher allowing someone to reach 2nd was more than made up for by the pitcher stranding that runner.

    It's good to get a nice explanation like yours, since most of what I've read about WPA has been more like math derivations than general explanations.

    ReplyDelete
  40. I think WPA is fun, but like KHAZAD, i don't like how situational it can be.
    If a player leads off the game with a solo homer, then his team wins 1-0, his WPA is probably pretty insignificant.
    If the same player hits a solo home run in the bottom of the 9th of a 0-0 game, his WPA is going to be a lot higher.
    But each player contributed the exact same amount to the win. The second player gets all the reward because of when he did it.

    ReplyDelete
  41. The name WPA needs to be changed. It sounds just like every other acronym in modern statistics, and that name alone will cause most in the media to roll their eyes and ignore it. Joe, you could start referring to WPA as Impact Percentage, or Clutch Factor, and that would help it become more generally popular. Most sportswriters and announcers would love to talk about something called Clutch Factor. At least then, they would have to base the narrative more on what actually happened.

    Seems like it'd be interesting to compute WPA not just on the impact a play has on one game, but the impact it has on the season as a whole-- how much did one play impact the team's position in the standings in terms of the pennant race? The numbers would be 162th of the size they are in regular WPA, but you could get a perspective of how important one given play is on a season-- a routine out would be virtually zero, but then something like Bucky Dent's HR or Yaz's hits in late 1967 would be relatively immense.

    ReplyDelete