THE BOOK--Playing The Percentages In Baseball

How about taking a LHP out of the game in favor of a RHP (with similar projections) with Cano coming to the plate?

Tango - on the chart, what is the difference between face him and do not walk?

Tom, this surprises you why?  I mean its a terrible decision, as the loading of the bases showed.  If you don’t trust your best pitcher not to give up a HR, you seriously have a bullpen problem.  And even then you should still not-walk A-rod, but it shows a big problem for the angels here.

When did he take out a LHP for a RHP with Cano coming up?

Managers are just clueless.  There is no hope for them.  They are obsessed with cliches and conventional wisdom and microscopic sample sizes. 

They are obsessed with, “How a pitcher is pitching.”

When Girardi was asked whether he considered taking out Burnett to start the 7th, he said, “We thought he was pitching well.” Of course it is correct to take Burnett out in the 7th.  Don’t managers ever look at how pitchers do the 3rd and 4th time through the order regardless of their pitch count, and wonder why there is such a difference?  Don’t they ever compare the stats of a starter the 4th time through the order versus even a run-of-the-mill reliever and realize the the reliever is almost always the better choice?

Not a game goes by where a manager doesn’t give up some win expectancy.  One of these days, someone is going to tally all of the mistakes that managers make in the course of a season.  IBB’s, like the one to A-Rod, leaving starters in too long, taking them out too early, not putting in a pinch hitter for a pitcher in the early or middle parts of a game, going by batter/pitcher matchups causing them to make a personnel mistake, bad batting orders, going with the hot hand, benching or sending down in the order a cold player, stealing too much, stealing too little, bunting too much, bunting too little, bad hit and runs, not sending runners home enough.  On and on and on.

Can someone tell me the advantage of pinch running for A-Rod in the 9th (and losing him if you tie the game) if he is not going to be stealing?

BTW, Tango, he already did that once this LCS.  This was the second time.

I ran it through my sim.  The sim should be pretty accurate with this scenario.

Fuentes pitching to A-Rod, ANA wins 95.42% of the time in 100,000 games.

With A-Rod on first, they win 93.67% of the time.  So Scioscia gives up around 1.75% WE which is a decent amount, but nothing to write home about.  Then again, any individual move never amounts to more than 1 or 2% at the most.  As moves go, this is a very costly one, again, in comparison to other bad moves that a manager can make.

For example, let’s say that Girardi pinch hits for A-Rod with Jerry Hairston.  How much does that cost the Yankees?  This is just to give you an idea as to how much the worst decision in the world, one that a manager would never make, is worth in WE.

If he does that, the Angels win 97.42% of the time, so that horrendous move only costs 1%, less than the cost of Scioscia walking A-Rod.

So we are supposed to be MORE outraged at Scioscia walking A-Rod then if Girardi pinch hit Hairston for A-Rod!

How about if he pinch hit Freddy Guzman for A-Rod?  The Angels win 97.81% of the time, still a more defensible move than the walk to A-Rod, in terms of the amount of WE given up!

Here’s another one for you.  Let’s say that he takes Fuentes out and puts in Sidney Ponson to pitch to A-Rod. Is that better or worse (for the Angels) than walking A-Rod?

Angles win 93.66% of the time in that scenario, exaclty the same as walking A-Rod.

How’s that for putting that IBB in context?

I’m glad that we think alike MGL. It was a bad move to IBB Rodriguez and put the tying run on-base with 2 outs. Even stranger was Giradi pinch running for Rodiguez. Rodriguez is already a good baserunner and a good basestealer.

I thought Scossia should of brought Oliver in to face Damon after Lackey walked Jeter on 4 straight pitches during the 7th innning. Lackey was visibly upset about the umpire not calling an blatant 3rd strike on the previous hitter, Posada. Scossia should have recognized that his starter was starting to lose his cool. Damon popped-out against Lackey, so not making the move to the bullpen worked out in this instance. Even though the Oliver move backfired against Teixera and A-Rod, I think Oliver was the right pitcher to have in the game at that point.

I was watching some of the post-game on MLB Network. Mitch Williams praised Lackey for “wearing his emotions on his sleeve” and not letting them affect his performance. He must of taken a bathroom break when Jeter was up there.

"I thought Scossia should of brought Oliver in to face Damon after Lackey walked Jeter on 4 straight pitches during the 7th innning.”

That is 100% correct.

And that pitch to Posada was borderline.  It probably should have been called a strike, but lots of umpires call it a ball.  If you watch a lot of games, you will see that with many umpires - any pitch that is within a few inches of the corners of the plate is a flip of the coin.  If you are looking for “consistency” among all or even most of the umpires, you are in for a Quixotic adventure.

For example, in game 5 of the Phillies and Dodgers game last night, Tom Hallion was behind the plate.  This guy is as inconsistent as you can possibly be on borderline pitches.  No chance of guessing which way he is going to go.  He is a terrible home pate umpire. Just terrible.

The pitch to Posada was probably a strike but it was definitely not a blatant miss by the umpire.  I would guess that 20% of the time that pitch is called a ball.  Any of the pitch f/x guys can check that.  Last time I nailed it and I think I am pretty close this time as well (although not as confident as last time).

Which pitch was it MGL?

3-2 pitch on Posada in the 7th.  This one:

http://img38.imageshack.us/img38/9383/lackeypoab.jpg

That pitch was .57 feet from the center of the plate horizontally and 1.758 from the ground vertically.  In 2009, there were 55 taken pitches within 3/100 of an inch to those coordinates to a left handed hitter, and they were called strikes 16% of the time.  You win again

Actually, you said ball, not strike.  Turns out you were way off on this one.

Right, I expected that to be called a strike 80% of the time.

The pitch was more than an inch and a half from the corner of the plate, yet it was called a strike only 16% of the time?

And most people were saying that it was definitely a strike?

And there were only 55 taken pitches in that location?

Something is not jiving.

How about pitches within like 1/2 of an inch of that location? 

3/100 of an inch is an awfully small area and thus the number of pitches is small. With 55 pitches, we’d have to say 16% plus or minus 13% at 2 sigma.  Of course I’d still be way off.

I was trying to give the benefit of the doubt to everyone who was saying that that was a bad call by the home plate umpire.  I mean Lackey was acting and speaking like it was obvious that it was strike 3.

Are we talking about the same pitch?  #8, right?

Right, #8.  Here is the data for that game, courtesy of Brooks Baseball.

http://spreadsheets.google.com/ccc?key=tJSyo4paUhxWSMyJD0SPnCQ

Scroll down to row 98 and you’ll see the pitch to Posada.

Setting the range to a half an inch in all directions, gives me 94 takes from lefties this year and 19 called strikes.  So 20%. 

When I think about it, that does seem like a very low amount of takes for a rather wide range of location.  Does anybody else want to double check?

MGL/3 -

After Oliver gave up the game-tying single to Matsui in the 7th, he removed Oliver for Jepsen with Cano due up (who promptly doubled to give the Yanks the lead).

I wish we had the TV graphic showing the pitch location. Is the strike-zone that Fox shows shows on replays bigger than the actual strike-zone of most umpires? Is it bigger than the strike-zone of pitch FX? On the TV graphic, the pitch to Posada clearly looked like a strike.

Culbreth called that pitch a strike to RHB about 60% of the time in 2009.

What about lefties?  Posada is a lefty.

Also, Mike, can you can run this query? 

http://www.codepaste.net/6qc419

It’s giving me wonky results, but I can’t find anything wrong with it.

Oops, scratch that.  I missed which side of the plate Posada was batting from.  Culbreth called that pitch a strike about 45% of the time to LHB in 2009.

Terpsfan, that pitch was clearly a rulebook strike.  But it was near the corner, which is a hard place for umpires to call correctly, so sometimes you get a strike call there and sometimes you don’t.

I believe the strike zone shown on Fox is the rulebook strike zone, although I’m not positive about that, particularly about whether they make any allowance for one radius of the ball along the edges like they should.

Nick, there’s nothing wrong with your query.  You are getting an accurate answer: 19 strikes and 76 balls.  If you expand to include 2007-2008, you get 51 strikes and 160 balls, or 24% strikes.

What my answer was predicated on was the bottom of the strike zone for Posada on that pitch as set by the PITCHf/x operator.  It was set at 1.871 feet. 

It turns out that the average sz_bot for Posada as a LHB is 1.58 feet, which is also the average for all LHB.  So I don’t know what to make of that.  If we set the strike zone bottom at 1.58 feet, Culbreth calls about 20% strikes on pitches like the one Posada saw.

Sometimes I really wish we had the ability to edit our posts here.  And I apparently shouldn’t try to run numbers this early in the morning.  I got half of everything backwards.

If you take sz_bot for Posada at 1.87 feet, Culbreth calls that pitch a strike about 10% of the time.  (The pitch was at 1.76 feet.) If you take sz_bot for Posada at his average of 1.58 feet, Culbreth calls that pitch a strike about 35% of the time, and the average umpire calls it a strike about 25% of the time.

I think that’s right, but ask me in a couple hours when I’m awake.

Thanks Mike. 

It turns out that the average sz_bot for Posada as a LHB is 1.58 feet, which is also the average for all LHB.  So I don’t know what to make of that.

I think that is fairly common.  The sz_top and bot are input by the stringers, so they obviously won’t be very consistent.  For Posada this year, the standard deviation on the bottom end of the strike zone is .14 feet, so the one that was input into that at bat is in range, although barely.

The sz_top and sz_bot are input by the PITCHf/x operator, not by the MLBAM stringer.  The Pfx operator is lining up a line with the hollow of the batter’s knee and another line with the batter’s belt, supposedly at the point in time specified by the rulebook:

The STRIKE ZONE is that area over home plate the upper limit of which is a horizontal line at the midpoint between the top of the shoulders and the top of the uniform pants, and the lower level is a line at the hollow beneath the knee cap. The Strike Zone shall be determined from the batter’s stance as the batter is prepared to swing at a pitched ball.

http://www.insidethebook.com/ee/index.php/site/comments/if_anyone_else_besides_me_has_thought_that_the_k_zones_on_tv_were_screwy_re/#38

In theory, the operator should be capturing real differences in a batter’s stance from pitch to pitch.  In practice, I think it’s about 20% real differences swamped with about 80% noise.  So I usually average a batter’s sz_bot and sz_top across the season for strike zone analysis.  There are probably times when checking his actual stance might be important.

"That pitch was .57 feet from the center of the plate horizontally and 1.758 from the ground vertically.”

The plate being 17 inches wide, anything less than .71 from center is over the plate.  1.76 inches from the ground is a rulebook strike to me, it comes to the center of my knee.  And I’m taller than Posada (6’3").  I’m really shocked that such a pitch is so rarely called a strike.

I can’t argue with the data showing it’s a ball 80% of the time, but I can understand Lackey, throwing that pitch into the what the strikezone is defined as, being upset about it.

Was the IBB in the 9th worse than the IBB to A-Rod in the 7th that put the go-ahead run on base?

"So we are supposed to be MORE outraged at Scioscia walking A-Rod then if Girardi pinch hit Hairston for A-Rod!”

This is not true according to the numbers you posted.

IBB to A-Rod: 95.42-93.67 = 1.75

PH Hairston: 95.42-97.42 = 2.00

Unless you made a typo.

“If you don’t trust your best pitcher not to give up a HR, you seriously have a bullpen problem.”

That’s pretty much it: The Angels have a bullpen problem.  If the Angels have any leads to protect in game 6 and Crom willing, game 7, I’d like to see John Lackey out there to close them.

What was Mike Scioscia thinking?

Tom, I think you’ve provided the answer before: he doesn’t want to face the criticism in the unlikely event that A-Rod hits a HR.  As you’ve said, a manager losing a game is almost preferable to being “second guessed.” Had A-Rod hit his HR then the media storm would’ve been unlike any other.  This way, had the Angels gone on to lose, Mike may have been second-guessed for bringing in Fuentes, but not for losing to Matsui or Cano.  “Their best guy didn’t beat us,” he’d say, and all the reporters would nod.

I’m also sure that he really believes his team’s in a worse position, 2 outs no one on and facing A-Rod, than having A-Rod on first.  He’s not consulting WE charts or running simulations.  The conclusions gained from them are often so counterintuitive that it’s impossible for many people to take them seriously.

I find it hard to believe Scioscia doesn’t want to face criticism.  If that were the case, he would not give Napoli and Mathis almost equal time, or pinch hit GMJr, which Angels fans have criticized all year.

He may be wrong, but I think he’s doing this because he thinks the team’s chances of winning are better not facing A-Rod there.  And I think he’s basing this on tow things: The Game 2 homer, and the way A-Rod has killed Angel pitching all series.

Agree with Rally here:  Scioscia must believe the IBB makes a win more likely.  Even if a manager fears criticism at some level, the safest way to avoid criticism is always to win the game. In addition to believing A-Rod has a “hot hand,” I suspect some managers actually think it’s a mistake to let the other team’s best hitter beat them.  And in a very crude way, that is the hitter most likely to hurt you, so other things equal you’d rather face someone else.  But of course other things are NOT equal, because you have to put a runner on 1B to avoid A-Rod, and that is what managers fail to give proper weight.

In fact, if you employ proper strategy, the other team’s best hitter will in fact “beat you” more than any other hitter.  The only way to prevent that, in general, is to increase even more the chance that someone else will beat you.  But that doesn’t seem to be well understood.

“If you don’t trust your best pitcher not to give up a HR, you seriously have a bullpen problem.”
But a weak bullpen is an argument against the IBB, not for it.  It means you have a real risk of A-Rod scoring, and even worse, the go-ahead run too.  The key point is that an A-Rod HR only ties the game here.  So you have to pitch to him, no matter how hot you think he is.

With no runners on base, why would it be so bad for Fuentes to at least throw pitches as far from ARod as possible?

Suppose for example that he managed to get ARod to chase the first pitch, and now ARod is 0-1.  Does the IBB make sense there (if it had somehow made sense at 0-0)?  Now does Fuentes go after him?  If not, and if ARod chased to 0-2, now does Fuentes go after him?

Similarly, if it was a close call as to whether to walk him, and Fuentes goes 2-0 on ARod, then it would be an easy call to then make it IBB.

I have no doubt that Scioscia thought of ARod as even better than he normally is, probably Bonds-esque, if not even better.  He put him at that level.  Once Scioscia did that, put him at that level, then of course it made sense to IBB him.

It’s not so much that he didn’t understand the win probability chart.  It’s that he didn’t understand ARod’s true talent level to begin with.

MGL: can you rerun your sim, and keep all of ARod’s forecast the same, but bump up his HR total.  How many HR (per 500 PA or whatever) do you have to give ARod in order for the IBB to be breakeven at that point?  THAT is what Scioscia would have to have implicitly believed.

It’s safe to assume that Scioscia elected to face Matsui because he feared the game-tying HR to Alex Rodriguez. But by walking him, he’s opening himself up for the more likely game-tying extra-base hit.

Over his career, Alex Rodriguez has hit a HR in 6.1 percent of his plate appearances net IBB. Matsui has hit an XBH in 9.1 percent of his plate appearances.

The only explanation for Girardi pinch-running Guzman is the marginal benefit of plating on a Matsui double since Guzman didn’t even attempt a steal.  Maybe Guzman doesn’t score 9.1% of the time but it’s pretty close to that and certainly more likely than a A-Rod homer.

"It’s not so much that he didn’t understand the win probability chart.  It’s that he didn’t understand ARod’s true talent level to begin with.”

We’re speculating of course, but I do think it’s likely some of both.  With 2 outs, he’s mainly worried about the HR.  And he figures A-Rod is more likely to hit one than Matsui (true), and say he inflates A-Rod’s true HR talent by 50%.  At that point, the probability of an A-Rod HR and a Matsui XBH (which also scores tying run with pinchrunner at 1B) are about the same (10.5% of ABs).  But the IBB is still a mistake, of course, because in the Masui XBH scenario Matsui is in scoring position or has alreay scored the go-ahead run, vs. bases empty if A-Rod hits solo shot. Not to mention possibility that pinchrunner advances to 2B on PB, or BB to Matsui, and a single can tie the game.  So I do think it’s likely Scioscia is focused too narrowly on the HR scenario, which is another way of saying he’s not thinking about WP properly.

oops, Tom beat me to it.....

Considering that borderline pitches often go either way, I wonder why the networks don’t reflect that in their “ptich track” graphics… I mean, it’s mostly blarney anyway, but I think that, given called-strike information based on location:

Start with the outline of the book strike zone; lightly shade the area representing “strikes called at least three-quarters of the time”;
give a smaller shaded border surrounding that, where the ball/strike calls tend to be mixed;
grade the tint (either by color or intensity) to reflect the increased likelihood of not getting the strike call, all the way to “just about never called a strike.”

I think this would give fans a more complete picture of what’s going on at the plate.  You could see if the last pitch was a call you really should expect to get based on the actual previous data - no use wailing about a textbook strike that’s actually ruled a ball 90% of the time.

As for different kinds of pitches being called differently in the same location… well, I’ll leave that to cleverer people.  I don’t know that I trust automatic pitch identification well enough to go with a color-coded dot or something on the little chart; besides which, there comes a point where the graphic would compete with the actual game, rather than enhance our understanding of what we’re seeing.  I’d save such a chart for replays, not have it live in the corner the way TBS is doing.  It’s ridiculous how crowded the screen is getting.  Watch a “classic” rerun of a game from the 80’s or something and it’s like landing on a different planet - you mean, nothing but the GAME on the game broadcast?  Madness!

Excellent way to put it guys: Arod has a 10% chance of hitting a HR, and Matsui has a 10% chance of hitting an extra base hit.  And in both cases, it ties the game.

In the Matsui case, it also puts a runner at 2b, 3b, or he-himself having hit a HR to take the lead.

On the other hand, by walking ARod you not only take away the chance he hits a HR, you also take away the chance he hits a double or triple (raising the % of times the game is tied on a Matsui single, which is more likely than a Matsui xBH).

You have have to put this into your equation as well, though perhaps it turns out to be negligible.

Josh: if you work it all out, you get exactly what MGL’s sim says you should get.  That would be the point of the sim.

How are the CAPTCHAs generated on this site (ie, the ones that appear when I am previewing my post before submitting it)? I have noticed that disproportionate amount of times I get a word followed by the number “69.” Is this indicative of a predilection for this website’s authors?

In the end, I agree here that the IBB to ARod in that spot is indefensible (unless Scioscia knew Girardi would do something stupid like take Arod out of the game, for a PR who does not steal, in which case, one wonders!)

On the other hand, “going after” ARod in that spot, with Fuentes on the mound, and Matsui on deck, seems a dubious strategy too. It is hard to ignore what ARod has been doing this series, whatever our doubts on “clutch” or “hothandedness” as a repeatable skill.  Fuentes does strongly depress slugging and HR vs. lefties, especially in 09 fwiw.

On this basis, I would have thought the correct approach was “pitching around” ARod as implied or stated by Tango. But if pitching around him is the advisable course of action, can we fault him that much just for putting him on? Note that Fuentes was so damn wild, that Scioscia would legitimately have concerns that Fuentes might have laid one in thigh high in the course of pitching around ARod(but this goes to the larger point: if you cant trust the pitcher there to be able to pitch around someone, why is he in the game?)

Contra walking Arod, it is surely of some relevance that Matsui had a strong 2009 reverse platoon split and neither Arod or Matsui over their career seem to care too much which hand the pitcher is throwing with. 

I think the bigger tactical error here is having such an execrable pitcher as Fuentes on the mound with the game on the line. He didn’t look like a big leaguer out there.

Another one which had me scratching my head more than the IBB: Figgins facing Rivera and not trying to bunt Willits home in the bottom of the 8th with one out. Surely Chone bunts well enough and hits fly balls infreqeuntly enough (esp. v. Rivera) that this was called for under WE principles?

When a Proven Closer™ looks like crap but still gets the job done in the end, it’s simply proof of his Proven Closer™ status.  This is why closers are so hard to dislodge these days; the cloud of mythology around them is thicker than that around any other MLB type…

The platoon situation definitely plays into the equation.  A-Rod is going to be better versus Fuentes because he is a LHP and Matsui is going to be a lot worse.  So the numbers quoted above (for HR and extra base hit rates) are not valid.  They have to be adjusted for each players true platoon differential.

But, the sim takes all of that into consideration.

BTW, when we talk about a manager being averse to criticism or risk averse, or any of those things, they are on a subconscious level and they influence his assessment of the situation.  Of course he thinks that his team is better off walking A-Rod.  It is just that that assessment (that his team is better off) is subconsciously influenced by the severe criticism he would receive if A-Rod hit another HR, as well as the second guessing he would do to himself.  That is the way those things work. 

Girardi not stealing Guzman after he pinch ran for A-Rod is the same thing.  He truly thinks that the steal attempt is not the right move there, but that assessment is influenced by the fact that he (all managers) is afraid to end the game on a CS.  Just like in life, managers’ decisions are bad because they are influenced and dictated by all kinds of common psychological forces that are part of human nature.  The only way to get around that such that managers make optimal decisions is to educate them on what the optimal decisions are.  The only reasons managers make bad decisions is because they have no idea what the right ones are in the first place.  So they have to use their guts and intuition.  And those things are influenced by the things we are talking about (being risk-averse, not wanting to get criticized, etc.).

MGL: can you rerun your sim, and keep all of ARod’s forecast the same, but bump up his HR total.  How many HR (per 500 PA or whatever) do you have to give ARod in order for the IBB to be breakeven at that point?  THAT is what Scioscia would have to have implicitly believed.

Good question. 

Remember that with a “normal” (current projection) A-Rod batting, the Angels win 95.53% of the time (I re-ran the sim a million times).

A normal A-Rod hits a HR around once every 16.7 non-IBB PA in a neutral context.  That is my projection.

Versus Fuentes in that park on a normal day, it is once every 20 PA.

If we bump that by 50%, such that he hits a HR every 13 PA against Fuentes, the Angels win 94.51%, so that still isn’t enough to justify an IBB.

If we double that, so that he hits a HR every 10 PA, the Angels win 93.48% of the time.

So basically, we have to double A-Rod’s HR rate for the IBB to be justified.

BTW, against Fuentes, Matsui hits a HR once every 33 PA, a double once every 25 PA and a triple once every 467 PA.

Again, these are all using current projections for players, including their projected platoon differentials.

So let me get this straight - the chart says not to walk Bonds with the bases loaded in the 9th inning of a tie game?  Chart fail.

"So basically, we have to double A-Rod’s HR rate for the IBB to be justified. “

What MGL is saying here is that you have to give ARod 1 extra HR (and 1 less out) per 20 PA, in order for the IBB to be breakeven.  That means adding .05 HR per PA.  In wOBA parlance, that means adding .100 per PA.

If MGL has ARod as say a .400 wOBA hitter, then Scioscia would have to believe that ARod if a .500 wOBA hitter (at that moment in time) for the IBB to be breakeven.  (Or if MGL has him as .380, then he has to be a .480, etc.)

Now, when he give margins for error, I normally just give 20 points of wOBA.  Basically, that’s about the platoon advantage.  Maybe, you can give out 50 points of wOBA if you REALLY REALLY think you know something that everyone else doesn’t.  Giving out 75 points of wOBA would be truly generous, and giving out 100 points of wOBA as a margin of error would be obscenely high.

The difference between Molina and Posada is 100 points of wOBA.  Can you imagine a scenario where you think that Molina’s context is such that you think he could hit as well as Posada?

So, this is what Scioscia is believing in going for the IBB here.

It amuses (and confuses) me that otherwise not stupid people think that they can “reason this out” and come up with the right answer.  That is why managers get these things wrong. Because they can’t be “reasoned out”.  Someone has to crunch the numbers as much as some people think that it can’t be done.  As I said, this one is pretty straightforward.  There really aren’t any intangibles or things we don’t know or can’t model.  Plus when the outcome is not even close (one alternative or the other is THE clearly better choice), you don’t have to worry about the uncertainty of the data that goes into the model or the robustness of the model itself. I have always said this (and it is an important concept). If a crude but correct analysis yields a clear cut answer, where one alternative is clearly better than another, then that is all you need to do.  If it yields a close decision, then you need to do a better analysis. If a very good analysis still yields a close decision, then you can flip a coin and move on to something else.

MGL’s rule # 4.817:

The difference between a smart person and an ignorant one is not that the smart person always knows the right answer.  It is that the smart person knows what it is he doesn’t know and the stupid one thinks he knows everything or at least a lot more than he really does.

Tango and I only know the answer to this question because we did the requisite research to figure it out (with some degree of certainty of course - IOW, there is finite chance that we are wrong).  NO ONE can figure it out without doing some numerically based research.  Opinions don’t matter. This is a factual issue.  Whether you like chocolate or vanilla ice cream is a matter of opinion.  Baseball strategy questions are NEVER a matter of opinion.  That is not to say of course that we know the answer to all those questions with 100% certainty in all situations.  We don’t.

It’s an interesting question, but I think we’re missing the bigger picture, which is what was Sciosca thinking (or maybe better, how dumb was it?) to lift Weaver for Fuentes with Rodriguez due to leadoff.  He could have left Weaver in for AT LEAST 1 more batter, then brought in Fuentes to face the lefty hitting Matsui (or even leave Weaver in to face Matsui).

And Uhg (stupid ME) I meant leave Weaver in UNTIL he at least gets past Rodriguez in the lineup, as obviously he didn’t leave off.

without looking anything up im gonna guess that fuentes is a better pitcher than weaver even against a rhb. of course that entirely depends on how much you want to bump weaver for pitching in relief. if u want to give him a full run per 9 or even close to that then he is definitely the better pitcher.

If I put Weaver in the sim versus A-Rod and give him about a .7 run bump for pitching in relief, the Angels win 94.9% of the time. Again, with Fuentes, it is around 95.5%.

Despite all the nonsense you hear, Fuentes is a really, really good pitcher. He is an elite closer.  Not on the level of Rivera of course, but he is the next level down.

Now, if Weaver faces A-Rod and then Fuentes comes in if A-Rod does not make out, then the Angels win 95.04% of the time.

So Fuentes is still the best option versus A-Rod.  Again, all the nonsense you hear about Fuentes is just that - nonsense.  Small sample nonsense. The same nonsense you hear about how Hammels (for the Phillies) is no good, and A-Rod is the best hitter of all time, and the Yankees bullpen is bad other than Rivera and Swisher is terrible.  All nonsense.

This small sample emphasis is magnified 100 times during the post-season.  That’s all you hear on sports talk radio, the MLB network, and ESPN.  All talk about “how players are doing in the post season” by all the idiot talking heads.  Mitch Williams wants to take Hammels out of the rotation.  Lots of people want Gardner in RF rather than Swisher.  All the talking heads are saying how unreliable Hughes and Joba are.  All nonsense.  Complete nonsense and B.S.

Did I say it was all nonsense?

"If I put Weaver in the sim versus A-Rod and give him about a .7 run bump for pitching in relief, the Angels win 94.9% of the time. Again, with Fuentes, it is around 95.5%.

Now, if Weaver faces A-Rod and then Fuentes comes in if A-Rod does not make out, then the Angels win 95.04% of the time.”

I was mentally using Tom’s 1.0 run bump for Weaver, and I was not at all doubting Fuentes is very good, but just that the starting RH pitcher Weaver who is already in the game should go ahead and continue at least thru Rodriguez batting before you would bring in Fuentes vs. Matsui, but looks like it’s roughly a coin flip, depending on how much better Weaver is as a reliever vs. starter, so I was wrong.

Yes, with a 1 run bump, I think it is roughly a coin flip.  And of course almost all managers will go with their closer in the 9th regardless of the handedness of the batters first coming up.  And that is fine if your closer is a lot better than any other pitcher you have available and/or he doesn’t have much of a platoon split himself.

Basically it is not that hard to estimate the true talent of every pitcher versus RH and LH batters.  Of course managers have no idea how to do that.  Just another example of someone wanting to construct the best house possible but all they have to do so is a screwdriver and a hacksaw.

When you say Fuentes is an elite closer, have you updated a projection including 2009 or basing it on preseason?  Because his strikeout rate dropped by 40% and he allowed a godawful number of baserunners.

I haven’t updated a projection, but by eyeballing it, seems like a marcel-type projection would have him maybe a bit above average for a reliever, but nothing more than that.

If I did the Marcel right, I’ve got Fuentes at a 3.62 ERA and 3.85 FIP.  MLB average for a reliever is 4.08.  Above average, but I think he’s a long way from elite.

Yes, I still have him as an elite closer including this year’s numbers.  With an average reliever at around 3.90 (and an average closer at around 3.30), I have Fuentes currently at 2.71.  Rivera is 2.14 and Hughes is 2.99 as a comparison.

No question that this year his numbers are much worse than in previous years.  His K rate is way down, his HR rate is up a little, and his BB rate is around at career levels.

His fastball is the slowest it has been since 2004, so maybe he is losing something - I don’t know.  He still looks like he has pretty good stuff for a closer to me, although it is a little hard to tell when watching a lefty to be honest, especially one who doesn’t have great command of the strike zone like Fuentes.

Even though he has had a pretty bad season, it is only 55 innings, the equivalent of 3 months for a starter and the equivalent of 6 full games!  That is not a whole lot to change a projection.  Hence one of the reasons why we have these love/hate relationships with short relievers.  They pitch so few innings that even if their true talent never changed, their performance per week or month, or even in an entire season can be quite volatile.

That is why despite the fact my research found that recent performance for pitchers has some predictive value (although I suspect that is mostly for health reasons), I pay no attention whatsoever to recent, short-term performance for any players, batters or pitchers.  I never want to get in the same trap as most people, which is to equate true talent and expected future performance with recent short-term performance (of course they take it to a ridiculous level).  That is almost a disease with commentators and fans during the post-season, as if the regular season never took place, let alone seasons before that.

I’m a little shocked Fuentes could have a projection of 2.71.  That’s better than his career best season.

Maybe out scales are different.  My projections above assume a typical recent season with an overall ERA around 4.50 and reliever ERA about 4.10.

MGL typically sets his forecast levels to a league ERA of 4.00.  Basically, he’s saying Fuentes is at 70% league average, which I agree seems hard to accept.

yes, my scale is always a league average of 4.00. You have to remember that Fuentes pitched for the Rockies for most of the last 5 years.

Here are his component ERA’s scaled to an average of 4.00, for the last 5 years, after park and opponent adjusting:

05 2.32
06 2.22
07 2.71
08 1.01
09 4.16

If we do a weighted average where each year is weighed 50% more than the previous year, we get 2.75.

Obviously my projection includes regression, but I also don’t weight recent years quite so heavily for pitchers, and I am doing everything by component.

Other than the fact that he may have gotten considerably worse this year (do we have much evidence for that other than the numbers, which the projection algorithm already takes into consideration - as well, the projection algorithm already takes into consideration to some extent that he may have gotten worse, true talent wise - that is why we do the weighting), a 2.71 projection (out of 4.00) is a reasonable number for a very good closer.  Your absolute best closers allow close to 2 runs less per 9 than the average pitcher.  An elite but not the best allows almost 1.5 runs less.  An average closer allows close to 1 run less and a very mediocre closer (should not be closing actually) allows a half run less per 9 than the average pitcher.

I am simply saying that Fuentes is at the lower end of still being an elite (not the best “level” - the second best “level") closer.

I don’t see how that is so hard to believe.  Mike Scioscia appears to like him a lot.