Lookout Landing

One that would make me both happy and incredibly annoyed at the same time

Graham MacAree - July 14, 2009 reply

I'm going to close this subthread before it devolves into annoying speculation

Let’s just leave it at “Hardy could end up a Mariner, but he will not do so at a low price.”

No.

Graham MacAree - July 14, 2009 reply

Before people misconstrue what I'm insinuating

I don’t think that Vargas will be Washburn’s equal for the rest of the year, because his command is a little worse, and he’s more likely to get fatigued after not throwing at all in 2008. The point of this is that, hey, Washburn hasn’t been that great himself, and RRS may very likely be able to replace him if he’s dealt.

tRA la la la la...

SethGrandpa - July 14, 2009 reply

I'm still confused as to how Washburn's LD rate is so high

it seems like he hasn’t been hit all that hard. Oh well, thank goodness for advanced stats, I guess

Maybe a bunch of fliners. I don't know

I don’t even think LD% over three months is that meaningful. I just included it because people would ask.

By the way, I think Washburn has a slightly better ability to turn missed bats into strikeouts than Vargas, and also a better ability to avoid walks, but once again, I’m not trying to call them equals; I’m just trying to call them similar.

But it's not as if fliners are particularly good for a pitcher anyway

yeah, they’re not great contact, but that type of hit is pretty likely to turn into a blooper.

I’d say this is a situation where Washburn has more “pitchability” or whatever the hell you want to call it, so yeah, he’s probably a bit better, but he also costs 20x what Vargas costs.

Plus, RRS!

I love that his velocity is getting better. For a few months he was practically a forgotten man.

89-92 on the TAC gun is probably around 88-91 from what I'm told at least.

Slurvey - July 14, 2009 reply

Which is right around where we want him to be

But 88-91 starting is where he's always been

it would be weird if he came up and suddenly was throwing 2 mph harder

Really?

I seem to remember liner after liner getting snared in the OF in NY and against Texas. Maybe Baltimore too…

Bearskin Rugburn - July 14, 2009 reply

Ichi-hawk!

section331 - July 14, 2009 reply

Pardon the stupid

but what’s the difference between tRA and tRA*? And which is better for evaluating a player?

tRA* (or any stat with *, typically) is regressed

Essentially it tries to eliminate luck. From what I understand, tRA* regresses everything, but the one that has the biggest effect is HR/FB rate.

Thanks

that’s been bugging me for a while.

Note that they're both regressed pretty heavily, since they only have a half season's worth of starts

but Vargas’ straight tRA is influenced by a fluky home run rate.

How do you say it in conversation?

tRA asterisk?

Regressed tRA

OlSalty - July 14, 2009 reply

That makes sense.

tRA star?

Jack Moore - July 14, 2009 reply

x

from SC

The regressed version of tRA. tRA* takes each of the pitcher’s major component stats and regresses them back to league average, the magnitude of regression being based on the amount of batters the pitcher faces. tRA* is not a measurement of a pitcher’s results per se, but should be seen as the system’s best estimate of a pitcher’s true talent level based on his stats in any given year at any given level. tRA* does not consider a pitcher’s statistics from other years and leagues.

appleshampoo - July 14, 2009 reply

Also, how large does the sample size need to be before tRA* is useful? Are there varying scales or usefulness or predictive value for varying sizes of sample...size (yeah...)?

(I remember at the beginning of the year that Graham and Matthew told us to ignore tRA* for at least a few months).

Decatur - July 14, 2009 reply

Thanks again

I probably should have checked StatCorner before asking, really. That glossary looks pretty darn helpful.

http://www.hardballtimes.com/main/statpages/glossary/#xfip

The Hardball Times' xFIP stat does a similar regression, I believe (hopefully this is relevant).

Expected Fielding Independent Pitching. This is an experimental stat that adjusts FIP and “normalizes” the home run component. Research has shown that home runs allowed are pretty much a function of flyballs allowed and home park, so xFIP is based on the average number of home runs allowed per outfield fly. Theoretically, this should be a better predicter of a pitcher’s future ERA.

Decatur - July 14, 2009 reply

I haven't looked into FIP yet.

I’m trying to get to grips with tRA as best I can as it seems to be the preferred stat around here.

It is definitely better

FIP considers all contact equal, even though we know that a pitcher that gives up more line drives sucks more. tRA makes up for that – it’s why Carlos Silva’s pitching effectiveness was in question – some referred to him as an average pitcher with bad defense, but tRA showed he gave up line drives like a pitching machine, and that implied he was just as bad as we thought he was. But it’s a good idea to learn FIP because you will currently see that in more places, like Fangraphs.

For a quick and dirty, both tRA and FIP roughly scale to ERA

so anything under 4 is good, 4-4.5 is mediocre and 5 and up is bad.

Wait, I thought tRA was scaled to TOTAL runs against, while FIP was scaled to EARNED runs agains.

Decatur - July 14, 2009 reply

Yes

Matthew - July 14, 2009 reply

What is the benefit of scaling unearned runs?

I thought the point of tRA was to eliminate things that the pitcher can’t control and errors would seem to be the top of that list.

Errors are subjective, and errors are included as a natural part of defense.

wOBA considers reaching on an error a positive value for the hitter for a similar reason. All outcomes that are not a walk, strikeout or homerun have a defensive aspect. Why should a player get punished when a fast third baseman drops a ball but rewarded when a slow third baseman can’t get to a ball?

But with regards to tRA, I am not positive that actual runs scored (earned or not) matters as much in the equation, because if I am not mistaken it gives “arbitrary” (sort of) run values that are derived from the overall run values of every event in baseball each year. Not really arbitrary, I guess, but not necessarily related to the actual result of the line drive itself.

So a pitcher's tRA

is only minimally affected by runs allowed and most affected by what types of hits that he gives up (line drives, ground balls, etc…). This is how it shows the difference between a pitcher who gives up a lot of runs because of the bad defence behind him and pitcher is just bad. Is that right?

That is exactly my understanding, yes.

But before you assume that’s right, wait for confirmation by Matthew, Graham or Jeff.

Follow up question for the trio that agrees on everything:

Though tRA may be doing its best to be defense independent, doesn’t defense still affect it slightly due to the number of plate appearances a pitcher experiences due to bad defense?

For example, though a ground ball isn’t worth a whole lot in terms of run value, if 11 ground balls in a row go for hits due to a shitty shortstop, won’t that affect their tRA simply because there have been far more plate appearances for those ball in play opportunities to occur?

This is a good question.

I’d also like to know how tRA* assembled? I understand that it has something to do with projecting the regression of HR/FB rate but what is used to make this sort of prediction? League average? Or do I have the definition of tRA misconstrued?

It regresses everything toward league average year to year (and by league), HR, GB, LD and more.

I do not remember if it addresses career norms, but all of the old fanposts I have say it does not.

Also, ignore tRA during most of the season, as small sample sizes are often regressed too much.

I think a little over half a year’s data was when it starts to be more reliable, and it is best at the end of a full year.

WOAH, sorry. Ignore tRA*.

I swear I had the * in there before.

Also, it is technically over half the season, so you can probably start looking at it now with some reliability.

But if you were looking at these tRA* numbers, say, over a month ago, then they would have been less accurate.

I'm hoping that eventually tRA* would regress from a larger sample size on a weighted scale

this would help it be more useful in the early season, if I’m understanding correctly that the run values of batted balls and HR/FB rates don’t fluctuate much from year to year.

But wouldn't regression toward league norms

defeat the purpose of tRA* as a means of finding determining who was actually good at pitching and who wasn’t? For example, isn’t the fact that Roy Halladay or Aaron Cook have such high GB/FB ratios what makes them good? Isn’t it exactly what makes the above league average?

League averages vary from year to year.

If GB% is up league wide and Halladay’s GB% increases by around the same amount, he’s not pitching any better.

Aaron Campeau - July 15, 2009 reply

That makes more sense.

Thanks everybody for the help. I appreciate your patience with me, I’m still trying to get a grasp of a lot of these terms.

The theoretical answer would be no

because tRA is a rate stat, not a counting one. Even though yeah, he gives up two more grounders because his shortstop was Yuniesky Betancourt, those two grounders contribute both an expected run value as well as an expected out value, leaving the overall ratio, and thus the tRA the same.

The practical answer is probably, since pitchers pitch worse when they are fatigued

and shitty defense causes them to get fatigued faster, as they need to face more batters and throw more pitches

Seperate topic, slightly related. Don't strikeouts have a slight negative value in tRA?

There was a comment about that recently. Is that due to the number of pitches required for a strikeout? Every time I think I have a layman’s grasp of tRA I realize I don’t.

Separate you illiterate billbug

To use two quotes written by other members:

Certain events (strikeout, infield fly) have negative run values

-Essby

tRA is not a dynamic model and therefore doesn’t account for the fact that if you strikeout 100% of the batters you face each K is worth less

- Graham

I do not know if these are the run values that Graham uses (there may be more than one type of calculation), but run values look, in essence, like this:

CapSea - July 15, 2009 reply

So what is it, all these numbers added per 9 innings?

I just realized I was reading the term negative value in the wrong light

You crazy stat geeks and your crazy math, stay out of my yard!

I'm out of your yard, OK?

I’M JUST YELLING FROM THE SIDEWALK NOW

It took me a while to figure out who essby was

HE SHOULD HAVE SPELLED IT IN MATHS

.

Theoretically a HBP isn't a batted ball but...

Home run isn't in there for some reason too.

But I didn’t make the chart and couldn’t find Tango’s data.

CapSea - July 15, 2009 reply

A question on how the run values are calculated

so is the run value of each event the average of [(run_value_start – run_value_end) + runs_scored] for each play?

For example, a ground ball with a runner on second and no out becomes a run-scoring single. The run value of that event would be [(1.189-.953)+1] = 1.236 runs?

And then you just add up the values of each play and divide by the total number of occurrences?

Basically.

Matthew - July 15, 2009 reply

Except it should be rv_end - rv_start + runs

Matthew - July 15, 2009 reply

Ah, got it.

That makes more sense.

It's not affected at all by runs allowed

Each event has an expected run outcome and an expected out … outcome, which can be calculated by looking at league averages and poking them with a big algebra stick repeatedly. Once that’s done there’s no need to look at runs or actual innings pitched.

Graham MacAree - July 15, 2009 reply

Minimally affected =).

Runs scoring on a play does affect average year by year data, at a very very minimal level.

CapSea - July 15, 2009 reply

Not affected to the point anyone notices

Graham MacAree - July 16, 2009 reply

Alright, I think I finally get it.

What about a foul out? That has to have a 0 expected run outcome, except for sacrafice flies, how is that calculated?

http://www.lookoutlanding.com/2009/7/14/949518/a-comparison#18247463

Except with Matthew's caveat that I switched the rv_start and rv_end places

They are generally popups.

Graham MacAree - July 16, 2009 reply

He's so smooth.

msb - July 14, 2009 reply

Since we're talking about a pitcher that has defied his component stats

Is there any early Hit/fx information on if some pitchers are capable of reducing the hit-speed of opposing hitters?

John Morgan - July 14, 2009 reply

I will leave this to Matthew to answer

Way too small a sample.

You’d need thousands of batted balls in your database to do a good W/WY comparison.

Matthew - July 14, 2009 reply

Yeah, and it seems like you'd need to control for pitch velocity too

otherwise you just get the obvious stuff like ‘speed off bat vs. Wakefield is lower than league average!’

marc w - July 15, 2009 reply

So Tim Wakefield - deserving All-Star!!

One thing I'm curious about as far as comparing pitcher to pitcher

Is how those pitchers fare against AL West opponents and possible playoff opponents. Did Vargas get a healthy dose of crappy lineups when he started? Has Washburn dominated the Angels? Is RR-S great on the road?

These are the types of questions I’m going to start looking into based on information already given.

It appears that Vargas' worst 3 starts came at Col, at LAD, and at NYY. And duh, Washburn sucks against the Angels. Im liking Vargas over Washburn more and more.

Generalizing from a small sample size is always a bad idea

Sorry, I tend to reach when looking for reasons to make up my mind.

Be open to not making up your mind.

Matthew - July 14, 2009 reply

Fuck, now Im confused.

Vargas had the flu in NY

LauraBu - July 14, 2009 reply

We can talk about trades all we want

but I can’t see Z being able to do shit with us still in the race and Wash having such a shiny ERA. The common fans would riot.

IceStormV1 - July 14, 2009 reply

Jason Vargas has a kick-ass changeup

A pleasant surprise:

According to Harry Pavlidis (http://www.beyondtheboxscore.com/2009/7/10/944659/expanding-on-harrys-whiff-rate), Jason Vargas has the 4th best changeup whiff rate in MLB. 45.9%, in the 75-149 swing category.

LauraBu - July 14, 2009 reply

I love that pitch so hard

Jeff Sullivan - July 15, 2009 reply

Interestingly enough Fangraphs pitch data hates it until this season.

Aaron Campeau - July 15, 2009 reply

Huh?

Not really.

Jeff Sullivan - July 15, 2009 reply

Am I reading the data wrong?

1.42 in ’05, -1.75 in ’06, -5.69 in ’07.

I’ll admit that looking at the pitch type run values is new to me, but aside from ‘05 and this season it has produced below average results. I still thinks it’s an awesome pitch.

Aaron Campeau - July 15, 2009 reply

The 05 number is the largest sample.

He threw 10 innings in 07, so I’m not sure that figure means anything. The ’06 number is interesting, though.

marc w - July 15, 2009 reply

Changeups in general or Vargas'?

What kind does he throw anyway? I know there’s a few but the only one I’m familiar with in any fashion is the circle change. Due to Moyer, he did a little clip showing the grip and discussing its action and why he liked it, had some slow motion video

Well, both, but I love Vargas' in particular

I don’t know what kind he throws, but it has a lot more tail than his fastball does, so it’s not a straight change.

Jeff Sullivan - July 15, 2009 reply

The general idea behind most changeups is the same - you want to get the ball as far back in your hand as possible, grip-wise

which increases the friction as it leaves your hand and adds additional backspin with the same delivery as a fastball. From my understanding and experience, the advantage to the circle change is that it tails much more to the pitcher’s arm side, which is why it’s pretty popular as well as effective against opposite handed batters

However, it should be noted that it's also a more difficult pitch to throw because the grip is pretty unnatural

I've got a book with Branch Rickey describing the mechanics of a "window shade" or "shutter" change.

I never know when to use quotes properly. The description alone gave me sympathy pains in my elbow

Why do you guys think it's so good (if it is)?

Is it that he doesn’t have any tell, he does exactly the same set of things as when he throws his fastball?
Is it the velocity difference compared to his fastball? (Note his fastball isn’t the super fast kind)
Is it that he throws it infrequently enough to surprise hitters?
Is it that he’s bold enough to throw it in situations that go against the conventional pitch wisdom?

LauraBu - July 15, 2009 reply

The grip/tail idea is interesting

Because that reminds me of this piece by Shannon Drayer (http://www.mynorthwest.com/?sid=177631&nid=374) where she mentions that Taka is explaining Uehara’s pitching to Vargas.

Cool that Jason is open to learning new stuff.

LauraBu - July 15, 2009 reply

My real question is will he be able to sustain these stats?

Or will opposing hitters figure him out?

LauraBu - July 15, 2009 reply

Fangraphs won't load on my computer at work but I suspect it does all of these things.

Most importantly, however, it produces high swingning strike rates. Which is the most telling sign, as well as good control, of a good pitch. I could be wrong here and I would wait for someone better at this to confirm it.

8 mph difference, 19.4% thrown

You’d have to ask someone like Matthew what the pitch is doing, but I suspect that since Vargas doesn’t go all the way over the top anyway, his change probably has some vicious arm-side run which helps it be more effective against opposite-handed batters