FanGraphs Baseball


RSS feed for comments on this post.

  1. What about Bochy letting Cain hit in the 7th with a runner on second, then pitch to two batters in the 8th? At that point in the game, any added chance at plating that 3rd run would have been significant.

    Unless, of course, Bochy foresaw Washington’s bullpen implosion.

    Comment by Andrew Kneeland — October 28, 2010 @ 11:17 pm

  2. What about David Murphy starting over Vlad? Also, what about the Giants likely winning the game 2-0 even if Ogando and Feliz had come in and gotten the third out without allowing any runs?

    Comment by WY — October 28, 2010 @ 11:18 pm

  3. For your first question, see .

    Your second point is quite true, having the best relievers in the world would have still meant a 2-0 deficit rather than giving up another touchdown in the 9th.

    Comment by Jason B — October 28, 2010 @ 11:29 pm

  4. ill say that is true, but still that lineup is capable of putting up 2 runs, so at least they have some chance. maybe they would have lost, but that’s the fallacy of the determined outcome. still they had a better WE by keeping 2-0 than 9-0, and you can’t fault the manager for using feliz to keep it there for his potential come back. of course, he didn’t go that way…

    Comment by phoenix — October 28, 2010 @ 11:48 pm

  5. As soon as Darren O’Day was pulled I was like “uh oh”. I liked waht I saw from O’Day, giving up a single shouldn’t have negated the fact that his breaking pitch was working to the inside and his command was on.

    If at the beginning of the season I informed you that a Rangers/Giants WS was taking place and asked you what bullpen core was better, what would you pick?


    The deadline-deal revamped Giants bullpen happens to be absolutely lights out all postseason, and the Rangers..have been consistently blowing games, whether or not they have the lead. Don’t even get me started on having Derrick Holland come in to walk 3 batters.

    Comment by exxrox — October 28, 2010 @ 11:55 pm

  6. I think Cain was the best option for the 8th. The SF bullpen didn’t really shut anyone down yesterday, and Cain was lights out.

    Comment by johng — October 29, 2010 @ 12:06 am

  7. I don’t understand the logic of telling Holland to walk 3 batters. That put the game away for the Giants.

    Comment by johng — October 29, 2010 @ 12:08 am

  8. Ogando threw 25 pitches yesterday…. isn’t that seemingly more important than inning count? Mark Lowe (who was used) threw 23 pitches but ‘only’ pitched 0.2 innings.

    Pretty poor that innings pitched would be used as a justification to not use Ogando without looking at the pitch count as well and the objective SABR guy apparently agrees. How many 20pitch innings do closers or other relievers have and then come back the next day with no problem? The only real difference is the extra 8 warmup pitches going out a 2nd inning.

    I’m amazed that Dave would resort to the he couldn’t use him because of the innings pitched and not look at the actual pitch count. I can understand Washington making this foolhardy mistake though.

    Comment by Joe — October 29, 2010 @ 12:10 am

  9. Yes, if only Ron Washington had been very smart, his team would have lost instead of losing. With the bases loaded and a two run deficit in the 8th, the Rangers had a 6% chance of winning. While Washington does not come out looking smart, let’s not pretend like he could have managed his way out of the situation when Holland walked the bases loaded and whacked that probability down to 5%. What does Neftali Feliz give you? A +10% chance of getting an out in a situation where you’re almost guaranteed to lose anyway?

    Comment by philosofool — October 29, 2010 @ 12:50 am

  10. I think 25 pitches over 2 innings is worse than over 1, because you sit, rest, cool down a bit and have to go out again, though that is balanced by relievers who warm up, sit down, warmed up and then pitched.

    That being said, unless there was a particular issue with Ogando, I think he could/ should have been used today, I might shy way from him if it was a regaular season game, but in the playoffs, unless you literally can’t use someone, you use them.

    Comment by quint — October 29, 2010 @ 12:52 am

  11. Its a WS game, that 10% could be the series. Its the world series, you have to give your team as much chance of winning any game where you are down by anything less than 5.

    Washington failed that, yes the Rangers almost certianly would have lost, but Washington managed in a way that didn’t give his team the best chance of coming back. In a WS game that close, that is criminal.

    Comment by quint — October 29, 2010 @ 12:54 am


    Comment by Ratto's Pants — October 29, 2010 @ 1:06 am

  13. What I’d be more irked about if I supported the Rangers:

    6 outs left, and you give one to Borbon and one to Andrus with Vlad available. FAIL

    Comment by Aussiedodger — October 29, 2010 @ 1:11 am

  14. Oh yeah, I don’t question that generally you want the best man for the job to maximize your chances of winning, even if they’re slim. A 5-10% chance of coming back from a 2-0 deficit in the 9th >> something like a 0.1 – 0.2% chance of recovering from a 9-0 deficit.

    Particularly with an off day following, and this being the known last series for both teams, no need to keep any bullets in the chamber, so to speak. Smoke ‘em if you got ‘em.

    (Smoke your…bullets? Mixed metaphors…)

    Comment by Jason B — October 29, 2010 @ 1:13 am

  15. After he loads ‘em up, how do you leave him in to pitch Huff? Why isn’t Feliz or someone, ANYONE else ready there? Holland is *clearly* off and has zero margin for error at that point. He had thrown like eight straight balls to walk the first two hitters!

    Comment by Jason B — October 29, 2010 @ 1:18 am

  16. Andrus has at least hit quite well during October. Agreed, Vlad >> Borbon.

    Comment by Jason B — October 29, 2010 @ 1:19 am

  17. Washington didn’t even have anyone in the bullpen warming until the game was out of reach.

    Comment by Daver — October 29, 2010 @ 1:40 am

  18. If he threw those 25 pitches over say 1.1 innings, my guess is folks would say he should have been available. Or if he had come in mid inning, pitched part the following inning (racking up 1 total IP stat-wise and 25 pitches), I’m also fairly certain people would say he should have been available.

    This is just simple “old school” perception of 2 innings being a lot of work to come back the next day from…. I expect that out of Washington but not folks who spend a lot of time hammering people for being slaves to old school conventions. You would think people would look at pitches thrown and/or # of batters faced and/or recent workload rather than just a simple determination that 2 innings = unavailable the next day

    Comment by Joe — October 29, 2010 @ 2:18 am

  19. Yep. You ride your hot hand. Run prevention was much more important than run creation at that point in the game. I love call to leave Cain in the game.

    Comment by this guy — October 29, 2010 @ 2:30 am

  20. There are somethings you can’t defend against. The Giants scoring 11 and 9 runs back to back is like a meteor strike, its just something you aren’t looking for. But yeah what was Washington saving Feliz for? He didn’t pitch yesterday, tomorrow is an off day, even if the game is out of reach get the kid some work. Feliz hasn’t pitched since the 22nd, next chance will be the 30th. I don’t think overwork should be a factor.

    Comment by Kazinski — October 29, 2010 @ 2:31 am

  21. He doesnt want to burn out his pen in what he hopes is a 7 game series. Philosofools point is quite valid. You don’t burn out your pen to preserve a loss, especially after your ace gets shelled and you have Hunter, who would be lucky to go 5 innings.

    Managing a short series is different, but you’re taking it too far. Feliz and Ogando NEED to be fresh if they want any hope of winning ANY games this series. Don’t let the regular season stats fool you. The Rangers pen is a patchwork unit of inexperienced kids, who are being asked to battle a lineup of savvy veteran hitters in extremely high leverage situations. On the flip side, the Giants have a pen full of veteran arms who have been doing it all season. 2-0 late in the game vs a red hot Cain/Wilson. Washington had no choice but to throw shit at the wall and hope it sticks.

    Comment by this guy — October 29, 2010 @ 2:43 am

  22. Indeed.

    Comment by this guy — October 29, 2010 @ 2:44 am

  23. Stealing in the 8th was quite stupid too. You’re down 2 runs with the top of the lineup batting. If Renteria gets there in time, Posey guns down Andrus by 2 full strides. Very stupid.

    Comment by this guy — October 29, 2010 @ 2:46 am

  24. Then why even bother with other pitchers, let Lee and Feliz pitch every inning until their arms fall off.

    I understand the drive to squeeze every ounce of opportunity here, but I saw in 2002 what happens when you ask your relievers to leave everything on the field in every single game. By the time you get to game 6, and they’re throwing BP fastballs and blowing a 5-run lead.

    Comment by Bhaakon — October 29, 2010 @ 3:10 am

  25. Look, I understand that the Giants have the worst offense of the postseason teams, but that’s not the same as a bad offense. This isn’t the 2009 team, they’re going to score runs if you pitch like garbage.

    Comment by Bhaakon — October 29, 2010 @ 3:14 am

  26. I reject the notion that throwing Feliz or Ogando tonight would have caused them to be overworked – because Feliz was used so often during the ALCS?

    We are talking getting 1 out (when the giants started getting runners on in the 8th) with 2 pitches. Feliz in particular hadn’t pitched much in the ALCS and its an off day the next day. Get him warmed up to get Cody Ross out.

    Then you wouldn’t have had to screw around with your game 4 starter getting warmed up in half a minute in the bullpen – that will only have positive benefits?

    Bhaakon – at the rate the Rangers are going, there won’t be a game 5, yet alone a game 6. We are not talking about burning a whole bullpen, Wilson went deep, and the bullpen got some outs, wanted to get 1 more.

    Comment by quint — October 29, 2010 @ 5:09 am

  27. That he hasn’t pitched much the last three weeks isn’t irrelevant, but it isn’t particularly meaningful either. What’s important isn’t not was he has done in the past, but what he might have to do in the future. If the next 2-5 games are close, and Feliz is needed in all of them, maybe in multiple innings, then suddenly using him tonight becomes a bigger deal. You punt tonight 10% advantage (or whatever the difference in win probability between 2 down and 9 down in that situation is) and save him for higher leverage situations down the line. Furthermore, the 10% difference is misleading, that assumes that 1) Feliz is infallibly perfect, and 2) The other relievers are perfectly fallible, neither of which is the case.

    Comment by Bhaakon — October 29, 2010 @ 5:49 am

  28. i was surprised by that one… renteria gifted that steal to ian… he was out by a mile if he just catches the ball.

    Comment by richard — October 29, 2010 @ 7:42 am

  29. I imagine that Dave was speculating that that was the reason why Ogando wasn’t used. It was Washington who decided, for whatever reason, not to use Ogando. The only logical conclusion to draw is that Washington decided not to b/c he threw 2 innings the previous day. But maybe he chose not to use him for some other reason. The reason why is immaterial; why you choose to use it as a focus of your comment is strange, to say the least. Ogando should’ve been in the game and whether it was b/c he threw 2 innings the previous day or b/c the Rangers were behind or b/c Ogando’s ERA on Thursdays is 8.47 is irrelevant.

    Comment by chuckb — October 29, 2010 @ 8:02 am

  30. Which is all the more reason why Texas’s best relievers should have been in the game rather than “punting” game 2, as you suggested earlier, and going with O’Day, Lowe, and Kirkman. It’s the World Series. There are 2 off days out of the next 5 days. Either Ogando or Feliz should have been in the game.

    Comment by chuckb — October 29, 2010 @ 8:06 am

  31. What Chuck said. It’s Torre syndrome – saving your best arms for the situation that never comes.

    (Although he’s by no means the only offender…just perhaps the most egregious.)

    Comment by Jason B — October 29, 2010 @ 9:25 am

  32. This is the World Series. Everyone should have been available regardless of pitch counts and innings pitched. That being said, thanks coach Washington for leaving Holland in there.

    Comment by Kriz — October 29, 2010 @ 10:27 am

  33. Amen to this. If Andrus had been deservedly caught, Washington would have been justifiably beaten around.

    Comment by Souldrummer — October 29, 2010 @ 10:27 am

  34. Because that one game sample really tells us so much.

    Comment by Wally — October 29, 2010 @ 10:33 am

  35. Have you ever thrown 50 or 60 pitches (warm up + pitches in game) one day and tried throwing the next. Throwing any amount of pitches will tire your arm and maybe Ogando was tired and did not feel 100%. Maybe he felt 80% and Washington thought someone else 100% > Ogando at 80%.

    Maybe throughout the year Ogando has never pitched more than 20 pitches one day and thrown the next day. Maybe he did a couple times and his “stuff” wasn’t quite as good.

    Maybe the Rangers pitching coach saw Ogando in warm up and realized his velocity was down, or his breaking pitches weren’t moving.

    It is not logical to say that he threw 2 innings yesterday and that is why, because he threw 2 innings, that Ogando was not called into the game. Their are many variables at play.

    Comment by John — October 29, 2010 @ 10:36 am

  36. Bhaakon,

    I no one was saying that putting in Feliz would assur that 10% chance of winning the game hits, while using other relievers would assure it doesn’t. If you think the use of that 10% was misleading, its because you misunderstood it.

    Using Feliz simple increases your chances to record that last out in the 8th. Holland has a FIP around 4 and was quite clearly struggling with his command on this particular night. Lowe has a FIP of 4.5. Feliz has a FIP of about 3. Those are some big differences.

    With the off day now today, Feliz having not pitched in 10 days(?) and this being the world series. I think you have to bring him in to get that out and keep it a two run game.

    Comment by Wally — October 29, 2010 @ 11:00 am

  37. Elephant in the room….

    Mark Lowe

    Why is he all of a sudden on the Rangers WS roster?

    Comment by Bill — October 29, 2010 @ 12:25 pm

  38. everytime i see Bhaakon post for some reason i think of that Star Trek movie where Kirk(?) says KKHHAAANNNN!!!!!

    Comment by SFG FTW!!! — October 29, 2010 @ 12:39 pm

  39. I disagree. Cain was OK last night, not lights-out. 2K! He allowed plenty of balls in play. lots of flies to the warning track. He got a bit lucky and had good D behind him.

    Comment by brendan — October 29, 2010 @ 12:44 pm

  40. I’d like to see a post examining the wisdom of bringing in a lefty to face Schierholtz given Schierholtz’s eye-popping career splits:

    vs. RHP .249/.295/.373 (.668 ops) in 603 PAs
    vs. LHP .354/.390/.500 (.890 ops) in 155 PAs

    Granted these are small samples (especially vs. LHP) but wouldn’t The Book still recommend treating him as a RHH if you regressed these numbers against the assumed platoon advantage? Two games in a row now Washington has brought in a lefty to face Schierholtz and it hasn’t worked either time.

    Comment by ElJimador — October 29, 2010 @ 12:50 pm

  41. it was an embarrassment.

    -should have pinch hitted with vlad, then if he got on, pinch run with borbon
    -should have brought in neftali to get that 1 out – since we have an off day tomorrow.

    ugh, can’t believe the rangers did so well and now are doing so poorly, i guess poor management decisions have finally caught up to them

    Comment by Rangers Fan — October 29, 2010 @ 12:58 pm

  42. Just would like folks who rail on old school decisions to not also rely on those same perceptions when it suits them….

    The 2inning stuff is speculation, especially looking at the low pitch count involved, only 6 batters faced, and no significant workload prior to that day. Maybe Ogando wasn’t available, but it was random speculation on Dave’s part (unless Dave spoke to Washington about it) and it wasn’t even necessary for the article, except to be an attempt to just take another shot at Washington.

    Also worth noting – while Game 1 ended up being a blow out, Ogando came in when it was 8-4, Lincecum was out of the game and there were 3 more innings left for the Rangers. Not exactly an easily winnable game, but not exactly a “blowout” at the time Ogando was brought in – this seems again like Dave trying to pile on and shade things when it really isn’t necessary.

    I’m not defending Washington – he’s made plenty of poor decisions in these playoffs, but there is no need for Dave to shade things to get that point of view across and hammer it home.

    Comment by Joe — October 29, 2010 @ 1:09 pm

  43. Wally – while those are big FIP differences… you need to put the #’s in perspective.

    How does that translate into a 1 inning run expectancy? If we use FIP as simply runs scored per nine innings, then a 4.5 vs 3.0 FIP amounts to a delta of 0.17 runs an inning. If you are talking about only getting 1 out the delta is even smaller, just under .06runs. (I’m simplifying this greatly by ignoring handedness, specific wOBA’s and the runners on base)

    Is that really changing the win expectancy of a 2-0 game that much?

    I know you should generally do whatever you can to give your team the best chance to win (without hugely impacting subsequent games), but some folks are painting this as if choosing pitcher A over pitcher B is a huge impact. When you consider they were down 2 runs with only 1 inning left to hit, it’s not really that large an impact.

    Should he have used his best pitcher? Sure. Is it criminal and the massive impact people are portraying it as? Nope.

    Comment by Joe — October 29, 2010 @ 1:22 pm

  44. “I disagree. Cain was OK last night, not lights-out. 2K! He allowed plenty of balls in play. lots of flies to the warning track. He got a bit lucky and had good D behind him.”

    That’s pretty much a Matt Cain game in a nutshell (obviously more then 2 strikeouts normally, but it’s pretty much what he does).

    Comment by Deelron — October 29, 2010 @ 1:23 pm

  45. They pulled O’Day because they didn’t want their sidearming righthander facing a lefthanded hitter. Ignoring the benefit of hindsight, that’s not a bad decision.

    Comment by JH — October 29, 2010 @ 2:02 pm

  46. It’s like Charlie Manuel said after Halladay’s no hitter: “I managed good tonight, huh?”

    Comment by bflaff — October 29, 2010 @ 2:06 pm

  47. Subbing out Burrell for defense is pretty much conventional wisdom and a well-established practice at this point, so I’m not sure how many atta boy points Bochy deserves there.

    It’s not like Washington is putting in batting practice pitchers to face the ’27 Yankees (WHO COULD MAKE SUCH A BONEHAED MOVE ?!1?!) so maybe credit goes to the Giants for just playing out of their minds.

    Comment by bflaff — October 29, 2010 @ 2:17 pm

  48. I think w/ such small sample vs. LHP, you would assume a league-avg platoon split. It’s probably safe to assume nate hits righties better.

    Comment by brendan — October 29, 2010 @ 2:19 pm

  49. At what point does a sample become large enough to go by then? That’s more PAs than any hitter in baseball had this year and you see managers make decisions based on season splits all the time.

    Comment by ElJimador — October 29, 2010 @ 2:52 pm

  50. Joe, but this wasn’t just a bases empty get 3 outs situation. Any difference in FIP is going to magnified in a bases loaded situation as compared to your average situation, regardless of the numbers of out in each. A single is worth 2 real runs in at that point in the game, and basically 5% WE, a walk was worth ~2.5% WE.

    In that situation we already had some information that Holland was not is usual self, so we should expect his walk rate on the next batter to be higher than his season average of 3.77/9IP. We don’t really know much about Feliz at this point, so lets just assume he’s his 2010 average at 2.34bb/9, so you cut the likelyhood of a walk in half. Why not take that while you can, especially given the very minimal effect this is likely to have on Feliz’s ability to pitch in future games this season? It might be a somewhat small effect, but pretty much all decisions like this are <.3% WE at least, maybe more like .1%. You need to keep this in perspective. The effects look small, but the relative differences are still large.

    "Sure. Is it criminal and the massive impact people are portraying it as? Nope."

    I'm not sure who's using this kind of hyperbole except you. This was obviously not a likely to change the outcome of the game, but it could have, and I see no rational reason it would effect the future WE of any of the remaining games. If Feliz is needed for all three of games 3-5, pitching yesterday probably doesn't change the fact that he very unlikely to go for game 5. So while the LI was not that high in this situation, it was likely this would be the highest leverage point in the remainder of the game, and with an off day tomorrow, there was no reason to not put your best pitcher out there and give your team the best chance of winning today.

    One could probably even argue that with 8 days off, Feliz's performance in games 3-5 might actually benefit from some work in game 2.

    So, yeah, this isn't a HUGE deal, but it still made no sense. And really, what other baseball do we have to talk about?

    Comment by Wally — October 29, 2010 @ 2:58 pm

  51. I guess I should also say, that I don’t see how pitching in game 2, with an off day today, would effect Feliz pitching in any two of game 3-5, either. He’d likely be fine going back to back in games 3 and 4 regardless of pitching in game 2. Maybe you could make the argument that if he’s needed for 2 IP in both games 3 and 4, he’d be more likely to pitch well without throwing in game 2, but I don’t know that you could support that. Plus, the more you increase the example work load, the less likely that work load is actually needed. And in game 2, there is a pretty good chance he comes in, throws like 5 pitches and gets the out. Don’t think you can argue that out come, which would happen at least 60% of the time, would effect him for games 3-5 at all.

    Comment by Wally — October 29, 2010 @ 3:07 pm

  52. With ~150 PA’s against LHP, I think you’d assume league average + about 1/3 of the observed average. Which would still be a lot, about .100 OPS total.

    Comment by Wally — October 29, 2010 @ 3:12 pm

  53. ElJimador,

    I believe the book says 600 PAs is the half way point, were you regress splits to the league average by 50%.

    And don’t know what kind of regression I was doing before, but its about a .070 split difference, if I recall the OPS league average split right.

    Comment by Wally — October 29, 2010 @ 3:16 pm

  54. A walk might shave 2.5% off the win expectancy but that doesn’t make the difference between Feliz and reliever “X” 2.5% as you mention (it kind of matters who the batter is, not just the pitcher walk rates).

    PS – Quint said in this very comment tree “In a WS game that was criminal” and I was responding to that. If you’re not sure who’s using the hyperbole but me trying reading the comments instead of attacking the person trying to bring some perspective to what is a very minor impact/deal.

    And you can’t compare effects that are close to in the noise and say they are relatively large effects compared to each other. That’s like a company that turns a 1million profit on 10Bil in revenue boldly claiming there profits are up 100% because they had made just 0.5mil the previous quarter… sure the relative jump is huge, but it really doesn’t matter. If you have the opportunity to double your win expectancy from 0.00001 to 0.00002 on a relative scale that is huge… in the real world it doesn’t really matter.

    This scenario there is a difference between 2-0 and 3-0 with one inning to play but the difference between Feliz and another pitcher getting one out is probably far less than 1 expected run (even with the bases loaded) so in reality you end up looking at 2-0score vs a 2.x-0 score.

    Comment by Joe — October 29, 2010 @ 3:35 pm

  55. Joe, maybe I wasn’t fully clear, but I never said that full 2.5% WE from a walk goes away with Feliz. Further I pretty explicitly said it goes down by roughly half comparing Holland to Feliz (which if you can keep up does NOT mean it goes to 1.25%). So, if you think there might be 1/4 chance of a walk with Holland right now, and that there is a 1/8 chance of a walk with Feliz, then WE difference is about 1/8th of walk in that situation, or .3% WE. I don’t see that as so trivial to ignore when you figure its just one PA in a game we’re talking about, and its only one possible out come, as you’d have to add up the marginal WE drop for the hit types, even wild pitches, errors from BIP, and what not.

    PS, Joe. Saying you’re using hyperbole is not attacking you. Its telling you what you’re doing. If I was mistaken about the “criminal” statement, that hardly deserves your own personal attack suggesting I haven’t read the comments. It is also possible I read it and forgot you know….So try not to make yourself out to be a hypocrite in one sentence next time, mmm K? Also, you’re only furthering the use of a hyperbole on quint’s part. In case you don’t know, regardless of who started it, attacking my argument by saying it isn’t criminal is still hyperbole, since it can both not be criminal but still be a bad choice.

    “And you can’t compare effects that are close to in the noise and say they are relatively large effects compared to each other.”

    What? How is it “in the noise?” Can you quantify that for me please? I didn’t think so. Stop making stuff up.

    “If you have the opportunity to double your win expectancy from 0.00001 to 0.00002 on a relative scale that is huge… in the real world it doesn’t really matter.”

    I’m sorry, you were getting worked up about being told you were using hyperbole? ….Just. Stop.

    “This scenario there is a difference between 2-0 and 3-0 with one inning to play but the difference between Feliz and another pitcher getting one out is probably far less than 1 expected run (even with the bases loaded) so in reality you end up looking at 2-0score vs a 2.x-0 score.”

    Probably? You know you have tools to try to estimate this right? We have rates for BB, 1B, 2B, 3B, HR, K with each pitcher, wOBAs against, even left/right splits to try to answer this question. I’ve given reasonable examples to attempt to quantify my case, all you can do is come up with some vague “probably…2.x” argument and some silly hyperbolic unrelated example in an attempt to prove what? While simultaneously ignoring the lack of negative consequences from gaining this advantage, how ever small you can’t quantify it to be?

    Sorry Joe, if you want to convince me, you’re going to have to do a lot better than this.

    Comment by Wally — October 29, 2010 @ 4:06 pm

  56. Wally… I’m not the one claiming it was a significant difference. If someone is making a claim that not using Feliz was criminal (not my words and not my hyperbole despite how you want to depict it) or that it was significant or that it was a boneheaded move… it is up for that person to back up their OPINION, it’s not up to me to disprove it.

    The loss probability difference of Feliz in that situation vs a league average reliever or even a below average reliever is minimal especially when you consider they were starting at a 95% loss expectancy if things worked out perfectly. You see what I did there? By putting it in terms of loss expectancy suddenly the relative change in loss expectancy is pretty minor (which is why you don’t simply look at relative difference in the real world.

    The only people getting worked up are the idiots who think this had a major impact on the outcome of the game (and I don’t think you are part of this crew). As I said before – did it have an impact? Sure. Is it criminal or as large as many are making it out to be? No.

    Comment by Joe — October 29, 2010 @ 5:03 pm

  57. Joe – i think using Lowe and Holland instead of Feliz had 0 effect on the game.

    That said, this isn’t a WP question in my mind, its simple risk v reward.

    You have to as the Rangers finish the 8th inning quickly, its a 2 run deficit, you need to give yourself a chance to tie the game in the 9th. Otherwise you are down 2-0. So that was the reward, a small chance to win a WS game. What was the risk? Using a relief pitcher in a relief situation for 1 out, on a guy who hasn’t pitched much the week and who won’t pitch the next day because its a travel day? Where is the risk in using Feliz?

    It doesn’t matter how small the chances of comeback were, you have to give yourself the chance. The Giants only have to win 2 out of the next 5 games now. That is a situation the Rangers should have done everything reasonable to avoid, and they didn’t.

    And – half of this isn’t even about Feliz in my mind, its why would Washington not have his closer and best reliever warm up in the bullpen, but had his game 4 starter down there?

    Comment by quint — October 29, 2010 @ 5:18 pm

  58. Joe, like it or not, if you’re responding to me, and making an argument against mine, and using the fact that its not criminal as some sort of point to prove your case, you’re engaging in hyperbole. Whether you or quint started it, is irrelevant. I’m not going to think it wasn’t a mistake because you say it wasn’t criminal. And you clearly deviated into hyperbole twice in your previous post in completely novel ways.

    Second, regarding the burden of proof. I’ve made an argument supported by what simple calculations I’ve been willing to do and some simple logic. You can accept it or reject it in your own mind however you like, but if you’re going to voice some criticism, you have to do a little better than your “2.0 to 2.x example.” If there is a flaw in my logic or you want to expand on the calculation of the true effect to WE, go ahead. But absent something on that level, I’m not going give any serious consideration to your criticisms.

    “The loss probability difference of Feliz in that situation vs a league average reliever or even a below average reliever is minimal especially when you consider they were starting at a 95% loss expectancy if things worked out perfectly.”

    This is completely subjective and ambiguous. What exactly do you consider “minimal?” Do you have any estimation of the true loss in WE?

    After all the words you’ve written on this subject I have yet to see anything beyond handwaving and hyperbole, and you’ve brought NOTHING up to make me reconsider my basic position that it was a mistake to give up on maybe .5% to even 1% of WE for this game with a lack of any negative consequences going into future games.

    If your only real point was to tell me “it isn’t criminal,” I have to wonder why you even bothered to write that original post to me, as even you’ve noticed I don’t believe it was “criminal?”

    Comment by Wally — October 29, 2010 @ 5:29 pm

  59. At least I’ve got that going for me.

    Comment by Bhaakon — October 29, 2010 @ 5:44 pm

  60. Quint – I agree with you in terms of risk reward. While the difference in raw #’s between pitchers may be “0.X” runs (as some have mentioned I’m too damn lazy to run the #’s), in the real world it’s the difference between a 2-0 game vs a 3 (or more) – 0 game. Assuming you aren’t mortaging the future (which Wash wouldn’t have been), you go with the best chance at keeping it 2-0 (even if is minor).

    My point is that while you make the move (mainly because there’s no reason not to), in the end it really doesn’t matter (on average). In other words the best chance at keeping it 2-0 is not really that much different than the chance of other pitchers in the pen also keeping it 2-0 in that situation.

    I just don’t get the other people arguing win expectancy #’s and thinking it’s significant (and on a side note I question if they even understand how that model works, yet alone the accuracy of it when they are asking me about noise). If people are looking at this from a win expectancy perspective and how Wash could optimally use the pen – then the move to make was to get Feliz in was after Wilson’s lead off walk in the 7th when it was 1-0. If not then, Feliz probably should have just started the 8th.

    In short (or long)… I think the right move was to go to Feliz at any point after CJ Wilson came out (the earlier the better), and since he hadn’t worked in a while that was probably as much of a justification as any. However people’s view of the 2 out bases loaded situation is perhaps colored by the eventual outcome of the situation as generally speaking the overall game outcome is the same >>99% of the time regardless of who came in to pitch in that spot.
    (note I’m saying the outcome would have been the same, not that Texas would have lost the game >>99% of the time in that situation)

    Comment by Joe — October 29, 2010 @ 5:57 pm

  61. O’Day 2010 vs. lefties: .229/.289/.271 70 ABs
    career: .249/.310/.359 350 PAs

    he does strike out a lot less lefties, but they don’t make great contact either. I assume Ron Washington went with your thought of conventional wisdom: righty sidearmer must only be good against righties. O’Day is a great reliever, though. Micro-managing has killed Washington twice.

    Comment by BlackOps — October 29, 2010 @ 7:31 pm

  62. Probably because you’re a little bit dyslexic.

    Comment by Wally — October 29, 2010 @ 7:46 pm

  63. Two things – The reason Nate does well against lefties, at least according to Giants announcer Mike Krukow, is that he hits outside stuff well, and that’s the way a lefty’s pitch runs. Granted, small numbers, but maybe there’s something there.

    Second, the other plus of the defensive switch is Ross moving to left over Burrell, and he made a good running catch (leadoff batter in the 7th?) that Burrell might have made, but not as easily, again, according to the announcers.

    Comment by Fan of the Panda — October 29, 2010 @ 8:52 pm

  64. Wally – like it or not randomly slapping some #’s together, throwing arbitrary correction factors and ignoring key variables is not “analysis” or using simple logic. I’d rather wave my hand then do what you did, but if you are serious about this let’s take a look at what you did:

    You determined (via logic/analysis) that Holland was twice as likely to walk Aubrey Huff as Feliz.
    - Holland’s career walk rate is 40% higher than Feliz
    - Throw out 2009 and now you got it up to 60% higher. Throwing out 2009 means you are using a sample size of 57innings (as opposed to 195innings) to determine a mean walk rate…. why this should be done is a bit questionable at best. To put this in perspective a difference of 2-3 walks all year would mean nearly a 10% shift in his walk rate. You think opponents faced or heck even one umpire who was squeezing him would dramatically change the walk rate you chose to use? Do you have any idea what the error bar on this # would be given the sample size you have chosen?
    - Then you add in the “he’s not himself” correction factor and voila…. walk rate is double Feliz. What exactly was the logic behind bumping the walk rate from 1.6X to exactly 2X? Truthfully you just arbitrarily rounded things to make it double, right?
    - Finally Aubrey Huff apparently has no impact on getting a walk (or at least isn’t a significant factor). We’ll revisit the validity of ignoring the hitter.

    So back to reality:
    1) There’s no reason to throw out over 2/3 of Hollands career innings to get a walk rate. all this is doing is reducing the sample size to such a small # that it has a monster error bar (and no I can’t define monster quantitatively) and can now be influenced by factors that generally are assumed to wash out over larger sample sizes.
    2) Last I checked the “he wasn’t himself” correction factor was not considered a viable SABR stat… Unless he was injured, how can you adjust it at all? What if he wasn’t himself in the regular season when he walked 4 batters against TB in 4.1 innings…. take out that one start where “he wasn’t himself” and his 2010 walk rate drops over 10%.- by similar logic should we go through and adjust his regular season #’s when he wasn’t himself?
    3) Aubrey Huff’s career walk rate is 40% higher against righthanded pitchers. And this is on a large enough sample size (over 5000 total plate apperances in his career, over 1600 against lefties) that I don’t think you can ignore the hitter. Simply put the handedness of Feliz vs Holland with respect to the hitter should come into play and not be ignored (for both walk and hit rates)

    So what do we have:
    - A left handed pitcher who walks batters at a 40% higher career rate than the right handed option Feliz
    - A batter who walks at an ~40% higher rate against righties.

    To most people this would be more or less a net wash….apparently I’m to believe instead Holland was twice as likely to walk Huff? Because you did simple calculations based on flawed assumptions and threw in a ‘he’s not himself’ correction factor?

    There is just not enough data to put together a # with any sort of statistical significance… You’re coming out with a 0.5-1% difference (which is somewhat cooked for the reasons given above) and somehow consider that meaningful given the data sets you are using (and choosing to ignore)?

    Comment by Joe — October 29, 2010 @ 11:59 pm

  65. Buy at least 200 to send beautiful T shirt

    And so on, buy more more sent.
    Time :2010.07.25-2010 .09.15

    Comment by dgdfgdfdgdfgfhg — October 30, 2010 @ 6:00 am

  66. Okay I googled and found the formula on some blog. Apparently according to The Book it takes 1000 PAs for a LHH vs. a LHP for the player’s true splits to be weighed at 50% of the league average platoon split of .027 wOBA for LHH. Dispensing with the whole formula and if I’ve run the numbers correctly, Schierholtz’s 155 PAs vs. LHH is too small a sample to assume he really does hit lefties better despite how pronounced his splits have been to date. Bottom line, if you assume Schierholtz’s lifetime .306 wOBA reflects his true talent level then weighing his splits against the league average you should assume him to be a .298 wOBA hitter vs. lefties and a .308 wOBA hitter vs. righties. Turns out he’d need to have near twice the number of PAs vs. LHH (around 280+) at his current splits to assume he really is a better hitter vs. lefties.

    I still think such a small difference begs the question of whether it was worth it to bring in a lefty in this case though. If going by The Book we should assume that Schierholtz is that poor of a hitter vs. righties and lefties then why not just trust the guy who was already on the mound and pitching well?

    Comment by ElJimador — October 30, 2010 @ 2:40 pm

  67. this is a good information,There are somethings you can’t defend against.Throwing out 2009 means you are using a sample size of 57innings (as opposed to 195innings) to determine a mean walk rate…. why this should be done is a bit questionable at best. To put this in perspective a difference of 2-3 walks all year would mean nearly a 10% shift in his walk rate. You think opponents faced or heck even one umpire who was squeezing him would dramatically change the walk rate you chose to use? Do you have any idea what the error bar on this # would be given the sample size you have chosen?

    Comment by Michal — November 1, 2010 @ 7:04 am

Leave a comment

Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current ye@r *

Close this window.

0.131 Powered by WordPress