31 Responses to “Post ’49 Batting Leaders Highest Game WPA Where PA = 0”

1. Andy Says:
   April 27th, 2011 at 10:46 am

   Whoa, this is a fascinating list. I'm guessing it's going to reveal a lot of the problems with WPA.

   Check out the first game above. Tavarez came in as a pinch-runner at first base with no other base-runners and no outs in the 9th, with the Marlins down by 1.

   During the next plate appearance there was a wild pitch and he took 2 bases, earning 0.19 WPA.

   Two outs and a walk later, he was balked in from 3rd, for 0.31 WPA.

   I'm not saying he doesn't deserve at least some of that WPA--for all we know, he caused the pitcher to balk by dancing off 3rd base. And maybe the wild pitch didn't go far from the catcher and he showed incredible hustle getting to 3rd base. But this is a case where it's really difficult to know how much he really deserves it.

   I am guessing this list will be littered with stuff like that.

2. BSK Says:
   April 27th, 2011 at 11:22 am

   Does this include playoffs? I'd have to guess that Dave Roberts should be on there for Game 4 of the 2004 ALCS.

3. BSK Says:
   April 27th, 2011 at 11:25 am

   Roberts had a WPA of "only" .107. I figured it would have been higher, but I guess he doesn't get credited for scoring from 2B on the single, just the SB.

4. deal Says:
   April 27th, 2011 at 11:28 am

   Love the #20 Jerry Hairston Quadfecta of PR/SH/HBP/BALK to get around the bases.

   fortunatly the Phils came back to win that one.

5. John Autin Says:
   April 27th, 2011 at 12:05 pm

   Without even checking the play-by-play, I'll bet that most of these games involve the featured player scoring the tying or go-ahead run in the 8th or later, on a play that he had virtually nothing to do with.

   Normally, I would segue right into my "here's what's wrong with WPA" screed. Instead, as an exercise in opening my mind, I will note that virtually every measure of player performance has "noise," i.e., credit and blame assigned when they are not truly deserved.

6. Joseph Says:
   April 27th, 2011 at 1:34 pm

   How about highest WPA without scoring a run and no plate appearance?

7. John Autin Says:
   April 27th, 2011 at 3:21 pm

   The postseason high in this category is 0.375 by George Foster in the 1972 NLCS, the deciding 5th game. Foster (surprise!) scored the winning run on a wild pitch, after entering as a pinch-hitter.

   No other postseason game of this sort had a WPA of 0.300 or higher.

8. John Autin Says:
   April 27th, 2011 at 3:47 pm

   @6, Joseph -- Highest WPA without scoring a run & no plate appearance:

   0.254 by Terry Shumpert on 05/02/1997. Shumpert pinch ran at 1B in the bottom of the 9th with no out and his team down by a run. On a passed ball, Shumpert moved all the way to 3B, but he was stranded there.

   http://www.baseball-reference.com/boxes/SDN/SDN199705020.shtml

9. Doug Says:
   April 27th, 2011 at 3:47 pm

   A bit baffling to me how high WPAs can result if your team loses the game. Nine of the 31 games on the list are for performances by losing players (incl. the two highest WPAs on the list).

   Number 2 (Kenny Lofton) was another PR at 1st base, stole second, advanced to 3rd on a slow roller to the 1st baseman, and scored on a wild pitch with two outs. This was in the bottom of the 9th, and Lofton's run tied the game. He was removed from the game for a defensive replacement in the top of the 10th.

   Lofton's run was critical at the time, but not in the context of the ultimate game result. Nor in the season context - this was the last game of the year between two also-rans.

   Should WPA make adjustments based on whether your team wins or not? I know a player can have a great game, do everything possible to help his team win, and have nothing whatever to do with why his team lost. Yet, FWIW, somehow it doesn't make sense to me to award a high WPA if, despite a player's heroics, his team still lost.

10. Johnny Twisto Says:
    April 27th, 2011 at 4:04 pm

    Should any player be credited with anything positive if his team loses?

    WPA is quite simple. The change in win-expectancy before and after an event. One can quibble with how that change is credited to particular players. But after every event, one team has improved its chances of ultimately winning the game. The batting team is trying to move players around the bases, and the pitching team is trying to get outs. If they succeed in their respective tasks, they improve their chances of winning. If you want to look back after the game and say it didn't matter whether they lost 1-0 or 100-0, that is your right. At the time the game was being played, two teams were trying their best to win, and WPA attempts to credit those players who had the most impact.

    If you want to only measure WPA in wins, you can. It's not WPA anymore, though. It's "WPA in WIns." I did something similar last season, tracking which Yankee attained the most WPA in each game they won, and who had the lowest WPA in each game they lost -- sort of a Hero/Goat of the Game tally. I don't know if the final results were that meaningful, but it was mildly interesting.

    As with any statistic ever made, the question is, "What are you trying to measure?" Answer that, and then you can decide which statistic(s) to use, or if you need something new.

11. Tmckelv Says:
    April 27th, 2011 at 4:06 pm

    Since we are talking about "unearned" WPA...here is MY biggest problem with the stat...Defense not accounted for at all. Here is a perfect example:

    For anyone that did not see the end of the Yankees/White Sox game last night:

    The White Sox brought in Santos in the bottom of the 9th inning, up by 1 run with runnes on 1st and 2nd.

    Santos gave up 2 screaming line drives to A-Rod and Cano, either of which would have won the game for the Yanks except for Brent Lillibridge's two lunging/diving catches that saved the game for the Sox.

    Santos' WPA = .342 (not to mention the SAVE!!! - oops, I mentioned it)
    Lillibridge's WPA = .015 (because of a 1-1 batting performance)

12. Johnny Twisto Says:
    April 27th, 2011 at 4:25 pm

    Tmckelv, agreed.

    WPA is perfect** for tracking the flow of a game -- which plays were most important in turning the tide, etc. It is imperfect in the accounting of how to credit the individuals. That is of course most notable in separating fielding from pitching. No statistic has figured out the best way to do this yet. WPA doesn't even try -- all credit goes to the pitcher. I can't even imagine how to go about dividing credit for those plays. Obviously it would require someone to observe every single play. Do you give the Lillibridge 75% of the credit? 98%? Who knows?

    What might work is tracking how often a play is made on a particular type of hit in a particular place on the field. This data is already roughly collected, I believe, and is used to create the advanced defensive stats like UZR, DRS, Total Zone. Once Field F/X comes along, that data will be improved. So if that last ball caught by Lillibridge would be caught 5% of the time, Lillibridge gets 95% of the WPA and the pitcher 5%. Something like that. It could never be divided perfectly but it could definitely be improved.

    **Perfect, to the best of our estimations. We can never "know" if the average team would win 75% of the time from a specific point in a game, because we can't play that game over between average teams a million times. But we can estimate pretty close, based on history.

13. Tmckelv Says:
    April 27th, 2011 at 4:58 pm

    @12 JT,

    I am even more drastic in the split of Defensive player and pitcher. In the Santos/Lillibridge example I talked about above, I believe Santos actually hurt his team so him getting 5% might be too much. I think if a pitcher gives up a hit ball that would normally be a hit without a great defensive play, then perhaps he should get a negative WPA for the play and the defender get even greater positive WPA.

    I know that is ridiculous and goes even beyond the realm of statistical analysis because you would need to further break the granularity down to the point of single pitch placement (for example, the pitcher makes a great 0-2 pitch but the hitter makes an off-balanced check-swing and bloops a bases-clearing double)...this is where the performances and results are diametrically opposed, and requires too much "official scorer-type" judgement beyond what can reasonably be accomplished statistically.

    Like you said, WPA is great for the flow of the game and the end results of each play, but there are some limitations as discussed in this and many other threads.

14. John Autin Says:
    April 27th, 2011 at 5:49 pm

    Another huge flaw in WPA: It fails to credit managers when they get their team fired up by deliberately getting thrown out of the game.

    Also, it fails to give me credit (blame) when I directly influence game events by making in-game observations, a la "the Phillies have gone 14 straight games without scoring 5+ runs."

    🙂

15. bluejaysstatsgeek Says:
    April 27th, 2011 at 6:34 pm

    Last year, Jose Molina caught 4 TBRays - one in each of the first four innings - trying to steal second, and the credit goes to the pitcher, Brandon Morrow. Sure, some of the credit should go to Morrow for delivering the pitch quickly and possibly some to the middle infielder that applied the tag.

    When hit f/x or play f/x is available, it seems to me that at the start of every pitch, we should be able to determine the positioning of every defensive player, and when the ball is hit, determine the probability that the ball will be caught, a single, double, triple or HR.

    We even have data on the effectiveness of pitches and I can see WPA being applied at the pitch level.

16. BSK Says:
    April 27th, 2011 at 7:49 pm

    TMCKELV and JT-

    But maybe we have to give Santos credit for allowing those balls to be hit hard enough to reach Lillibridge! If he was a better pitcher, they likely don't make as good a contact and maybe those balls drop in.

    It's sort of the way a guy can perform TOO well on a given play and ultimately fail. A screamer that is caught is worse than a dribbler that you reach on. Now, in the big picture, you'd prefer the screamer because the odds of it resulting in a hit are higher than the dribbler. But there are always times where it pays to be bad.

    Similar to a guy with 7-2 off suit beating a guy with pocket A's. If you are on the losing hand, you'll bemoan yourself for not having had worse cards to start with. Which is silly. Of course, if the best starting hand (i.e., best player) won every time, how much fun would that be?

17. Neil L. Says:
    April 27th, 2011 at 7:59 pm

    Wow, this blog brought out the sabermetric heavyweights. Awesome analysis. I'll just sit on the sidelines and learn.

    @6, John A., your observation about the presence of random noise in any particular baseball event is a cogent one.

    @10 Doug, there shouldn't be any issue with assigning a positive WPA for a player even if his team loses. After all, as I understand it, this statistic attempts to isolate the effect of an individual plate appearance from the performance of the team and the outcome of the game.

    @11 Tmckelv, your post highlights (again) the fact that offense and defense on a particular play are two separate worlds. The pitcher is not the only defender.

    Discussion on a very high level in here. Thank you, Steve, for such an interesting blog.

18. dukeofflatbush Says:
    April 27th, 2011 at 9:30 pm

    @ 10 Johnny,

    I agree with you.
    I hate when people argue a stats' merits, or its usefulness.
    A stat simply states what it is. Exactly like WPA, which is nothing more than what it implies: A change in outcome expectancy.
    I can only agree with some detractors who argue that attributing something like a balk to an individual's WPA, when clearly he simply was lucky. Maybe things like balks, should be attributed to a team. rather than to an individual, but even including 'luck stats' - over a long enough timeline, all things equal/level out.

19. dukeofflatbush Says:
    April 27th, 2011 at 9:40 pm

    @10,
    Back to my original point, I once argued on this site the merits of the triple crown. My antagonist was so adamant in saying the TRIPLE CROWN proved nothing. But I kept saying, "yes it does" - it shows who led in RBI/HR/AVG- a neat stat, but nothing more than it claims.

20. Neil L. Says:
    April 27th, 2011 at 9:47 pm

    @18
    But, Duke, everybody who develops a new uber-stat always has a vested interest in declaring it is the ultimate metric. Not saying that WPA is like that, but reputable baseball analysts throughout time have argued a stat's merit.

    How can you not have an emotional opinion about a particular statistic?

21. dukeofflatbush Says:
    April 27th, 2011 at 10:34 pm

    @ 20 Neil,

    Good Point.
    It does get annoying when someone overreaches a stats' point or value by trying to simplify or quantify the entire game into a single number.
    Every stat is helpful in its own context, it depends on what an individual is trying to prove with that #.
    It does get overwhelming when each month a new, perfectly useful stat, is thrust onto the baseball vernacular, claiming to have this or that advantage over previously held beliefs in measurements, but a yard is three feet, no matter how you slice it.

22. John Autin Says:
    April 27th, 2011 at 11:40 pm

    How about the flip side of Steve's theme?
    Since 1950, lowest WPA by a pitcher without officially facing a batter:

    -0.373, Morrie Steevens, 9/19/1964./
    http://www.baseball-reference.com/blog/archives/10864

    Phillies fans of a certain age might remember this game -- though the memory may be buried under the trauma of the 10-game losing streak that started 2 days later.

    The Phils had eliminated the defending champion Dodgers in the series opener, and held a 6.5-game lead in the NL race with 15 games to go. But LA would take a little revenge over the next 2 games. They won the next game with 4 late runs. The next game would be a marathon that was settled on a sprint.

    Bottom of the 16th, 2 out, potential winning run on 3rd in the person of Willie Davis. Gene Mauch brings in the young lefty Morrie Steevens -- just up from the minors and making his first MLB appearance in 2 years -- to face the LHB Ron Fairly. And Willie Davis steals home! It was the 2nd (and last) steal of home among his 398 career steals. Ironically, the bottom of the 14th had ended with Fairly at bat and Willie Davis out trying to steal home.

    Another morsel from this game: The Phillies had a scoring threat in the 9th, getting runners on 1st & 2nd with 1 out. After Johnny Callison fanned, Walter Alston had his lefty relief ace Ron Perranoski give the IBB to the rookie Dick Allen, preferring to face the lefty Wes Covington (a pretty good hitter himself) with the bases loaded. This time it worked, as Covington grounded out.

    I don't know if Alston had the LH/RH data (did they have that stuff in 1964), but Allen hit .372 vs. LHP that year (.287 vs. righties). Covington was a platoon player by this point in his career, and should have been one all along; he hit just .219 against southpaws, with minimal power.

23. John Autin Says:
    April 27th, 2011 at 11:57 pm

    Meanwhile, back in the present....

    The Padres were shut out again Wednesday, their 7th shutout loss in their first 25 games. In the entire game-searchable era (1919-present), no team has been shut out more in its first 25 games of a season; 5 other teams have matched that mark, none since the '87 Royals.

    San Diego is now hitting .208 as a team and averaging 2.8 R/G. It was the 14th time the Pads have scored 2 runs or less so far, and they've lost all 14. On the bright side, they're 9-2 when they score 3 or more.

24. Johnny Twisto Says:
    April 28th, 2011 at 12:48 am

    My antagonist was so adamant in saying the TRIPLE CROWN proved nothing. But I kept saying, "yes it does"

    I feel ya. It drives me crazy when I see people write "RBI are meaningless." And I see it a lot. They're absolutely not meaningless. Those are real runs scored, not theoretical runs -- and I say this as a person who is fascinated by all various runs-created metrics and who has a spreadsheet set up to calculate them myself. RBI may not mean what some people think they mean, but they do mean *something*.

25. Gerry Says:
    April 28th, 2011 at 1:42 am

    Willie Bloomquist, the only player to make the list twice.

26. Gerry Says:
    April 28th, 2011 at 1:44 am

    Mostly one-run games - no surprise there, but the Knoblauch game is very different, final score 8-4.

27. Nash Bruce Says:
    April 28th, 2011 at 5:23 am

    (disclaimer: please forgive the following, as the musings of someone, who is not so famiIiar, with the most recent baseball sabermetrics- I love baseball, and your site, but I'm not a statistician......)
    I can't wait, until defensive analysis, catches up with everything else. I've seen it said (was it Bill James??) that defense, means (at most?) 17%, or something, of baseball outcomes.....that most of baseball, comes down, to the "three true outcomes".....but I would suspect, that this would change through eras. For example, maybe defense meant less in recent steroid times. Now, however, with homers on the way down, creating runs becomes more important, as does defensing the attempts at doing so. Furthermore, there will be more pitchers, willing to throw strikes- as well as more pitchers, able to do so, as we get further and further away from expansion times.
    I don't know if this "hit f/x or play f/x" is the program, that analyzes every pitch ,and the tendencies of hitters, to hit whatever pitch, wherever. I did see some sort of program, on something similar, though, and can only say, that baseball is far behind football on this idea: on "this" count (down)we need to throw "this" pitch(defensive formation) because "this" hitter.....and then the adjustments of the hitter, and THEN the counter-adjustments of the defense.....etc etc.
    And, the defense will be positioned appropriately, according to where the pitch is thrown, and what the pitch is, and the tendency of the hitter.
    I think that defense is going to be valued at a much higher rate than 17% (if I did get that number correct) in the near future.
    I guess that point of my rambling, is that I hope, that defensive statistics catch up, by that point. (Ask a pitcher, how important having a good defense behind him is, when he has to throw a strike, in a hitters count.)

28. dukeofflatbush Says:
    April 28th, 2011 at 8:44 am

    @ Nash,

    You said it better than I've been thinking it.
    The other day, CHISOX/YANKEE game, and Dunn hits a bullet up the middle, that 99% of the time is a clean single. But the Yanks (wisely) employ the shift, and Jeter makes a routine play on the first-base side of second.
    A few of your points are evident here.
    1. Jeter made the play simply by manager/coach positioning, but his range will go up, as will his dWAR.
    2. Dunn gets an out, even though he hit the hardest ball of the game, in a place hitters are told to aim for.

    I've thought for some time that a simple rating on fielded balls could greatly show a fielders contributions better than they are shown today.

    For instance; 1-4, 4 being the most difficult.
    So If two shortstops both have 9 chances in a game, but one guy has gone in the hole, dove, over the shouldered or other 'above average play, it could be weighted.
    Obviously this will need a criteria, a certain objectivity and an official or unofficial scorer.
    Because lets face it, a Super Sam Fuld catch has more value than a pop to right but shows up the same on the boxscore, and after a week of ESPN highlights, is generally forgotten.
    To touch on your point again, with other sports, for example, the NFL, realizes not all catches are equal, so they have a stat called 'YARDS AFTER CATCH' - So the receiver who turns a 6 yard route to the flat into a 30 yard gain, gets more credit than the Quaterback who had little impact on the play once the ball left his hands.

    just a pre-coffee thought

29. John Autin Says:
    April 28th, 2011 at 12:35 pm

    @28, Duke -- It would be nice to have an objective rating of every defensive play. The trouble is getting it truly objective. For instance, we're all biased towards diving catches, and in the heat of the moment we almost can't help rating such plays highly. But not all diving catches are great plays; Bill James wrote that, while some third basemen have the range of "one step and a dive," Carney Lansford's entire range was just the dive. And not all great plays involve diving, tumbling, or otherwise creating a "web gems" spectacle.

    Another example: A couple of nights ago, Rick Ankiel made the ESPN highlights with what seemed to be a terrific running catch in right-center of a Carlos Beltran drive. But if you saw the whole play, you saw that Ankiel misread the ball; he changed direction in mid-pursuit, cutting back towards the wall just in time to grab it. Some CFs (such as Beltran in his prime) would have made a better read and made that catch in their hip pocket.

    I do enjoy juicy highlight plays. But I prefer my defensive analysis dry: Give me a count of plays made, along with the context to weigh that number, and I'll trust that the diving catches even out over time.

30. John H. Says:
    April 28th, 2011 at 3:39 pm

    I don't have a problem with giving positive WPA in a losing effort. If a guy is helping his team at the time, he should get credit. However, here's a different scenario that WPA doesn't quite pick up on:

    Suppose that a team is winning 5-0 in the 5th and a guy hits a grand slam to make it 9-0. Well, he won't get much WPA for that because his team was already pretty far ahead. But then suppose that it turns into a wild game, the bullpen almost blows the lead, and they end up winning 9-8. That grand slam looks pretty important now, doesn't it? What may not seem like a key play at the time, could turn out to be huge. Any ideas on how to account for that?

31. Joseph Says:
    April 29th, 2011 at 6:33 pm

    @ # 8, John: Thanks John. Interesting.