Nationals Arm Race

"… the reason you win or lose is darn near always the same – pitching.” — Earl Weaver

Archive for the ‘joel zumaya’ tag

Starting versus Closing

5 comments

Should we try Clippard as a starter? Absolutely! Photo: NationalsDailyNews/Meaghan Gay/DCist.com

Baseball writer extraordinaire Tom Verducci posted a fantastic article today talking about Neftali Feliz‘s proposed move from the Rangers closer to the starting rotation.  The article touches on a topic that I’ve been meaning to write about for a while; Starting versus Closing.  It also is literally the best summation I’ve seen yet describing why the save is over-rated, closers are overpaid and why you’d rather have starters versus relievers.

Lets face it; for the most part relievers are failed starters.  A few get drafted or signed as relievers (Washington’s Drew Storen being one local example), but most starters are drafted as starters and work their way through the minors as starters.  Some starters discover that they can’t develop secondary pitches, but their primary pitches are so fantastic that the club (rightly) turns them into relievers.  This especially allows hard-throwers (think someone like Joel Zumaya) to have a career despite the fact that they only really have one pitch and throw with such effort that they could not possibly last 6+ innings.

Minor league relievers definitely make the majors, but most often as either LOOGYs or rubber-armed replaceable right-handers (think Miguel Batista) out of the bullpen.  In recent  years the desire to have more and faster throwing arms out of the bullpen has led to more pitchers opting to become relievers sooner, but they still are converted out of starting roles for either performance or fragility.

Two items from his story that I’d like to comment on:

1. Managers don’t use Closers in the most high-leverage situations. I could not agree more.  When is the best spot to use your best, most reliable reliever?  In a one-run game in the 6th when your starter runs out of gas and loads the bases with one out?  Or at the beginning of the 9th inning of a 5-3 lead?  Verducci is right; managers in the modern game are slaves to the save statistic and will not bring in their closer unless its a “save situation.”   But he also notes what is common knowledge; that you could be putting out the 12th man in your bullpen and probably have only a slightly worse chance of getting 3 outs without losing the game for your team.  Per the article, 94% of 2-run leads in the 9th inning are won irrespective of who you put out there, and that percentage has not changed significantly over the past 50 years of baseball.  Joe Posnanski also wrote about this same topic in November with similar results, finding that teams in the 50s closed out games with the same regularity as teams now, but without high-priced one-inning closers.

Luckily for the Nats, we look to have 3-4 different guys who are of sufficient quality who we CAN bring in to a game in the 6th and get a high-leverage situation.  Storen, Clippard, Burnett or newly acquired Henry Rodriguez all seem to fit the bill.  But that doesn’t mean that we don’t have a manager in Riggleman who is in the “slave to the save” category.  Matt Capps was brought in to be the closer and he closed games.  That’s it.  It is safe to say that if Riggleman decides on a closer, that’s going to be his role and that’s that.

The save stat is ridiculous and most people know it.  You can get a save in a game where you give up 2 runs and 5 hits in a 1/3 of an inning.  You can get a save when you perform mop up duty but let the score get too close while you rubber-arm your way through a meaningless blowout.  The save takes nothing about the pitcher’s performance into account; only whether or not the game ended while he was on the mound and the win was preserved.

But the save stat, and its monster creation the specialized one-inning closer, are here to stay.  Prospects come up through the ranks specifically to be closers, free agent players will only play for certain teams if given “the chance to close.”  Closers are well paid, and their pay is directly tied to this flawed save statistic.  Statisticians have tried to create a better set of metrics for middle relievers (“Holds” mostly) but the reality is that closers have high leverage in salary situations while middle relievers are lucky to get paid a bit more than the veteran’s minimum.  Verducci touches on this ridiculousness, pointing out that Papelbon‘s higher salary in 2011 than Cole Hamels despite the relative levels of production for their teams.

Ironically, some Major League managers *know* this fact, but continue to trot out their best reliever for a 3-out save at the beginning of the 9th inning in a 3-run game.  They do the same as the other 29 managers because the radical idea that backfires directly leads to termination.  No manager is willing to risk their job to try to do something the right way.  To say nothing of the reaction of a highly-paid FA closer who is suddenly told he’s going to be primarily used in the middle of the 7th to clean up the starter’s mess.

It makes you wonder if there’s a better way.  Here’s two radical suggestions:

1. Comprise a bullpen with no named closer role, and tell the entire 7-man bullpen they’re doing closer-by-committee.  It may infuriate fantasy baseball players and the union (since saves translate to salary for their FAs), but it probably placates an entire roster of wanna-be closers.  Imagine if 5 of the 7 guys in your bullpen (leaving out the LOOGY and long-man) know they may be brought in to rescue a game in the 6th or close it out in the 9th, and their roles change on a daily basis based on use.  That to me is a far better situation than pre-naming a closer (which invariably is the best guy out there) and then never using him until the 9th.

2. Comprise an ENTIRE pitching staff of long-men relievers.  Imagine if you didn’t have starters at all, but an entire bullpen of guys who were geared to pitch 2-3 innings every other night.  You would never have a need for specialized closers or even high-priced starters.  You’d rotate through who got the start, the starter would go 2-3 innings, then the next guy would go, and you’d repeat this until the game was over.  It’s kinda like spring training but all year.  Since these guys are only throwing 2-3 innings, they should be able to repeat this task multiple times in a week.

There’s 54 regular innings to be had per week mid-season (6 games at 9 innings per).  54 innings divided out by 12 guys in the pen means about 4.5 innings per WEEK per pitcher.  If you split those 4.5 innings up across three games you’d be pitching (say) 2 innings on monday, 1 on thursday then 1.5 on saturday.  That’s pretty manageable.  Plus if everyone else is doing the same, you can rotate through the guys and slightly adjust based on how they’re pitching that day.

Plus, think about how CHEAP this pitching staff would be.  12 middle relievers could not possibly cost your team more than about $15-20M annually in salary, even if they were mostly on veteran contracts.  Roy Halladay makes more than that in 2011 just by himself.

Coincidentally, this is exactly what Tony LaRussa tried at one point in the early 90s with the Athletics.  Unfortunately his experiment ended quickly, failing less because of execution and more because of lack of support from his players and management.  Its just a matter of time before someone tries it again.


Here’s the second item i’d like to comment on:

2. Starters are FAR more valuable than Relievers or Closers.  Last year in the midst of Clippard’s fantastic middle-relief run I asked myself, “Why isn’t Clippard in the rotation?”  He pitched 91 innings spread out over 78 appearances and only gave up 69 hits.  He maintained an 11.1 K/9 ratio, which is better than any starter in 2010.  91 innings was good for 4th on the entire staff in 2010.

The leading argument i’ve read for Clippard staying in the bullpen relates to the nature of his stuff.  He’s got a sneaky good fastball, a decent curve but his bread and butter pitch is the change-up.  Apparently the knock on him is that hitters adjust to him more quickly and thus he makes more sense in a relief role.  In a starting role hitters would be getting their third crack at him in the 5th or 6th inning, right when he’s tiring and right when he’s vulnerable.  In relief, he can “show” all his pitches in one at bat and work each batter individually, then leave the game before his “stuff” is exposed.

Clippard was a starter his entire minor league career, and his minor league numbers were pretty good.  He always maintained a small hits-to-IP ratio, a good k/9 ratio.  It wasn’t until he reached the majors that suddenly he couldn’t start.  I think perhaps he’s either gotten pigeonholed or he’s psychologically set in the reliever mind-frame now.

A quality starter gives your team 6+ innings, works through the opposing team’s batting order nearly 3 full times and keeps your team in the game.  6-7 innings at a 3.00 era is invaluable for your team’s psyche as it tries to win game after game.  Leaving just 2-3 innings a night for a bullpen staff of 7 means that there’s fewer days when your staff is over worked and you have to give up games for lack of bullpen arms.

How about using career WAR as a bench mark?  I don’t really like the career WAR analysis (since it is an accumulator stat and a mediocre guy with 22 years of experience appears to be better than the best pitcher of his day who only had a 15 year career).  But it is telling in this situation.  Here’s a link to career WAR for pitchers at baseball-reference.com.  And here’s the rank of the 5 best relief pitchers of all time (the 5 relievers currently in the hall of fame), along with the rankings of some of their active contemporaries who seem likely for the hall.

Lname Fname Career WAR Rank
Smoltz John 38
Eckersley Dennis 46
Rivera Mariano 69
Wilhelm Hoyt 121
Gossage Goose 133
Hoffman Trevor 215
Wagner Billy 238
Sutter Bruce 315
Fingers Rollie 325

Smoltz and Eckersly both started for large portions of their career, hence the high rank.  Mariano Rivera is clearly (in my mind) the greatest reliever who has ever played and his career WAR shows.  But notice how low closer-only guys like Sutter and Fingers are on this list.  Both are currently below modern day starters Ted Lilly and Kevin Millwood, again guys who are hardly listed as being among the game’s elite.

By means of comparison, Trevor Hoffman, who is ranked 215th all time is ranked just ahead of one Freddie Garcia in all time WAR.  Now, is Freddie Garcia a serious hall of fame candidate?  Not likely; he’s currently on a minor league contract offer with the Yankees after nearly washing out of the game two years ago.


Oh, coincidentally, I absolutely think Felix should be in the rotation.  As should Aroldis Chapman in Cincinnati.  Because they’ll be able to help your team win on a much more frequent basis.  You always want the chance of 180 innings of quality versus 60.  Its that simple.

Did Chapman really throw a ball 105mph?

one comment

Chapman is amazing. Photo via wiki/flickr SD Dirk

Chapman is amazing. Photo via wiki/flickr SD Dirk

(Editor’s Note: Originally published on 9/2/2010, edited several times since, including links to the 105.1 pitch later in 2010 and him hitting 105.1 again in 2016).

If you believe twitter and these published reports, Cincinnati’s Cuban prospect Aroldis Chapman “sat” at 103 and hit 105mph in his final AAA appearance before getting called up for the 9/1/2010 roster expansion.  I had a hard time believing it; the picture in the Yahoo looks doctored frankly, and stadium guns are notoriously “jacked up” to get the crowds excited and to spark interest in hard throwers.

There are two kinds of radar guns used by stadiums and scouts today.  A “fast” gun measures the speed of the pitch as it leaves the hand of the hurler while a “slow” gun measures its speed by the time the pitch crosses the plate.  For obvious reasons you’ll take the “fast” reading if you’re trying to hype the player, and you take the “slow” reading if you’re trying to evaluate the pitcher.  Oh, and you take readings over and again, compare first inning to last inning, etc.  My thoughts were that Chapman was on a fast gun that was over-exaggerated to add a couple MPHs (case in point; Jordan Zimmermann‘s radar readings on the telecast two nights ago put him at 96mph when his pitch f/x maximum was 94).  Click here for an excellent overview of the way we’ve measured velocity has changed over the years.

Now, that being said, check out the Pitch F/X data from Chapman’s first appearance in the majors on 8/31/2010, and his second appearance on 9/1/10. In his major league debut he threw 6 fastballs with an average of 100.65 and a max of 102.7mph.  This in and among itself would have been amazing and would have tied the fastest ever recorded pitch f/x data (that of Joel Zumaya hitting 102.7 in a game in June 2009)

However, look at the data from 9/1/10. He threw 7 fastballs that AVERAGED 102.11 mph with a peak of 103.9mph.  103.9!!  Now, I know there’s a ton of links out there talking about what the fastest ever reported pitch is on record (see links here or here at HardBalltimes or this google cache’d document as well as wikipedia or Guinness Book of World records links).  Nolan Ryan‘s “record” of 100.9mph has stood as some sort of altar to the speed ratings for years and years.  You’ll see radar gun readings as evidence that various players have thrown the ball 102, 104, whatever.  Steve Dalkowski and Bob Feller being “recorded” using primitive measures in the 107 or 110 range.   Walter Johnson was known for his power in the early part of the century; his fastball was estimated to be in the upper 90s by lore, and 91mph by testing.  Its a shame that modern technology didn’t exist back then.

So, what is the “proof” that Feller threw 107.6 in 1946?  Well, its complicated.  This ESPN article covers it: basically Feller threw a ball through a Lumiline Chronograph device, which hyper-accurately measured speed at the time.  His best measure was 98.6mph … but modern pitches are measured at the point of release, not where it crosses the plate.  So, you have to estimate how much velocity was lost from the point Feller tossed the ball to the device.  If you look at the photo (in the ESPN article), Feller was not throwing from the mound, but somewhere well forward of it.  So, to standardize speeds to some point closer to the release point of the pitch, we have to estimate the percentage loss of Feller’s pitch from an estimated delivery point based on a photo, then back into a number.  Some estimate this to be in the 103 range, others claim 107.6, which would imply he was throwing off the mound and experiencing the full expected velocity loss (which is clearly not true from the photo).  So I’d believe Feller was throwing roughly 103-104 based on where he released the ball … but its an estimate.  Add back 1mph for every 7 feet, and Feller seems to be standing at about 35 feet, so add perhaps 5mph to his 98.6 and you get the 103-104 rough estimate.

Coincidentally; the new standard is speed of the pitch measured at 50 feet from home plate.  This standardizes to allow for different release points by pitchers of different heights.

How about proof that Ryan threw 108.1, as is throw around the internet?  Well, It comes from the famous 100.9 reading he put up in a start in 1974.  However, that was reading was made using technology that, again, measured the speed closer to the plate than we currently do.  Per this article (which summarizes findings from a documentary called Fastball), the radar technology measured 100.9 ten feet from home plate, meaning we have to back up to the 50 feet estimate.  So, if the speed was measured at 10 feet, and we normally measure at 50 feet, we should add back “40 feet” of velocity loss, at around 1mph per every 7 feet, so call it somewhere between 5-6 mph … and we’re somewhere in the range of 106-107mph.  Not 108.1 as everyone claims; that’s adding back too much velocity and/or giving Ryan 10 feet more of credit (the measurement at the plate versus 10 feet in front).  These are still estimates of course; these speeds are also affected by the weather at the time (it was in Sept of 1974 in Anaheim … was that a humid day, an arid day?  Anaheim’s typical September weather is humid, hot; was it really humid that night?  Was it relatively dry?  That makes a huge difference in aerodynamics of a baseball and potential velocity loss.  So … we’ll never know.

Coincidentally, these velocity loss estimates are best documented by a physicist at the University of Illinois named Dr. Alan Nathan.  He estimates that baseball pitches lose 9-10% of their velocity from the point of release to the plate.  Example; if you release a ball at 100mph, it will generally measure at 90-91 by the time it crosses home plate.  Or, to put it easier … a pitch loses about 1MPH every 7 feet.

(Side Note: if you’ve never heard of Dalkowski … his wikipedia page is a pretty good read).

Other rumors and records for fast pitches suffer from perhaps hype and estimates, not science.  Zumaya reportedly hit 104.8 in the first game of the 2006 ALCS, but it was on the stadium gun, not Pitch F/X.  Ryan and Goose Gossage both reportedly hit 103mph in the 1978 All-Star game.  Stephen Strasburg reportedly could hit 102 as an amateur but rarely goes about 97 post-arm injury.  Justin Verlander may not have the triple-digit records but amazingly maintains his upper-90s throughout games, often hitting 100mph in his last inning of work.

I think Chapman’s speed last night is the new standard.  And a pretty amazing accomplishment.

(Editors notes: Chapman set a new MLB record on 9/24/10, after this article’s initial posting, throwing a documented 105.1mph.  He also threw a ball at 104.6 on 7/28/14.  Not quite as fast as his best ever … which I’m guessing even Chapman doesn’t think he’ll ever beat, since he got the reading tattooed on his wrist.  Then on 7/19/16,  years after his first record, he tied it again, throwing another ball 105.1.  Here’s the StatCast link for this pitch.  Coincidentally, this 105.1 pitch is now officially registered at 105.8 by MLB, because the technology we use now standardizes the pitch speeds at a specific distiance, and Chapman’s 105.1 was measured further down from where we now standardize.

Post-posting related link: BA’s J.J. Cooper posted this Q&A session on 4/28/15 talking about Chapman and 101+ MPH pitches.

Jan 2017: Boston pitching prospect Michael Kopech, who is renowned his velocity, hit 110 MPH during a “max velocity” workout early in spring training by crow hopping.  Impressive, but not the same as throwing off a mound.  Holy mackerel!  How long before this kid gets to the majors and has a gun on him??

May 2018: we may have a new gold standard; Jordan Hicks threw a reported 105.1 MPH fastball, but Pitch FX shows it at 106.1.  (Unfortunately the PitchFX data at BrooksBaseball site now does not work, so my “evidence”is gone).  However, subsequent reporting on this pitch have lowered it to the mid 104s, so it remains unclear what its real velocity was to this day.

May 2022: University of Tennessee’s Ben  Joyce threw a pitch recorded at 105.5 in a college game.  Does Tennessee have Pitchfx?  Was this a juiced gun?  We’ll have to see.

Written by Todd Boss

September 2nd, 2010 at 12:21 pm