Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Friday, May 30, 2014

Even A Team’s Fifth Starting Pitcher Matters When Trying to Make the Playoffs

New study shows statistical evidence that anybody throwing 170 innings is important to a teams’ won-loss record. Groundbreaking discovery!

Joyful Calculus Instructor Posted: May 30, 2014 at 03:45 PM | 24 comment(s) Login to Bookmark
  Tags: jonah keri

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. SoCalDemon Posted: May 30, 2014 at 05:27 PM (#4716262)
Ummmm, so I read the whole article, and....am I missing something? More specifically, am I missing the people that they are arguing against? (although I suppose I do disagree with their whole 6 WAR pitcher thing. There essentially is not, and can't be, any such thing as a 6 win 5th pitcher, in any meaningful sense. I am phrasing this wrong, but, first, such pitchers are really rare to begin with, and if your expected 5th starter puts up that kind of performance, he quickly becomes your ace (maybe this subset are just young but clearly good pitchers?), and on any staff where your 4th and 5th starters are expected to be anywhere near that good (am thinking of the 2011 Phillies here), first, thats really rare, and thus really, really small sample sizes, and second, unless your offense is horrid, it seems like an academic point, you are going to win a lot. I think I am rambling, but this article made my brain angry.
   2. Baldrick Posted: May 30, 2014 at 05:36 PM (#4716267)
What a mess. The only way you can have a 'really good 5th starter' is if you also have 'really good 1-4 starters.' Otherwise he's not your 5th starter. It's also pretty rare to HAVE a '5th starter.' Most teams give a smattering of starts to a ton of guys at the bottom end of the rotation. Because what else are you supposed to do?

It's possible that this article makes these points somewhere. But I really had no clue what they were talking about, what question(s) they were trying to answer, and why I should listen to any of it, so I gave up partway through.
   3. SoSHially Unacceptable Posted: May 30, 2014 at 05:48 PM (#4716270)
In most cases, a team's sixth, seventh and eighth starters are pretty important also.

   4. madvillain Posted: May 30, 2014 at 05:53 PM (#4716272)
In most cases, a team's sixth, seventh and eighth starters are pretty important also.


Yep, you hope like heck you don't have to dip much past the 6th guy, but for most teams hope isn't gonna work. Chicago is already on plan 7 and 8. Thanks to a small miracle, Don Cooper, and Rick Hahn's scouting team, plans 6 and 7 -- Andre Rienzo and Hector Noesi -- have actually been cromulent. Scott Carroll, plan 8, is thankfully banished to the long man role after Sale came back. Not even Cooper could fix'em.
   5. SoCalDemon Posted: May 30, 2014 at 05:58 PM (#4716273)
Also, their eureka, we found something thing, that 6 WAR 5th starters is worth 1 win more than you'd expect; there is no way in hell that is statistically significant with the sample sizes involved.

Following on what other people have said before, I think I might hypothesize that "having" a true 5th starters (something like at least 25 or 28 starts from your top 5 guys) would be a bigger predictor of playoff success than quality of your 5th starter. Somebody other than me should run that study.
   6. Greg Pope thinks the Cubs are reeking havoc Posted: May 30, 2014 at 06:06 PM (#4716274)
There essentially is not, and can't be, any such thing as a 6 win 5th pitcher, in any meaningful sense.

Is there any concept that replacement level is different for your #1 vs. your #5 starter? Correct me if I'm wrong, but replacement level is different for SS vs. 1B, right? So it could be argued that replacement level should be higher for your ace, since every team has an ace (their best pitcher, whatever you want to call them). In that case, your 5th starter might actually be 6 wins better than the worst 5th starter, without actually being that good.

Of course, I'm guessing the reason nobody does this is because you don't slot your pitchers 1-5 and then keep them that way all year, and have them perform in that order. Many years your nominal #3 ends up better than your ace, etc.

   7. Jacob Posted: May 30, 2014 at 06:20 PM (#4716278)
Somebody other than me should run that study.


Poz touched on the subject a bit.

http://joeposnanski.com/joeblogs/full-time-starters/
   8. Moeball Posted: May 30, 2014 at 06:21 PM (#4716279)
So let me see if I understand this - keeping your starting rotation healthy all season long so that even your #5 starter can get in enough innings to qualify for the ERA title - this is a good thing with a reasonably good correlation for team success?

Wow, I never would've guessed.
   9. Walt Davis Posted: May 30, 2014 at 06:44 PM (#4716285)
Reading it now ...

1. It starts off dumbly. With the idea that maybe the Ms should have signed Ervin Santana to be their _fifth_ starter. Hopefully it's just sloppy writing (or sloppy reading) and what he means is that signing Santana as a 3rd starter would move everybody down a slot, providing better depth.

2. Really it starts off worse than that since we don't seem to have a definition of "5th starter". There's "5th intended starter" and there's the guy who end up being your 5th best starter. The latter obviously provides a lot of information about how the team did but it's after the fact of course. But if Santana is this author's idea of a 5th intended starter, they're still off the mark.

3. When he finally gets around to defining 5th starters, he still doesn't tell us what the definition is. No, not even in the footnote which I checked. The footnote tells us they did not just look at season's end data, find "the worst guy" and declare him 5th starter ... which is good since the worst guy was probably the 8th starter. But other than telling us they used retrosheet, he doesn't tell us how 5th starter is defined. 5th pitcher to start a game that season? With or without adjusting for early-season injury? Starter with the 5th most starts? (presumably not) Using retrosheet implies they did not use ZiPS/Pecota projections to identify the projected #5. Did they distinguish between #5 and #6, etc.?

There's no "right" way to do this, there probably are wrong ways but you've got to tell us how because it makes a big difference in what the results mean.

4. I can't follow their table. But near as I can tell, it's the opposite of how they're interpreting it. Their table appears to show that it doesn't really matter. Put a replacement-level 5th starter out there and it results in a run differential of a about -20. But that's -20 vs. 0 (average), not replacement (about -20). If anything, putting a replacement-level 5th starter out there costs you only 1 win vs. average, not 2. That effect is pretty much constant up the line -- a 1 WAR starter gets you 2 wins above replacement; a 2 WAR starter gets you 3.

It's true that somewhere around that magical 6 WAR 5th starter the effect goes up .5 WAR. Hooray.

5. So why are 5th starters "worth more" than they seem?

6. They list some specific big "5th starter" seasons. The best listed is 4.9 f WAR. Where's our 6-WAR 5th starter? Our 9-WAR one?

7. In that list you have a lot of pretty big names, usually as youngsters or in their 30s so it is plausible they were their team's intended 5th starter and/or 5th guy to start a game that season.

8. From that table, it looks possible that the 5th starters that provide extra values are ones on high-scoring teams. That might make sense in that they're more likely to leave the game with a lead and the team is more likely to use good relievers to protect the lead, resulting in a greater average run differential. I wonder what the actual team W-L record was in those games and whether the extra runs were really necessary.
   10. SoCalDemon Posted: May 30, 2014 at 07:09 PM (#4716291)
Re #6: I think that doesn't work, because you need multiple starters. There is no "5th" starter position, there are just 162 slots you need to fill, and at least in the modern game, no one guy is going to fill more than 35 of them. Furthermore, you don't really care how good your #1 or #2 or #3 is, you care how well your starters are in aggregate. If your 5th starter was 6 runs better than the replacement 5th starter, he just would become your 2nd or 3rd or 4th starter, in terms of how we think of those terms. But there is no "5th starter". There are usually 5 guys that you start the season with, the ordering doesnt matter; it all gets scrambled in real-life soon enough by off-days in the best case, and injuries in the worse case.
   11. SoCalDemon Posted: May 30, 2014 at 07:13 PM (#4716292)
For position players, I could imagine a finding like this. In theory, anyway, I could imagine perhaps that the runs a shortstop saves, are worth more than the runs a DH creates (beget youcause runs saved are worth more than runs created, all other things being equal: 770 R, 700RA will get you fewer wins than 700R, 630RA). But for starters, you only influence the 30 or so games you are in, and they are essentially independent from any of the other games, so I can't imagine conceptually why there would be any 1st or 3rd or 5th starter effect.
   12. Greg Pope thinks the Cubs are reeking havoc Posted: May 30, 2014 at 07:27 PM (#4716301)
Re #10: You're right. If your #2 starter goes down, you don't just plug in a "replacement #2 starter". You plug in your "replacement #5 starter". I was just thinking that maybe if you looked at it after the fact and organized it this way, you could come up with a 6-WAR 5th starter. Unfortunately, it wouldn't mean anything practical. It would just sort of show why a team was good. Because their 5th starter was 6 games better than other teams' 5th starters.
   13. cardsfanboy Posted: May 30, 2014 at 07:41 PM (#4716307)
I'm not sure what this article is saying. But there is so much wrong with it.

They identify fifth starters by their turn in the rotation(I guess) and not by their quality, which has to be the dumbest way ever to identify a 5th starter. The best way to identify a fifth starter is based upon lowest era (or fip) among the top five pitchers on a team in innings pitched. Even then that ignores the fact that large percentage of the teams don't have a fifth starter instead of a stop gap fill type of starter or use the "fifth" spot in the rotation to groom young pitchers.

The concentration on the "exceptional" fifth starters really doesn't say much and of course it was an incredibly difficult read, as if the writer has never tried to publish for a mass audience before. I'm not remotely sure what the point of the article was as it never really had a concluding statement. Something about an exceptional fifth starter is worth more runs than you would expect. It really doesn't say much unless they run the same studies against all "roles". (which might indicate ultimately that regardless of role, having a high war pitcher results in more value than you would expect...a completely non-controversial concept, considering that there is a good argument to be made that War undervalues starting pitchers)



   14. Baldrick Posted: May 30, 2014 at 08:50 PM (#4716325)
Is there any concept that replacement level is different for your #1 vs. your #5 starter? Correct me if I'm wrong, but replacement level is different for SS vs. 1B, right? So it could be argued that replacement level should be higher for your ace, since every team has an ace (their best pitcher, whatever you want to call them). In that case, your 5th starter might actually be 6 wins better than the worst 5th starter, without actually being that good.

Of course, I'm guessing the reason nobody does this is because you don't slot your pitchers 1-5 and then keep them that way all year, and have them perform in that order. Many years your nominal #3 ends up better than your ace, etc.

You have to have a SS and you have to have a pitcher. You don't have to have a '5th starter.'
   15. Dan Lee is some pumkins Posted: May 30, 2014 at 08:55 PM (#4716327)
I guess I just don't understand how the authors expect teams to know going into the season who the "good" fifth starters are going to be. Going into the 2014 season, which of these guys would you rather have?

a) 25-year-old pitcher who, in 30 MLB starts, has an ERA+ of 89, has allowed 1.2 HR/9, 4.6 BB/9, 7.6 K/9, and a 4.78 FIP.
b) 24-year-old pitcher who, in 21 MLB starts, has an ERA+ of 89, has allowed 1.2 HR/9, 2.6 BB/9, 7.2 K/9, and a 4.26 FIP.

Both have had arm problems, but both came into 2014 as healthy as could be reasonably expected. Both have been up and down in the minor leagues, normally good when healthy, but neither has been spectacular for any extended period of time.

It's pretty close, right? You'd probably rather have B, but it's not obvious.

Pitcher A is Drew Pomeranz, who's won the fifth starter job in Oakland and has an ERA+ north of 270.
Pitcher B is Erasmo Ramirez, a "terrible option" (according to the article) on whom the Mariners have been "wasting starts".

Fifth starters are lottery tickets. Sometimes you win, most of the time you lose, and there's really no way to know in advance which is going to happen. Saying teams should go into the season with better fifth starter options is like saying people who play roulette should pick better numbers.
   16. TVerik, the gum-snappin' hairdresser Posted: May 30, 2014 at 10:03 PM (#4716358)
I've always thought that the first few turns through the rotation are more or less set up #1-#5, with the caveat that expectations aren't the same as results.

But if I'm the manager, in the first few weeks, I'd be tempted to go #5, #1, #2, #3, and then #4. The theory is that I am more or less guaranteed to lose the first game (their ace versus my #5), but if we're about evenly matched, I'll have the big advantage in the other four games (my #1 being better than their #2 and so on). If I don't get fired for this, I'd even set it up that way in the postseason.
   17. Cleveland (need new name) fan Posted: May 31, 2014 at 12:12 AM (#4716441)
Good points. But our line of thinking was that we wanted pitchers who were considered 5th starters *going into the season*. To that end, we looked for guys who made their team's 5th unique start of the season AND had the fifth-best preseason projected wins (with lower ERA as the tiebreaker) of anybody in the rotation. Thats a pretty good (albeit imperfect) algorithm for fifth starters, as viewed from the beginning of a season.


One of the authors posted this in the comment section explaining their definition of a 5th starter.

In 2008, Cliff Lee started the Cleveland Baseball Club's 6th game (Sabbathia started games 1 and 5) and posted a WAR of 7.1. Although Lee had a WAR of 4.9 in 2006, he was in the minors for most of 2007. He would seem to be the classic 5th starter they were looking for, but he wasn't listed in their table of best years from a 5th starter. So its unclear if they followed their own definition in analyzing the players. If anyone has the projections for 2008, it would be interesting to see what they projected Lee at versus the rest of the Cleveland Baseball Club starters.
   18. PreservedFish Posted: May 31, 2014 at 01:02 AM (#4716459)
I'd be tempted to go #5, #1, #2, #3, and then #4 ... I'd even set it up that way in the postseason.


In the postseason? That seems unwise.
   19. cardsfanboy Posted: May 31, 2014 at 01:40 AM (#4716470)
In 2008, Cliff Lee started the Cleveland Baseball Club's 6th game (Sabbathia started games 1 and 5) and posted a WAR of 7.1. Although Lee had a WAR of 4.9 in 2006, he was in the minors for most of 2007. He would seem to be the classic 5th starter they were looking for, but he wasn't listed in their table of best years from a 5th starter. So its unclear if they followed their own definition in analyzing the players. If anyone has the projections for 2008, it would be interesting to see what they projected Lee at versus the rest of the Cleveland Baseball Club starters.


That is an absolutely stupid definition of a fifth starter.. I have officially decided any baseball analysis taken at 538 is beyond silly from this point forward. To change my opinion they would need to
1. Hire better rational analyst
2. Hire people who can write.

as this article basically proves that they don't have anything in the way of either.
   20. cardsfanboy Posted: May 31, 2014 at 01:43 AM (#4716471)
In the postseason? That seems unwise.


I didn't bother commenting on that because it really doesn't work...if the series goes 7 games, you basically guarantee your worst pitcher is the the one available. Of course the post season is a 4 member rotation, so he was implying 4,1,2,3...either way it's not really a viable option.

good pitchers still lose 7 or so games a year.... automatically surrendering to the better pitcher only hurts you, not helps.
   21. cardsfanboy Posted: May 31, 2014 at 02:01 AM (#4716474)
If you think about it...assuming that on through the scale every starter is equal.
Meaning your number one is equal to their number one, and their number five is equal to their number five... you follow the logic in post 16...

Over a stretch of 5 games, going with the traditional way, you will win 2.5 games. (each game was 50/50)
obviously it's hard to argue the true math, but assuming that each pitcher is 10% better...that means a matchup between a number one and a number five would be 70/30 in favor of the number one.

This means having your number five start the first game, you are giving the opposition a 70% chance of winning. Putting the Ace the second game etc....only changes your odds of winning roughly 10%... This boils down to over a five game stretch you will win on average roughly 2.5....break even point there. I guess it really depends on how much of a difference is involved in the average team.
   22. boteman Posted: May 31, 2014 at 02:25 AM (#4716479)
I can't wait until Plan 9 comes to Chicago.
   23. GuyM Posted: May 31, 2014 at 11:04 AM (#4716519)
They identify fifth starters by their turn in the rotation(I guess) and not by their quality, which has to be the dumbest way ever to identify a 5th starter.

There are any number of reasonable ways to define "5th starter." But there is one clearly wrong way, which no serious analyst would ever employ: "identify a fifth starter based upon lowest era (or fip) among the top five pitchers on a team in innings pitched." You simply can't use hindsight to do it and reach any useful conclusions. (Among other things, "fifth starters" would probably have negative WAR on average, seeming to indicate that many/most teams are regularly using sub-replacement starters). This really isn't controversial at all among analysts...

What's potentially interesting in this study is the fWAR results, which suggest that each marginal fWAR is worth 12 runs. I doubt very much this has anything to do with 5th starters being more important than any other starters. If the 12:1 ratio is true for other rotation positions as well (and this isn't just a small sample fluke), it would indicate that fWAR is undervaluing starting pitchers by something like 20%. That could possibly be true, given it's reliance on FIP.
   24. cardsfanboy Posted: May 31, 2014 at 03:51 PM (#4716591)
There are any number of reasonable ways to define "5th starter." But there is one clearly wrong way, which no serious analyst would ever employ: "identify a fifth starter based upon lowest era (or fip) among the top five pitchers on a team in innings pitched." You simply can't use hindsight to do it and reach any useful conclusions. (Among other things, "fifth starters" would probably have negative WAR on average, seeming to indicate that many/most teams are regularly using sub-replacement starters). This really isn't controversial at all among analysts...


Actually you should base it upon backwards looking data. The real point of my comment is that there aren't such a thing as a fifth starter for the most part. It's an evolving position over the course of a season. It's often times filled by the 7th or 8th best starting pitcher in the organization, it's not a defined role in the same way that team ace is defined. (Yes I realize that team ace isn't always consistent, but generally it is or at least debatable between two guys)


5th starters having a negative war on average is not really a problem(except the fact that once again, War undervalues starting pitchers in this day and age.)it's part of the expectation.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
JE (Jason)
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogFan Returns Home Run Ball to Ishikawa; Receives World Series tickets
(15 - 6:56pm, Oct 20)
Last: snapper (history's 42nd greatest monster)

NewsblogCalcaterra: So, if you’re not a fan of the Royals or Giants, who ya got?
(73 - 6:53pm, Oct 20)
Last: A triple short of the cycle

NewsblogOT: Monthly NBA Thread - October 2014
(271 - 6:52pm, Oct 20)
Last: Jimmy P

NewsblogOT: Politics, October 2014: Sunshine, Baseball, and Etch A Sketch: How Politicians Use Analogies
(2754 - 6:43pm, Oct 20)
Last: zenbitz

NewsblogOT: NFL/NHL thread
(8361 - 6:42pm, Oct 20)
Last: Russlan is fond of Dillon Gee

NewsblogPitch from Zito helped sell Hudson on Giants | MLB.com
(4 - 6:39pm, Oct 20)
Last: Willie Mayspedester

NewsblogMorosi: Could Cain’s story make baseball king of sports world again?
(97 - 6:24pm, Oct 20)
Last: BDC

NewsblogAngell: Gigantic
(36 - 6:22pm, Oct 20)
Last: Jolly Old St. Nick Is A Jolly Old St. Crip

NewsblogESPN: Brian Roberts retires
(22 - 6:19pm, Oct 20)
Last: Captain Supporter

NewsblogCould the Yankees ever be Royals? Young and athletic K.C. is everything that Bombers are not - NY Daily News
(23 - 6:13pm, Oct 20)
Last: You Know Nothing JT Snow (YR)

NewsblogOT: NBC.news: Valve isn’t making one gaming console, but multiple ‘Steam machines’
(841 - 6:13pm, Oct 20)
Last: DJS and the Infinite Sadness

NewsblogSielski: A friend fights for ex-Phillie Dick Allen's Hall of Fame induction
(60 - 5:53pm, Oct 20)
Last: RMc is a fine piece of cheese

NewsblogWhy Royals great Frank White no longer associates with the team whose stadium he built - Yahoo Sports
(17 - 5:35pm, Oct 20)
Last: Zach

NewsblogDealing or dueling – what’s a manager to do? | MGL on Baseball
(8 - 5:17pm, Oct 20)
Last: Walt Davis

NewsblogHitting coaches blamed for lack of offense - Sports - The Boston Globe
(12 - 5:17pm, Oct 20)
Last: RMc is a fine piece of cheese

Page rendered in 0.4579 seconds
52 querie(s) executed