42 Responses to “Highest WPA in a blown save”

1. Mark R. Says:
   January 15th, 2011 at 9:15 am

   The definition of blown saves is totally bogus. Notice that most of the pitchers on these lists allowed no runs. It is totally asinine to charge anyone with a blown save for allowing runs that he didn't allow.

2. BSK Says:
   January 15th, 2011 at 10:03 am

   How can you have a WPA higher than 1?

3. Andy Says:
   January 15th, 2011 at 10:12 am

   BSK, I think your question actually is how can it be higher than 0.5. At the beginning of the game, the odds either team wins is 50%, and over the course of the game, one team nets +0.50 WP and the other team nets -0.50 WP. The WP can be spread among pitchers and hitters and will consist of both negatives and positives that total either +0.50 or -0.50.

   A single player can easily achieve more than 0.50. Think of this extreme example. The visiting team scores 3 runs in the top of the first, so they already have a good chance of winning going into the bottom of the first. The home team's cleanup hitter ends up hitting a grand slam in the bottom of the first, a big +WP event. Then, over the next couple of innings, the visiting team rallies and goes up by 3 runs again. The next time the home team cleanup hitter is up, he hits another grand slam. Another big +WP event. Let's say that happens two more times, including in the bottom of the 9th, when the same guy hits a walk-off grand slam rallying his team from 3 runs down. That batter's WPA for the game is going to be really high--well over 1.0 I'm sure and maybe over 2.0.

   There are lots of other scenarios where a player can get a really high (or really negative) WPA---basically if his teammates are scoring a lot of WPA in the other direction. In the example I gave above, the cleanup hitter's team's pitchers earned a lot of negative WPA by pissing away lead after lead.

4. Dr. Doom Says:
   January 15th, 2011 at 10:32 am

   The bottom five on your third list are actually the ones I would consider most similar, because they all went 2 or fewer innings. That was one of the things that John Autin pointed out in the other thread - that Mariano's postseason blown saves have all been in two-inning situations. And, frankly, it's a lot easier to acquire more WPA the longer one has to pitch. But this is all very interesting. Particularly for the Guetterman game. He entered in the bottom of the eighth with no outs and the bases loaded, and induced an Eric Young grounder to short, who took the force at second, on which the tying run scored. Then, Charlie Hayes grounded into a double play, and the inning was over. The next inning, he allowed single - double - intentional walk - sac fly, and the game was over without him recording another out. So 1 inning, 2 hits, one walk, and it looks okay. But the result wasn't. And as one looks at that game, it seems like it shouldn't be worth all that WPA. But here were the leverage indeces on those six plays: 4.00, 3.67, 2.34, 3.11, 4.47, 4.90. Those are some heavy-leverage situations!

5. illdonk Says:
   January 15th, 2011 at 1:56 pm

   I'd guess you can have a WPA higher than one by pitching more than nine innings. Which makes perfect sense on one hand and no sense on the other.

   As for the definition of a blown save, it's somewhat unfair in the statistical sense, but I'm not sure how you can have a pitcher be eligible for a save yet not liable for a blown save.

6. DavidRF Says:
   January 15th, 2011 at 2:34 pm

   WPA > 1 doesn't have to do with the length of the game. Its just sum of the individual contributions. Your teammates would contribute negatively to counterbalance.

   For a 9-inning game, the highest I found was Jim Pagliaroni at 1.301 on 1965-09-21 (sorry no link... paste not working, I need to reboot my computer).

   Multiple game-changing two out hits in the late innings, one of them a two-out walk-off HR.

   For a pitcher who pitches a full inning, its harder to do because you're responsible for both the good and bad events of an inning, but the same concept of it "adding up" is still there. Pitchers get 0.13 WPA for pitching a scoreless half-inning in a tie game from innings 9 and on.

7. SJBlonger Says:
   January 15th, 2011 at 4:41 pm

   What the heck: http://www.baseball-reference.com/boxes/PIT/PIT196509210.shtml

8. Johnny Twisto Says:
   January 15th, 2011 at 8:07 pm

   The definition of blown saves isn't bogus or asinine. It is what it is. The pitcher was on the mound when the lead was blown. It doesn't mean he's the one most responsible. Obviously not all blown saves are bad games.

   Most career blown saves with WPA > 0: http://bbref.com/pi/shareit/xplFY

   Since 1990: http://bbref.com/pi/shareit/9x45a

9. John Autin Says:
   January 16th, 2011 at 2:33 am

   Further to DavidRF @6 -- Pagliaroni's second big hit was a two-out, two-run walk-off HR that erased a deficit. (Far greater WPA than if the game had been tied when he hit it.)

10. John Autin Says:
    January 16th, 2011 at 2:51 am

    Mark R @1 and Johnny Twisto @8, re blown saves:

    I think you're both right. The problem isn't with the specific definition of a blown save; it's the fact that, in common usage, a blown save is treated as if it always represents a performance of X quality, when in fact it can represent anything from A to Z.

    It's similar to the problem with the definition of pitcher wins and losses (and, of course, saves). And I do think that those who have the power to define the terms bear some blame here. The definitions of the events are not specific enough to justify the emotion-laden words they have been attached to. The definitions do not sufficiently target good and bad performances.

    I won't go as far as Mark did in absolving every reliever who allows an inherited runner to score. For instance, if you come in with 2 out and a runner on first, and you let that runner score, I think you did a bad job and deserve to get a blown save.

    P.S. Trivia question:
    In a tie game with no runners on base, Pitcher A is relieved by Pitcher B. Pitcher B gets charged with a blown save, and Pitcher A is charged with a loss. How can this happen?

11. John Autin Says:
    January 16th, 2011 at 3:07 am

    Getting back to the original list, I just have to say: "SATCHEL!!!"

    By the way, that wasn't the longest relief outing of Satchel's MLB career, nor even of that season. Ten days after that game (and after four intervening appearances, with three saves and a win), he went 10.2 innings of relief in a 19-inning game against Cleveland, his former team. Paige came on to start the 9th in a tie game and kept Cleveland off the board for 10 innings. He was in line for the victory when Jim Delsing drove in the go-ahead run in the top of the 19th. But with 2 out in the bottom half, Al Rosen doubled home Bobby Avila to tie the game, and (after an IBB to Larry Doby), Hank Majeski singled in the winning run.

    Satchel worked mostly in relief that year, as usual, but he made 6 starts and tossed 3 CG and 2 shutouts, including a 12-inning, 1-0 affair.

    (Random Satchel feat: In 179 MLB games, he handled 79 chances and was never charged with an error.)

12. John Autin Says:
    January 16th, 2011 at 3:14 am

    P.S. Satchel's MLB career ERA+, WHIP and K/BB ratio were all better than Bob Feller's, and his shutout percentage was over 50% higher. (I'm just having a little fun, guys. I know there was some cherry-picking of opponents for Satchel to start against.)

13. Andy Says:
    January 16th, 2011 at 3:42 am

    RE: JA's trivia question...first thought is if Pitcher A goes to a 3-ball count on the batter, is relieved by Pitcher B, who then walks the batter. I believe the runner then belongs to Pitcher A and can be the losing run.

14. John Autin Says:
    January 16th, 2011 at 5:02 am

    Andy -- Yes, that's the principle. But only 2 balls by the first pitcher are necessary to charge the walk to him, if the count is 2-0 or 2-1.
    -------------------------
    Rule 10.18(h) A relief pitcher shall not be held accountable when the first batter to whom he pitches reaches first base on four called balls if such batter has a decided advantage in the ball and strike count when pitchers are changed.

    (1) If, when pitchers are changed, the count is
    2 balls, no strike,
    2 balls, 1 strike,
    3 balls, no strike,
    3 balls, 1 strike,
    3 balls, 2 strikes,

    and the batter gets a base on balls, charge that batter and the base on balls to the preceding pitcher, not to the relief pitcher.

15. Andy Says:
    January 16th, 2011 at 9:16 am

    Yeah I just gave 3 balls as an example. Another good trivia question is how a pitcher can get a win or a save without throwing a pitch, and I can think of a way a pitcher can blow a save without throwing a pitch too. I have blogged about some of these before.

16. BSK Says:
    January 16th, 2011 at 12:28 pm

    Andy-

    Thanks. So the net gain for a winning team will be .5, but individual players might achieve more than that because players on their team will have a negative WPA (from allowing the other team to accrue positive WPA)? Do I have that right?

17. bells Says:
    January 16th, 2011 at 1:10 pm

    Yeah BSK, it's situational. The example Andy cites is perfect, but maybe it would make more sense with numbers, if anyone else isn't clear. It might be simplest just to use an example of 2 at bats (not realistic, but you get the point).

    Let's say this guy gets a grand slam when his team's down 3-0 in the first. The teams Win Probability goes from .35 (I'm making these numbers up for simplicity) to .65. Guy gets WPA of .30.

    Then his team gives up the lead, and are down by 3 in the bottom of the 9th, win probability is back at .30 for his team. The pitchers on his team have borne the responsibility, via negative WPA, of the -.35 difference. Guy still has .30 WPA from his previous at bat. He hits a grand slam, sending the win probability from .30 to 1.00 (because they won), giving him WPA of .70 for the play, and 1.00 for the game. There's no real upper limit, if they played a 22 inning game and the same player kept making plays to get his team to come back, it could theoretically be a huge number of WPA by the end of the game, because it just tallies the probability added for each play that the player's in, and that player, because of the back-and-forth possibilities of games, can be inserted into situations where WPA can be maximized.

    I think that's right, but correct me if it's not.

18. BSK Says:
    January 16th, 2011 at 1:17 pm

    Does the positive WPA accrued by the batter correlate directly to the negative WPA accrued by the pitcher? If that batter hits the grand slam and gets .3WPA does it follow that the pitcher allowing the grand slam gets -.3WPA?

    It's my understanding that the primary shortcoming of WPA is that it assigns all credit to pitcher and batter. Is this correct?

19. Andy Says:
    January 16th, 2011 at 1:42 pm

    Good example, bells, and BSK you are right on both accounts. After each play, the batter and the pitcher get assigned the same WPA for the play that happened, only one will be positive and one will be negative. For any given game, the sum of WPA across all of one team's players (hitters and pitchers) must be either +0.50 (if they won) or -0.50 (if they lost).
    And yes, the biggest shortcoming of WPA is that it doesn't differentiate between a line drive, a swinging bunt, or even an error if the before-and-after score and baserunner scenarios match. So a pitcher can get negative WPA that's really his defense's fault or the batter can get credit for offense he doesn't deserve.

20. BSK Says:
    January 16th, 2011 at 3:04 pm

    Thanks, Andy. I had a basic sense of WPA but didn't know all the details.

    It sounds like it's probably more accurate to say a given play had a WPA than a specific player, though this obviously doesn't do anything for individual player valuation. In the grand scheme, it's still a really useful tool as is.

21. oneblankspace Says:
    January 16th, 2011 at 7:34 pm

    I once saw Lance Painter come on in relief and finish an inning without throwing a pitch. He did, however, throw to first, and the runner was not back in time.

    If Baker relieves Abel in a tie game, Baker would be the pitcher of record (unless an inherited runner scores to take the lead) and would not be entering in a save situation. Baker is not eligible for a blown save or a save (or a hold, for that matter).

22. Andy Says:
    January 16th, 2011 at 7:50 pm

    If Baker enters a tie game with a runner on third, an errant pickoff throw or balk could score the runner, earning a blown save.

23. kds Says:
    January 16th, 2011 at 9:37 pm

    Only exception to the explanations of WPA above (#18,!9); if there is a stolen base, caught stealing, pick-off, wild pitch, passed ball or balk; the base runner will get the WPA + or -, in all cases involving a plate appearance the batter gets the credit/blame. On the other side, the pitcher gets everything, even with an error or some other unusual play. The WPA published here, (and most other web sites with which I'm familiar), may be a little artificial. For example, there is no home field advantage. If the Yankees are playing the Royals at NYS the game starts 50-50, no awareness is given to team quality, or to the quality of individual players.

24. Andy Says:
    January 16th, 2011 at 9:40 pm

    The other thing that it doesn't take into account is batting order position that's up. In other words, for a given situation, the WPA will be the same regardless of whether the #8 or #3 hitter is up.

25. Johnny Twisto Says:
    January 16th, 2011 at 10:31 pm

    If the Yankees are playing the Royals at NYS the game starts 50-50, no awareness is given to team quality, or to the quality of individual players.

    Of course, if it were adjusted for individual players, the effect would be to penalize good players for being good. So it is more impressive for the Royals to come back against Mariano Rivera than against Todd Jones, but if you give Jones more credit for closing out the same game as Rivera, and adjust every player and situation in this manner, it seems like the practical effect would be that eventually everyone ends up with 0 WPA because they did what they were going to do.

    Adding home field advantage would probably be an improvement however.

26. BSK Says:
    January 16th, 2011 at 11:03 pm

    JT-

    I once mused that it'd be interesting to see what WPA would look like when adjusted for individual players.

    For instance, WPA is partly based on looking at the team's chances of winning given a certain situation. What if we could say that a team down by 2 in the 9th has a 10% chance of winning, all things being equal, but versus Mariano Rivera, they only have a 9% chance of winning. As you point out, it'd risk penalizing these guys, but I think it'd be fascinating to see what an actual teams chances of winning are versus what the historical data tells us about all teams.

27. John Autin Says:
    January 17th, 2011 at 12:41 am

    Sorry to anyone who tried my trivia question -- I completely butchered it. Obviously, a reliever who enters with the game tied cannot be charged with a blown save. I was trying to be clever with the concept that a pitcher can be charged with a walk when he departs in the middle of an at-bat with a count of at least 2 balls and behind in the count, if the reliever goes on to complete the walk. Thus, a pitcher can be charged with a loss even if he leaves with the game tied and no one on base. But I overreached for the blown save. Should never try to be clever late at night.

    P.S. As a matter of sheer accounting, does anyone know for sure whether, in the "split walk" scenario I described above, the pitcher charged with the walk is also credited with a batter faced (BF)? -- with the corollary that the pitcher who completes the walk (but does not have it charged to him) would not be credited with a BF?

28. Andy Says:
    January 17th, 2011 at 7:48 am

    #27, I believe another sharp reader once posted a trivia question about that, i.e. how to throw pitches in a game without getting credit for a batter faced, and the answer lies in your scenario (although I do not remember if the pitch separation that determines who owns the walk or strikeout also determines who gets the batter faced.)

29. Andy Says:
    January 17th, 2011 at 8:04 am

    I am fascinated by #25 and #26 above. Can you guys see the difference in your perspective?

    JT is looking at it from how players are measured after their performance whereas BSK is looking at it from in-game prediction based on situation and historical league-wide WPA values.

    Of course you guys are talking ultimately about the same things, and I know you both understand that. But it's neat to think about how this stat is being used differently from others.

    During the early part of Wade Boggs' career, when his career average hovered around .360, I doubt many people watched him come up to the plate and think---ooooh, yeah, Boggs has a 36% chance of getting a hit here, whereas if Dwight Evans were up, he'd have only a 27% chance. We all recognize that the likelihood of a particular batter getting a hit in any given plate appearance depends hugely on circumstances, including the identity and fatigue level of the pitcher and the baserunner/out situation. We all realize that going just by basic batting average is not appropriate because there are some situations in which a batter will hit consistently above his own career average and some where he'll hit consistently lower.

    Most fans still use the basic batting average at the end of the season as an indicator of what happened.

    So, the question is--does WPA really have much predictive value? I tend to think not SO much--especially in really specific situations. If it's the bottom of the 9th, Cardinals down by 1, 2 outs, runner on 2nd, and Albert Pujols is hitting off Oliver Perez, the Cardinals probably have a significantly better chance of winning the game than what WPA would tell us, just like if it's Cesar Izturis hitting off Mariano Rivera, there's probably a lower chance of Izturis' team winning than what WPA would say. I'm not sure that the WPA is anything better than a very broad indicator in cases like that...are the odds 1%, 10%, or what? In reality they may be 50% higher or lower, but an order of magnitude is helpful.

    WPA is probably more meaningful earlier in games when there are still a lot of events that need to happen, or when considering large numbers of games in an average sense. In other words, if we take those bottom-of-the-ninth scenarios, it's true that looking at a whole pile of those games together, the outcome will probably match the WPA's prediction pretty closely.

30. Mark R. Says:
    January 17th, 2011 at 10:17 am

    A statistic should at least attempt to measure a player's actual individual performance, not just "something that happens to him while he's on the mound". Blown Saves is very poorly defined and it should be improved.

31. Johnny Twisto Says:
    January 17th, 2011 at 5:18 pm

    If you want to get anal about it (as I often do), nothing really measures a player's individual performance. A guy frozen by three unhittable pitches off the low-outside corner has the same "K" next to his name as someone who whiffs on a hanging curve.

    Your problem isn't really with the definition of a Blown Save, which has no definition of its own. It's with the definition of a Save. Fair enough, you're certainly not the first one to say the Save is a flawed stat. But most people's complaints with saves is that they are too easy to get. Why credit a pitcher for holding a three-run lead for an inning? The save opportunities in the games above are the ones that most people think are legitimate -- come in with runners on base, gotta squirm out of the inning without letting them in. So if that's a legitimate save, then failing to do it must be a blown save.

    You just need to accept that not all blown saves are equal, just like some wins are shutouts and some are ugly 5-inning efforts, like some hits are liners smoked up the middle and some are swinging bunts, etc. At least with blown saves, we have the boxscores available to show us which were tough and which were just poor pitching, and we can further break them down that way if we choose to. We don't always know if a hit was well-hit, if a stolen base was due to an ump's blown call, if a triple was just misplayed by the outfielder.

32. BSK Says:
    January 17th, 2011 at 11:32 pm

    Andy-

    At the risk of bringing up the infamous "Phillies are Done" post, I think it's interesting to consider whether a stat is truly predictive or simply presents a historical model. As you pointed out, in the history of the sport, a given base-out situation may have yielded wins X% of the time. But put me on the mound, Barry Bonds at the plate, and Adam Dunn manning every outfield position, and that % is pretty useless. The question is, can that % be calculated with any degree of accuracy? And, if it could, I think a really interesting question would be what would we consider statistically significant. If I told you that Mo Rivera offered his team a 1% better chance of winning than any other closer in history, would that impress you? Would you think it too low? Too high? Just right? I can't even wrap my head around what would be a reasonable expectation for how an individual player could impact a game or even a specific play.

33. BSK Says:
    January 17th, 2011 at 11:35 pm

    To clarify, when I say "can that % be calculated" I mean, can the % of winning be calculated for the specific situation, independent of what history has told us.

    I always think about when announcers trot out things like, "In the history of the sport, no team has come back from 3-0 down," or something. The implication is that it can't be done because teams in that scenario have a 0% success rate. Obviously, it CAN be done, it's just really, really unlikely. But previous success rate rarely = likelihood.*

    * Of course, this blurb is ignorant of the Sox in '04, but it's the easiest example I could offer of the top of my head.

34. kds Says:
    January 18th, 2011 at 2:11 am

    BSK, If I were to try to calculate "individual" win expectation numbers, I would do something like the following. For pitchers, in general, I would calculate RA+, similar to ERA+ but looking at all runs, and taking team defense into account. Then I would take that and use it as a run environment to manufacture new WE, (Win Expectation), charts to give us WPA. So, for example, if Rivera's RA+ = 200, then his run environment would be half the usual. If the usual were 5 runs/game then with Mo pitching it would be 2.5/game, which is < Astrodome in the '60s. The WE table would change accordingly; a 2 run lead might well have a higher win % then a 3 run lead in the 5 runs/game context. For hitters it is a little more complicated since we don't usually rate hitters on a scale like RA+. OPS+ is not best for a variety of reasons. RC+ (Runs Created), is what we want except that it also includes baserunning, which is not relevant for our hitters. (Take batting runs above average, figure what the average number would be for that league, park, and number of plate appearances, add in the braa, divide by the average number we got, and multiply by 100.) Again we are figuring a specific run environment to get the individual WE and WPA. We could even go further and split things by handedness; Pujols vs RHP separately from Pujols vs LHP etc. Various analysts have pointed out that since facing Rivera is like playing in deadball conditions, deadball strategies are more likely to be correct than facing a pitcher who provides a higher run environment and different WPA.

35. Mark R. Says:
    January 18th, 2011 at 9:08 am

    Don't put words in my mouth. My problem is NOT with the definition of a Save. I disagree with Blown Saves. And NO I do not have to accept a ridiculously flawed definition. If a guy comes into a game with his team clinging to a 1 run lead and there are already runners in scoring position, it should not be a blown save if the inherited runners score. In many cases, if a guy allows just 1 of the inherited runners to score and gets out of the inning he has done a better job than average.

36. Johnny Twisto Says:
    January 18th, 2011 at 1:24 pm

    Relax son.

    A blown save is the failure to convert a save. That seems fairly obvious. If a pitcher enters in a save situation, he can either get a save, a blown save, or a hold. They are all based on the definition of a save. So if you have a problem with any of these things, you must have a problem with the definition of a save. Otherwise, please tell me how you would redefine a blown save.

    Hitting a line drive is doing a better job than average, but it can be caught. Allowing 1 run in 9 IP is better than average, but can result in a loss. These statistics weren't defined to be "fair," just to record what happens. If blown saves doesn't tell you what you want to know, use a different stat.

    But once you got all upset about words being put in your mouth (?!), you apparently didn't read or consider anything else I wrote, so whatever.

37. John Autin Says:
    January 19th, 2011 at 5:09 am

    @36, Johnny Twisto -- "These statistics weren't defined to be "fair," just to record what happens."

    But since wins and losses, saves and blown saves, are the most frequently cited pitcher stats, shouldn't they be designed to be, if not perfectly "fair," at least as meaningful as we can make them?

    Question for you: If we simply stopped assigning "wins" and "losses" to individual pitchers, would our perception of pitcher performance and value be better or worse, overall?

    I think that terms like "base hit" and "home run" are fundamentally different from the pitcher decision terms. We need the former in order to describe the events of the game. But do we really need the latter? If you want a brief account of a game and who was instrumental in the outcome, is it important to know what pitcher was in the game when the winning team took its final lead?

    And if we do want to retain the concept of pitcher "wins," why should we feel attached to the current definition, which we know is not terribly
    meaningful? Surely we could make it better without getting too complicated.

    (More on this at a later date.)

38. Johnny Twisto Says:
    January 19th, 2011 at 12:01 pm

    You are certainly right that "hits" and "wins" are different types of stats.

    While acknowledging that wins and saves are imperfect, I have no desire to get rid of them or even change them. They help tie together the history of baseball. What would happen if Cy Young no longer had 511 wins? I doubt I could even walk straight.

    I just look beyond those stats to other information to find out what I want to know. Now if "wins" and "saves" didn't exist, would the general public have a better perception of pitcher performance? Perhaps. More likely, someone would come up with new stats, even more ill-conceived, for the public to latch onto. It is a soundbite culture and most people aren't interested or energetic enough to learn the nuances of anything. Just give them the main idea in digestible portions -- even if it is misleading.

    In conversing with the misled, we can reach out and explain why Stat X is more meaningful than Stat Y. Some will be interested and want to learn more. Most will be satisfied they already know what they need to know. We won't change human nature.

39. Mark R. Says:
    January 20th, 2011 at 6:07 pm

    First of all, the concept of the Save was invented without any thought or mention of Blown Saves or Holds. So yes I can agree for the most part with the Save definition but not agree with those derivatives. The Hold could be a good stat if it had a better defintion but the current definition is an absolute joke.

    Suppose that a reliever enters the game in the 9th with a 3 run lead. He gets the first batter out, but then he gives up a double, a 2-run homer, then walks the bases loaded before getting yanked. Then after he leaves all 3 runners come around to score. He gets charged with 5 runs in one-third of an inning, totally blows the lead and probably gets tagged with a loss and yet somehow that's a "Hold". Totally asinine.

40. Mark R. Says:
    January 20th, 2011 at 6:14 pm

    Also, suppose that one reliever again enters with a 3 run lead, gives up 2 runs while recording only 1 out then walks the bases loaded. Then the next reliever enters the game and gets the next batter to sac fly out (which ties the game, but is not charged to him) then retires the side with no more damage.

    The pitcher who MADE the mess gets a "hold", and the pitcher who CLEANED UP the mess gets a "blown save". This is just STUPID, STUPID, STUPID. The current definitions of Hold and Blown Save are totally unacceptable.

    Why are some people persistently against changing a definition of a stat that was stupidly designed by stupid people in the first place? That is just stupid.

41. Mark R. Says:
    January 20th, 2011 at 6:20 pm

    Here are some changes that need to be made to the Hold stat.

    First of all, a pitcher should have to pitch at least 2/3 of an inning. 1/3 of an inning is just not long enough. A one batter appearance is frivolous anyway. It cheapens the stat way too much to allow anything less than recording 2 outs to be a Hold.

    Secondly, a pitcher should never get a Hold if the runners that he leaves on base come around to tie the score or lose the lead. You didn't "hold" the lead if the runners you let on base gave up the lead and you were pitching so ineffectively that you had to be yanked in the middle of the inning with runners on base. It's plain and simple.

42. Mark R. Says:
    January 20th, 2011 at 6:25 pm

    Also, with the Hold stat, a middle reliever should get credit for a hold if he enters a tie game, pitches at least 2/3 of an inning and gives up no runs in the appearance.

    Pitching in a tie game is very high leverage, and the Hold stat should reward relievers who keep the game tied so that their offense can win it later. A Hold does not have to be defined by Save opportunities. The Hold can and should be its own stat.