THE BOOK--Playing The Percentages In Baseball

According to baseball-reference, Bush has always had a tougher time of it with runners on base http://www.baseball-reference.com/pi/psplit.cgi?n1=bushda01&year=00
compared with the other starters, so perhaps NERC isn’t the best way to assess pitchers like Bush who presumably are less effective from the stretch.

2006 is a perfect example of this.  Great component stats for an NERC of 3.42, but his splits here do him in.  http://www.baseball-reference.com/pi/psplit.cgi?n1=bushda01&year=2006

None of the other Brewer starters have this pattern.

http://www.fangraphs.com/statss.aspx?playerid=4635&position=P

Look at the LOB% number in the second chart.  There should be a direct relationship between LOB% and pitcher quality.  His LOB% in 2008 is very poor, meaning that he pitched poorly with men on base.  But, if you look at this career, it looks about right (69%) for someone of his quality (FIP=4.46).

The first question to ask is whether Bush is among the 5 best available starters right now.  Sheets and Gallardo are in their own league.  Suppan is obviously a gritty gamer who knows how to win, so he’s not getting removed.  That leaves two spots, currently held by Manny Parra and Carlos Villenueva.  Here are their respective ZiPS projected ERA’s for 2008:

Bush: 4.64 ERA
Parra: 4.31 ERA
Villenueva: 4.46 ERA

Both guys have little major league experience but have enough “upside” to make a reasonable case to be chosen over Bush. 

So if we concede that it’s at least possible that Bush is the 6th best available starter right now, we stil have to ask, why the hell is Bush not at least in the pen?  I think the answer is that Melvin may still be concerned about Sheets’ health (he just missed a start with triceps tightness) and might want to make sure that Bush can jump in to make a spot start if needed.  That’s really the only reasonable explanation, although I’m not sure if it’s a good one.  If it’s a short term situation, I’m fine with it.  But if Bush is still in the minors in July, with both McClung and Turnbow still in the Brewers’ bullpen, it will be laughable.

I generally like Melvins’ moves but sometimes he’s a hard guy to figure out.  I mean, the Brewers have had 14 pitchers on the 25 man roster the last 2 weeks!

I’ve been with Rob on this one for awhile now. It’s clear from watching nearly all of Dave Bush’s starts that he is not the same pitcher from the windup that he is from the stretch.

I don’t know if it is that from the stretch he loses some of his deception, his breaking ball flattens out, he loses velocity, or command, but something happens. this, i think, is a great candidate for a pitchfx study.

I believe there is a real reason, attributable to Dave Bush, that with nobody on base opponents have hit:

.253/.307/.420

and with a man on first they’ve hit:

.314/.355/.546.

In the NL in 2007, the difference was .030 points of OPS between those two base states. For Dave Bush’s career it’s .174 points of OPS, he’s a different pitcher when he is worried about the runner taking second.

Now, if you accept the premise that Bush is a much worse pitcher from the stretch, and you acknowledge that his command, this season, is not as good as it has been in the past, meaning he is more likely to pitch with men on base, then, it makes sense, doesn’t it, to give him some time to work on improving his mechanics?

As Melvin, what would you have done MGL?

Personally, i thought moving Parra to the bullpen and ditching McClung was the right move, but that meant possibly losing a player. Also, here, i’m assuming the command and walk rates are not symptoms of something else, than can be fixed with time and training.

The splits mean nothing without knowing the number of PA. 

Bases Empty, 1649 PA, .253 .307 .420.  That’s roughly a .320 wOBA.  1 SD = .012

Man on 1B only, 431 PA (excludes the 10 SH), .314 .355 .546.  That’s roughly a .390 wOBA.  1 SD = .025.

Those numbers are a bit less than 2 SD from the league mean (to either side).

Assuming I’m doing this correctly, those two sample means are 2.5 SD apart, which makes this fairly significant.

Trumping this in any case would be scouting observation that shows he is actually doing something different.

Agreed that this looks like a PITCHf/x candidate.

***

Granted all that, I don’t think any minor league rehab should last more than 2 weeks.  Even if his 1B-only splits are below average (they are likely better than .390), he’s still better than replacement.

Tango, lately you have been looking at the SD of various splits and then saying whether they are “significant” or not.  You are working backwards! 

FIRST we have to determine how much skill there is in the population and THEN we can look at the split and make the “regression correction!”

If there is little or no skill in the population, then whether an individual is 1 SD or 3 SD, it makes little difference in terms of estimating his future split.  You CANNOT look at the number of SD for any split in isolation.  For example, a 2 SD platoon split for a RHB is completely different from a 2 SD split for a LHB. Same with clutch (a 2 SD split, say for one season, still gets regressed almost 100%).

So let’s please first determine the regression amount and then you can take your 2 SD or 3 SD splits and tell us how significant it is, in terms of an estimate of true talent.

It is misleading to take the guy the with the highest split (that is why people are concerned - this is a classic case of extreme selective sampling) and then say, “See, with such a high split, it must mean something!”

That is like looking at 600 series of 100 coin flips, taking the most extreme one (the one where you got 60/40 heads) and saying, “See, this one must mean something, it is so rare.”

I don’t have to explain to you the flaw in looking at someone’s splits in isolation (without first determining the spread of skill in the population).

If we want to talk about whether and by how much Bush’s “true” splits are with and without runners on base, there is only one way to do that.  First determine the regression equation and then apply that to his splits.

Just looking at the splits in isolation, and seeing that they are large, tells us virtually NOTHING.  That is especially true, since he was selected in the first place, at least in part, for having a large split (that is why his ERA and NERC are different, supposedly at least - they really are not - and why he is perceived by MIL and their fans to not be an effective pitcher).

All that being said, his mean ERA+ in 4 years of starting is 102, which is almost exactly what his NERC is!  Where is the evidence that his ERA is NOT commensurate with his components??

A mean ERA+ of 102 is a solid #3 starter at least and probably worth 8-10 mil a year as a FA!

As far as BP’s projections go, I am not a big fan of their pitching projections, but I would guess that the ERA projections for Parra and Villanueva are at least partly as relievers.  As we all know, there is a gigantic (1 run in ERA) difference between when a pitcher relieves and when he starts.

Here are my NERC projections for Parra, Villanueva, Suppan (who is a vastly overrated pitcher) and Bush, all as starters:

Bush 4.16
Parra 4.24
Villanueva 4.46
Suppan 4.60

Tango, as far as I am aware, he is NOT on a minor league rahab.  There is nothing wrong with his health I don’t think.  They sent him down because they think he has not been pitching well, and they don’t particularly like him in general.

If they don’t like him, they should trade him.  A smart team will give them probably 6 mil worth of FA value for him (discounted for his status as an arb eligible player, or whatever he is).  That would be a steal for the team that gets him, and since they consider him a 5th starter type, at best, that would presumably be a steal for the Brewers.

MGL, have you considered changing your ERA standard of 4.00 to something closer to current reality--say 4.40? It would make all of your results easier to interpret.

yeah, I know, but I use it for a million other things, so for me it is not that practical.  I guess I can just add .4 or .5 or do a ratio or whatever, whenever I post NERC numbers.  It is usually to compare one pitcher to another anyway, so it doesn’t matter of course.  I am used to using 4.00 for the average pitcher. I’ve been doing that for almost 20 years, but I understand that other people are used to seeing ERA’s in the 4+ range.  Of course, there has to be a standard for both leagues and across years and eras.

watching him; he’s worse with men on, he had a knack for the key homer that lasted more then a season.  backed up by his HRs/9 recently, 1.21 career but best years were the league switch and rookie year.  his best year is the year he switched leagues.  last year 217 hits, 186 IP.  that’s bad in front of milwaukee’s defense.  that gets you to 100 pitches by the 5th and even if you got some unearned runs in there your done by then.  this year he’s walking people and has an option left unlike the rest of milwaukee’s staff sans Parra.

Yost said in the pregame show tonight, and i’m paraphrasing, that Bush is going down for 2, maybe 3 starts, hopefully it won’t be too long. Right now he’s struggling with his command and that’s really important for Bushy so we’re going to give him a few starts in Nashville to work it out.

Yost is a huge fan of Bush, especially his mental makeup, and it was a difficult decision. They’ve identified some things mechanically that he needs to work on, and he needs to do it in AAA and not the majors.

MGL: when I said “minor league rehab”, I didn’t mean to imply that there was a physical health issue, but simply to “rehabilitate” for whatever reason.

http://wordnet.princeton.edu/perl/webwn?s=rehabilitate

rehabilitate: restore to a state of good condition or operation

MLB Collective Bargaining Agreement, ARTICLE XIX, section C (which talks about minor league rehabs):

C. Disabled List-Assignment to Minor League club

(1) There shall be no assignment of a Player by a Major League Club to a Minor League club while such Player is on a Major League Disabled List; provided, however, that with the Player’s written consent, a copy of which shall be forwarded to the Association, and with the approval of the commissioner, a Player on the Disabled List may be assigned to a Minor League club for up to a maximum of twenty days (thirty days for pitchers) for each injury, or reoccurrence of an injury, for the purpose of rehabilitation, subject to the limits contained in Article XIII(H). Separate consent shall be required for a rehabilitation assignment for a new injury or a reoccurrence of an injury. No consent shall be effective for longer than twenty days (thirty days for pitchers).  Players may not be reinstated from the Disabled List for purposes of assignment to a Minor League club until they are ready to play.  Players who are injured and not able to play may not be assigned to a Minor League club.

(2) Notwithstanding Section C(1) above, a Player who is injured and not able to play may be assigned to a Minor League club:

(a) During the period immediately following the close of the championship season and before the filing of Major League Reserve Lists under Major League Rule 2(a), if:

(i) the Player’s Major League Uniform Player’s Contract does not cover the next succeeding season; and

(ii) the Player, if he otherwise would have been eligible for salary arbitration as a “Super Two” Player (see Article VI(F)(1)), may elect free agency under the procedures contained in Article XX(D) in lieu of accepting the assignment; provided, however, that a Player who accepts the assignment shall not have a right, by virtue of such acceptance, to elect free agency following the next succeeding championship season.

(b) During the period immediately following the filing of Major League Reserve Lists and before the 15th day prior to the start of the next championship season, if:

(i) the Player has less than three years of Major League service;

(ii) the contemplated assignment would not be the Player’s second (or subsequent) career outright assignment since March 19, 1990;

(iii) the Player had no Major League service the prior championship season; and

(iv) the Player was not selected by the assignor Major League Club in the immediately preceding Rule 5 Draft.

(3) Any service with a Minor League club while on rehabilitation assignment shall be deemed to be Major League service as defined in Article XXI. A Player so assigned shall continue to receive his Major League salary and the other rights and benefits of such Player shall be in accordance with past practices relating to assignments to Minor League clubs; provided, however, that all such players shall be treated as if they were Major League Players on the road for purposes of hotel accommodations and the daily meal and tip allowance. Such assignment shall not be counted as an optional assignment under Major League Rule 11 or for any other purpose, and waivers shall not be required.

So perhaps we should only say “minor league rehab” we should mean: on the DL, but playing in the minors to get back to the Majors.  Certainly not a big deal, but it would makes things clear, as Bush was sent down via optional assignment.

I posted the whole section because I figure there are some people that don’t know how it works.  My apologies if it is too long.

Obviously, I am not there, and I don’t know the situation and the personnel, but given how obsessed most teams and people in general are with short term fluctuations, their inability to separate random from non-random events, and their obsessions with imputing physical and mental “causes” to likely random events, my best guess is that sending a 4 or 5-year veteran to the minors to work on his command is assinine.  Again, the fact that he has an ERA near 7 and is 0-3 is the likely catalyst for such a panic move.

As I have said many times in the past, although not lately, one of the signs of a poor organization (one that is obsessed with short-term fluctuations) is continually shuffling around their pitchers.

One thing this does for sure is screw Dave Bush.  And it maybe why he got sent down.  Bush entered the year with just over 3 years of service time.  Meaning if he spends what, 6 weeks(?) in Nashville he doesn’t get a full year of service time for 2008 moving his free agency date from 2011 to 2012.

End/14: yowza.  Isn’t there an exception that if you are sent down for 20 days or less, that it still counts towards your years of service?  Note also that a MLB season is around 183 days or so, but a service season is 172 days, so everyone gets a “free” 11 days.

***

MGL, Eric at Fangraphs was posting that Bush was down to throwing 58% strikes (based on 400 pitches), while his career is at 66%.  1 SD = 2.4%.  So, that is alarming for someone to make that kind of change after being successful at a certain level.

Your point stands though that if his ERA was 3.50 AND his strike% was down to 58%, they wouldn’t have made the move.

14 If that were to come about, it would be 2010/2011 not 2011/2012.

MGL, Eric at Fangraphs was posting that Bush was down to throwing 58% strikes (based on 400 pitches), while his career is at 66%.  1 SD = 2.4%.  So, that is alarming for someone to make that kind of change after being successful at a certain level.

For one thing, we’d need to know the league average.  If it is less than 66, then we would expect his career % to regress a little.  Also, we don’t take the 8% difference and say that it is over 3 SD.  Both are samples, so we take the sum of the variances and then the square root of that as our “expected SD from chance.” OK, that only makes it 2.55, assuming he had around 10,000 previous pitches, so not much difference.

I’d also want to know against what players those 58% strikes were as compared to his career.  Maybe (not likely of course) he has faced a pool of batters so far this year that require fewer
strikes.  Hey maybe they just need to tell him to throw more strikes. The one thing that a pitcher can easily control is their percentage of strikes.  A pitcher basically chooses how much on the corners he wants to pitch.

If I am a pitching coach or team, I am basically looking at 3 things:  One, velocity.  If that changes (for the worse), and the pitcher does not do that on purpose, than something is wrong and his performance will be worse, everything else being equal.  Two, the movement on his pitches - that will obviously vary a little, but I basically want to know if all his pitches are basically the same as they always were.  Three, his mechanics, although if his mechanics are “off” that should show up in something else, and if it doesn’t, who cares, other than he might be a greater or lesser injury risk.  Finally, I want to see or know if he has changed anything about the “way” he pitches, including his repetoire - is he throwing more or fewer of certain kinds of pitches in certain kinds of counts. Along with that of course is his “mixing up his pitches.” Is he being more or less predictable than before.  Now, those things do not tend to change other than if a pitcher deliberately changes them (his repetoire and randomness), so I would mostly focus on the former things, and probably just ask him if he has changed his style.

I mean if a pitcher is throwing fewer strikes, walking more guys, and his performance is suffering, but he is healthy and his mechanics are the same, what can I say/conclude?  All I can do is assume that it is just a fluke and let him pitch - maybe let him know that we want him to throw more strikes, but that is about it. If it continues, then we might have to evaluate.

One of the most important things in using “correct strategies” when you are faced with randomly fluctuating sample performances every minute and every day, is to be PATIENT - to EXPECT random fluctuation and not worry about it too much!

That is why bad teams continually shuffle their relievers back and forth between the majors and minors (they don’t understand the above concept), and why I assume that good teams, if there are any in this respect, leave their good pitchers (as determined by scouting and long-term stats) alone and let them pitch through all of the day to day fluctuations you should expect in pitching, and not let your bad pitchers pitch, or let them pitch in low leverage situations only, regardless of how good or bad they have been pitching “lately.”

Bush had an ERA over 6 after his first 10 starts last year and the Brewers didn’t give up on him.  I think he’ll be back soon.

After Suppan’s (who, as I said, is actually by far the worst starter on the team) performance today, it might be sooner than later.

It is not really apples and apples, but you have Bush in the minors and Ponson starting tommorrow for Texas.  Of course, Texas is one of the worst organizations in baseball…

Well, Bush is already back up but only because Gallardo tore his ACL. If we assume a 1 run/9 difference between Yo and Bush, that’s about 2 less wins over the rest of the year?

Real quick, off the cuff here - ZiPS has Gallardo as a 3.66 ERA pitcher, and Bush as a 4.64 ERA pitcher, and a league average pitcher as 4.55 ERA. That means that Gallardo is a .599 win percentage pitcher, and Bush is .490. (I may have been a little aggressive with the rounding because I don’t have so much as a spreadsheet or a graphing calculator to hand.)

Assuming 28 starts for the rest of the season, that’s a 3.74 WAR for Gallardo or a 1.79 WAR for Bush (I’m presuming the same distribution of innings per start for both of them - I have no idea how close that is to the reality of the situation.) So, yeah, about two wins.