Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Primate Studies > Discussion
Primate Studies
— Where BTF's Members Investigate the Grand Old Game

Thursday, December 11, 2008

Blown Lead Percentages, Multiple-Inning Relievers, 1954-2008

Following up on an earlier discussion, this list includes pitchers who entered the game with a lead and who pitched in more than one inning. This includes all appearances of more than one inning, and also includes appearances of one inning or less that bridged two innings - for example, a pitcher who enters a game in the seventh with two outs and the lead, gets the last out of the seventh and then leaves with one out in the eighth will be included on this list. This table shows the number of leads, the number of leads lost, and when those leads were lost (in the pitcher’s first inning, or in a later inning). Minimum 100 leads, sorted by percentage of leads lost.

Player		Leads	Lost	% Lost	First	Later	% Later
Shields, S	101	11	10.9%	9	2	18.2%
Veres, D	101	14	13.9%	11	3	21.4%
Foulke, K	116	17	14.7%	7	10	58.8%
Sullivan, S	100	15	15.0%	6	9	60.0%
Nelson, G	111	17	15.3%	8	9	52.9%
Rojas, M	139	22	15.8%	8	14	63.6%
Henneman, M	149	26	17.4%	15	11	42.3%
Eichhorn, M	120	21	17.5%	17	4	19.0%
Quantrill, P	119	21	17.6%	12	9	42.9%
Stanton, M	163	29	17.8%	11	18	62.1%
Burgmeier, T	129	23	17.8%	12	11	47.8%
Lopez, A	122	22	18.0%	13	9	40.9%
Moffett, R	109	20	18.3%	11	9	45.0%
Hernandez, W	162	30	18.5%	16	14	46.7%
Rhodes, A	107	20	18.7%	11	9	45.0%
Stoddard, T	123	23	18.7%	9	14	60.9%
Bedrosian, S	182	35	19.2%	22	13	37.1%
Niedenfuer, T	104	20	19.2%	10	10	50.0%
Quisenberry, D	244	47	19.3%	30	17	36.2%
Burke, T	109	21	19.3%	11	10	47.6%
Lefferts, C	140	27	19.3%	14	13	48.1%
Andersen, L	118	23	19.5%	14	9	39.1%
Reed, S		117	23	19.7%	13	10	43.5%
Giusti, D	121	24	19.8%	12	12	50.0%
Rivera, M	165	33	20.0%	21	12	36.4%
Miller, S	148	30	20.3%	18	12	40.0%
Bair, D		108	22	20.4%	9	13	59.1%
Brantley, J	117	24	20.5%	12	12	50.0%
Jackson, M	175	36	20.6%	18	18	50.0%
McDowell, R	202	42	20.8%	20	22	52.4%
Howe, S		120	25	20.8%	15	10	40.0%
Smith, L	251	53	21.1%	29	24	45.3%
Borbon, P	108	23	21.3%	10	13	56.5%
McGraw, T	168	36	21.4%	19	17	47.2%
Nelson, Jeff	140	30	21.4%	23	7	23.3%
Weathers, D	112	24	21.4%	20	4	16.7%
Hume, T		107	23	21.5%	10	13	56.5%
Henke, T	148	32	21.6%	17	15	46.9%
Smith, D	120	26	21.7%	15	11	42.3%
Reardon, J	220	48	21.8%	30	18	37.5%
Belinda, S	110	24	21.8%	17	7	29.2%
Assenmacher, P	128	28	21.9%	18	10	35.7%
Martinez, T	114	25	21.9%	16	9	36.0%
Myers, R	134	30	22.4%	19	11	36.7%
Crim, C		116	26	22.4%	16	10	38.5%
Eckersley, D	152	35	23.0%	22	13	37.1%
Howell, J	147	34	23.1%	23	11	32.4%
Radatz, D	129	30	23.3%	13	17	56.7%
Forster, T	140	33	23.6%	16	17	51.5%
Sutter, B	278	67	24.1%	39	28	41.8%
Reed, R		140	34	24.3%	14	20	58.8%
Carroll, C	151	37	24.5%	22	15	40.5%
Marshall, M	208	51	24.5%	26	25	49.0%
Minton, G	187	46	24.6%	20	26	56.5%
Face, R		126	31	24.6%	14	17	54.8%
Orosco, J	174	43	24.7%	28	15	34.9%
Plunk, E	133	33	24.8%	21	12	36.4%
Ward, D		124	31	25.0%	17	14	45.2%
Jones, T	100	25	25.0%	18	7	28.0%
Jones, D	189	48	25.4%	27	21	43.8%
Charlton, N	114	29	25.4%	13	16	55.2%
Thigpen, B	121	31	25.6%	18	13	41.9%
Worrell, Todd	113	29	25.7%	20	9	31.0%
Tekulve, K	218	56	25.7%	31	25	44.6%
LaRoche, D	128	33	25.8%	20	13	39.4%
Perranoski, R	166	43	25.9%	18	25	58.1%
Gossage, R	321	84	26.2%	52	32	38.1%
McMahon, D	146	39	26.7%	23	16	41.0%
Wilhelm, H	211	57	27.0%	31	26	45.6%
Montgomery, J	148	40	27.0%	31	9	22.5%
Fingers, R	319	87	27.3%	54	33	37.9%
Abernathy, T	135	37	27.4%	19	18	48.6%
Davis, R	141	39	27.7%	21	18	46.2%
Stanley, B	201	56	27.9%	31	25	44.6%
Hernandez, R	114	32	28.1%	24	8	25.0%
Plesac, D	170	48	28.2%	27	21	43.8%
Timlin, M	152	43	28.3%	32	11	25.6%
Garber, G	226	64	28.3%	25	39	60.9%
Knowles, D	155	44	28.4%	18	26	59.1%
Campbell, B	189	54	28.6%	33	21	38.9%
Kern, J		105	30	28.6%	16	14	46.7%
Wickman, B	104	30	28.8%	16	14	46.7%
Worthington, A	100	29	29.0%	8	21	72.4%
Righetti, D	175	51	29.1%	29	22	43.1%
Aker, J		120	35	29.2%	22	13	37.1%
Lyle, S		239	70	29.3%	39	31	44.3%
Franco, J	151	45	29.8%	31	14	31.1%
Locker, B	104	31	29.8%	15	16	51.6%
Sosa, E		107	32	29.9%	18	14	43.8%
Brewer, J	117	35	29.9%	19	16	45.7%
Lindblad, P	106	32	30.2%	15	17	53.1%
Harris, G	121	37	30.6%	24	13	35.1%
Clear, M	110	34	30.9%	15	19	55.9%
Gordon, T	103	32	31.1%	19	13	40.6%
Lavelle, G	171	54	31.6%	20	34	63.0%
McDaniel, L	175	57	32.6%	34	23	40.4%
Williams, Mitch	103	34	33.0%	22	12	35.3%
Hiller, J	157	55	35.0%	40	15	27.3%
Aguilera, R	112	41	36.6%	28	13	31.7%

As expected, there are almost no modern closers on this list, indicative of the trend to single-inning closers - Foulke (who wasn’t always a closer) and Rivera stand out as the exceptions.

John Hiller, who looked very good on the other list, looks very bad on this one - wonder if anyone ever though of nicknaming him “The Vulture”? Phil Regan, who did have that nickname, just missed getting onto this list (93 leads) but he’d have been pretty low on it, too.

Among all such relief appearances (including those by pitchers not on the list above), the lead was blown about a quarter of the time, with a 53/47 split between being lost in the first inning worked vs being lost in a later inning.

One thing I need to do is separate this out by closers and non-closers. This query takes so frickin’ long to run against my DB that I have to do it overnight, so it might take me a couple of days to get to it.

Mike Emeigh Posted: December 11, 2008 at 03:15 PM | 55 comment(s) Login to Bookmark
  Related News:

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. Slinger Francisco Barrios (Dr. Memory) Posted: December 11, 2008 at 04:46 PM (#3026114)
Length of lead, game state, and quality of opposition are other obvious things to correct for.
   2. Tango Posted: December 11, 2008 at 05:01 PM (#3026144)
Mike, if you post (or email) your query, I'll see if I can tweak it. It would help if you use the same schema I've got posted at my wiki, but it's not necessary.
   3. Jeff K. Posted: December 11, 2008 at 05:03 PM (#3026146)
Well, Mike noted that the size of the lead had a very low r-squared for the single inning percentage. If that's true, is there any reason to believe it has a much greater effect on multiple inning saves? When you're up by 3 in the 8th, which part of that is more important? Mike's findings would seem to point towards the fact that you're in the 8th is much more important than the fact that you're up 2 runs.

Mike, I work for UT, and we have a terminal server online that has SAS, SPSS, Matlab, uh...Minitab, and more. It's up for people to run big projects on that don't want to pay $800 for one of the stats packages. If you'd like, I could run whatever you want on that server, as long as you're explicit about how to use whichever program. I may have supported those programs when I was a student, but that doesn't mean I know how to use 'em. :)
   4. Mike Emeigh Posted: December 11, 2008 at 07:25 PM (#3026402)
Mike, if you post (or email) your query, I'll see if I can tweak it.


Won't be necessary, but thanks (Jeff K, also). I need to do some indexing, is all, and I've been just too lazy to do it.

-- MWE
   5. Jeff K. Posted: December 11, 2008 at 10:28 PM (#3026770)
Not a problem. Let me know if you'd like to run it on the server. Packages on it, in case you're curious:

Adabas
AMOS
Brio Query
Cognos
DBMS
HLM
ILOG
IMSL
Lisrel
Maple
Mathematica
Matlab
Minitab
MPlus
SAS
SPSS
Stata
SUDAAN
   6. Mike Emeigh Posted: December 13, 2008 at 04:08 AM (#3028204)
Here's the gang of 50: the guys I consider as career closers.

Aguilera, R
Aker, J
Cordero, C
Eckersley, D
Face, R
Fingers, R
Franco, J
Gagne, E
Giusti, D
Gossage, R
Graves, D
Harvey, B
Henke, T
Henneman, M
Hiller, J
Hoffman, T
Hrabosky, A
Isringhausen, J
Koch, B
Linzy, F
Lown, T
Lyle, S
Marshall, M
McGraw, T
Miller, S
Montgomery, J
Myers, R
Nathan, J
Nen, R
Percival, T
Perranoski, R
Quisenberry, D
Radatz, D
Reardon, J
Regan, P
Righetti, D
Rivera, M
Rodriguez, F
Sherry, L
Smith, L
Sutter, B
Thigpen, B
Urbina, U
Wagner, B
Wetteland, J
Wilhelm, H
Williams, Mitch
Wood, W
Worrell, Todd
Wyatt, J

The criteria are at least 300 career appearances in relief, with at least 2/3 of them in the ace relief role as best I can define it for the era. It seems to be a reasonably complete list, from what I can tell, but if I missed anyone important let me know. This is limited to 1954-2008 because I need the comprehensive event file data to do the analysis I need to do.

-- MWE
   7. Walt Davis Posted: December 14, 2008 at 01:05 AM (#3028524)
Where's Borowski? (only kidding)

I would guess, and my impression is the list supports this though it's hard to tell, that AL relievers are more likely to have multi-inning appearances, perhaps especially modern guys going 3 or fewer outs spread over a couple innings.

The best measure to look at might simply be number of batters -- they did this in the Book didn't they? OPS (or whatever) by batter faced for relievers? These days (and maybe always), I'd guess you want to yank a guy after no more than 5-6 batters. Rivera's advantage might not be greater endurance per se but that he's much more likely to retire 5-6 straight batters.
   8. Mike Emeigh Posted: December 14, 2008 at 08:40 PM (#3028761)
Based on comments I got from some folks, I'm adding Joe Hoerner, Ted Abernathy, and Lindy McDaniel to the list, and taking Wilbur Wood off of it. Wood was almost 50/50 in his career between starting and relieving and was an ace reliever for only about three years, so it's really not appropriate to put him in this grouping. I missed on Abernathy because he had two seasons that were split among multiple teams where he was actually pitching in high-leverage roles for all of them but because he pitched for more than one team my filter didn't pick them up as closer seasons. McDaniel and Hoerner had long careers as secondary relievers, but both were pitching in high-leverage roles for much of the early part of their careers and had somewhere around 400-500 appearances as ace relievers, so it's not quite right to leave them out of the discussion even though they weren't closers for a sizeable part of their careers.

-- MWE
   9. Don Malcolm Posted: December 15, 2008 at 04:02 AM (#3028927)
Mike,

Other possible additions: Eddie Fisher, Steve (Nose Candy) Howe, Clay Carroll, Steve Bedrosian, Gregg Olson, Rod Beck, Doug Jones.
   10. Mike Emeigh Posted: December 15, 2008 at 06:12 AM (#3028976)
Agree Beck should be in, and I need to take another look at Jones (he's like Hoerner and Kent Tekulve in some ways, with a long closing career and a long non-closing career which balance each other out). The other pitchers were pitching in other roles for large parts of their careers, and I have fairly high standards because I don't want to skew the results by mixing in guys who weren't always the primary go-to guy.

Fisher was only an ace reliever for part of one season with Baltimore, and part of another season in Chicago.

Howe wasn't a closer for most of his career after 1983, either - probably because of the nose candy.

Carroll and Bedrosian shared the closer role with other pitchers, and pitched in a variety of other roles in addition to closing games.

Almost all of Olson's post-Baltimore career, except for one year in Arizona, was spent in secondary relief roles.

-- MWE
   11. Steve Treder Posted: December 15, 2008 at 07:10 AM (#3028985)
Some possibilities:

- Al Worthington
- Jim Kern
- Clem Labine
   12. Mike Emeigh Posted: December 15, 2008 at 07:47 AM (#3028995)
Jones is in. I'm leaving Tekulve out, because he really wasn't a closer all that long and in Tanner's bullpen scheme he was used in a variety of roles in addition to closing relief even when he was a closer.

Labine was the Dodgers' ace reliever only from 1956 until Larry Sherry came up in 1959, which covers less than half of his career relief appearances, and he would share the load with guys like Bessent and Roebuck.

Worthington also had two careers. Until 1963 in Cincinnati, he was a utility starter/reliever who pitched very little high-leverage relief. In 1963, he replaced Jim Brosnan as the Reds' righty ace (he shared with Bill Henry, as Brosnan did), then when he went to Minnesota he took over as the Twins' ace. The ace portion of his career covers about half of his career relief appearances, and that's a little on the low side for me.

Kern was a closer for essentially three years, two in Cleveland (he shared with Dave LaRoche his first season) and one in Texas, which again represents less than half of his career.

-- MWE
   13. Don Malcolm Posted: December 15, 2008 at 08:03 AM (#3028996)
Mike,

Seems as though there was a lot more platooning of "closer-types" in the 60s/70s than from the 80s on. Since I'm not sure what the scope of your study really is, I'm just speculating on whether it's better to capture as much of the "closer-type" activity that was ambient in those years. If it's meant to study just the set that conform as closely to the more recent usage pattern, then that's fine. It's just that those innings totals for those guys (60s/70s time frame) are in the same region, and their involvement in protecting leads might well be relevant enough to include.

For potential outputs in such studies in addition to what you've been displaying thus far, let me suggest that you revisit what Doug Drinen was doing in the 99-01 BBBAs. His terms might need to be replaced with what's more current, but the basic framework should still be sound, I'd think. Plus the innings breakout that were in the reliever boxes (along with eventual team W-L records) were rather interesting (not to mention unique: I've never seen anyone else reproduce such work.) It might be worth it to reproduce some of that for the full Retroset, in addition to your planned outputs...
   14. Mike Emeigh Posted: December 15, 2008 at 08:26 AM (#3029002)
Seems as though there was a lot more platooning of "closer-types" in the 60s/70s than from the 80s on.


Not really; it was limited to just a couple of teams. By 1960, most teams had identified one guy as the primary "go-to" late-inning reliever, and it's been pretty constant since then; the definition of the role has changed, but the idea that you have a single key reliever hasn't. When teams had a role-sharing arrangement, it was usually temporary, and one guy usually took over. It was rare to find a team that shared the role for longer than a year or two.

The parameters of the key relief roles are actually fairly easy to define, for most of the 1960-2008 time frame. The two major flash points where the role changed dramatically are in the late 70s/early 80s and the early 90s. Both time periods, oddly enough, are marked by expansion and by labor unrest culminating in a strike, and it appears that both may have had an impact in the role redefinition.

-- MWE
   15. Steve Treder Posted: December 15, 2008 at 05:37 PM (#3029195)
Both time periods, oddly enough, are marked by expansion and by labor unrest culminating in a strike, and it appears that both may have had an impact in the role redefinition.

My thesis has long been that contract/risk exposure considerations were the driving force both times. As the market price of ace reliever/closers climbed, teams ratcheted up the attempt to limit the top $$ guy to a role that would minimize his injury risk. Whether that was a successful endeavor or a wise one is debatable, but I'm quite confident it's what it was about. In the case of the Cubs' changing the usage role of Bruce Sutter post-1978, the change that stimulated the first transformation, the Cubs were entirely explicit about it.
   16. Don Malcolm Posted: December 15, 2008 at 08:59 PM (#3029535)
And yet, if we look at the percentage of saves concentrated in the top man in 1960, 1970, 1980, 1990, we see that while the role had certainly been defined, there is no real change in the concentration percentage until around 1990:

<u>"SAVE CONCENTRATION" BY LEAGUE, 5 selected years (1960-2008)</u>
Year  NL Tot NL Top Pct. AL Tot AL Top Pct.
1960     213    119  55%    222     83  37%
1970     401    223  56%    467    248  53%
1980     445    240  54%    457    259  57%
1990     476    251  53%    640    446  70%
2008     607    451  74%    574    432  75%

"Top" (in "NL Top" and "AL Top") means the league total for relievers with highest saves on an individual team.

The NL's 1989 percentage for this was much higher (66%), but there was some period of retrenchment that occurred thereafter; the 1990 total takes a dip (as shown) above, with 1991-92 remaining well below the '89 percentage (54% and 60% respectively).

We can look at this another way by breaking out those league-years by levels of concentration. Let's look at all of the years (including NL for '89, '91, and '92) with nunber of teams with 70+% concentration, 50-69% concentration, and less than 50%:
Year/Lg  70+ 50-69 -50
1960 NL    2     0   6
1960 AL    1     0   7
1970 NL    1     7   4
1970 AL    0     6   6
1980 NL    1     6   5
1980 AL    5     3   6
1989 NL    4     6   2
1990 NL    1     7   4
1990 AL    6     5   3
1991 NL    4     2   6
1992 NL    5     2   5
2008 NL    8     7   1
2008 AL    8     4   2

Data like this, turned into a chart, would do a good job of mapping the changes as they occurred over time.

I think it would be a mistake to impose an economic determinism on this issue prior to 1990. It's clear that the front-office component of baseball management did not get particularly sophisticated about salaries and risk issues until at least a decade after the imposition of free agency. (Mid-80s "collusion" was not about risk issues; that came later.) Likewise, field managers didn't climb aboard this issue in a truly uniform way until sometime after 1990. While that economic determinism has clearly taken hold between 1990 and the present, trying to backfit it into the late 70s would be an overstatement.
   17. Mike Emeigh Posted: December 15, 2008 at 09:58 PM (#3029667)
And yet, if we look at the percentage of saves concentrated in the top man in 1960, 1970, 1980, 1990, we see that while the role had certainly been defined, there is no real change in the concentration percentage until around 1990


That is because the first major change - the late 70s/early 80s change, which actually started a couple of years before the Cubs did it with Sutter, I think in response to Mike Marshall's 1974 season - transferred non-save situations from the ace reliever to the rest of the bullpen. The distribution of save situations didn't change much.

The second change - the late 80s/early 90s change - changed the distribution of save situations as well as the distribution of non-save situations.

-- MWE
   18. Don Malcolm Posted: December 15, 2008 at 10:36 PM (#3029720)
Forman's distribution charts in PI for "relative score at the point closer-types enter the game" would be the place to test the assertion about Marshall. Arguing against the Marshall influence, though, is Rich Lederer's point (some one else's research that he was referencing...) that Marshall was brought into a high preponderance of games in '74 where the Dodgers were tied or behind, which doesn't sound as though it would prompt managers to eliminate non-save situations from the ace reliever.

I took a look at the data for 1985, per the presentation in #16, and both leagues have a 59% concentration, with the AL having 6 70+% concentrations (out of 14), 2 at 50-69, and 6 at -50%. The NL was 3, 5, 4--with the pennant winner, the Cardinals, in the -50% group. The AL that was totally bifurcated by division: in the East, only Detroit was using a highly concentrated save specialist (Willie Hernandez), while five of the seven AL West teams were doing so.
   19. Mike Emeigh Posted: December 15, 2008 at 10:51 PM (#3029739)
In the case of the Cubs' changing the usage role of Bruce Sutter post-1978, the change that stimulated the first transformation, the Cubs were entirely explicit about it.


You know what's odd about this? The Cubs DIDN'T change Sutter's usage pattern much at all.

-- MWE
   20. Steve Treder Posted: December 15, 2008 at 11:05 PM (#3029765)
You know what's odd about this? The Cubs DIDN'T change Sutter's usage pattern much at all.

Not as dramatically as they said they would, that's true. Herman Franks's announcement was that Sutter would only be used when the Cubs had the lead, and the reason for it was to limit his usage so that he wouldn't break down in the second half the way he had in both 1977 and '78.

But they did modify his usage. In 1977-78 he'd finished 75% of his appearances, and gotten the save 46% of the time; in 1979 he finished 90% of his appearances, and got the save 60% of the time.
   21. Mike Emeigh Posted: December 15, 2008 at 11:07 PM (#3029769)
Arguing against the Marshall influence, though, is Rich Lederer's point (some one else's research that he was referencing...) that Marshall was brought into a high preponderance of games in '74 where the Dodgers were tied or behind


Those are, by definition, non-save situations. And when Marshall was hurt and noticeably less effective the following season, I think people took note.

-- MWE
   22. Mike Emeigh Posted: December 15, 2008 at 11:24 PM (#3029786)
But they did modify his usage. In 1977-78 he'd finished 75% of his appearances, and gotten the save 46% of the time; in 1979 he finished 90% of his appearances, and got the save 60% of the time.


And yet, the timing of his appearances was virtually the same - he came into games when the team was trailing or tied about the same percentage of the time, he came into games when the score was within three runs about the same percentage of the time, he entered the game before the ninth about the same percentage of the time, and he had about the same percentage of his outings that bridged multiple innings (actually, slightly higher).

-- MWE
   23. Steve Treder Posted: December 15, 2008 at 11:32 PM (#3029797)
the timing of his appearances was virtually the same

Well, something changed. Those differences in GF rate and SV rate are fairly substantial.
   24. Mike Emeigh Posted: December 16, 2008 at 02:46 AM (#3029996)
While Sutter's distribution of innings before the ninth was virtually the same between 1978 and 1979, his distribution from the ninth inning on was very different. Sutter came into 14 games in 1978 in the ninth, and eight games in extra innings; in 1979 he entered 24 times in the ninth, just twice in extras. Furthermore, more of those post-eighth inning entries, in 1979, were with the lead.

And while the distribution of innings or number of leads didn't change much pre-ninth, the size of the leads did. Sutter was working in significantly lower-leverage situations when he came into games before the ninth inning in 1979 than he did in 1978 - with the end result being that he finished more of those games and earned more saves in those games.

-- MWE
   25. Steve Treder Posted: December 16, 2008 at 03:03 AM (#3030018)
Sutter came into 14 games in 1978 in the ninth, and eight games in extra innings; in 1979 he entered 24 times in the ninth, just twice in extras. Furthermore, more of those post-eighth inning entries, in 1979, were with the lead.

Sounds to me as though, considering not just 1978 but also 1977 in comparison with '79, the change in Sutter's usage patten was meaningful.
   26. Mike Emeigh Posted: December 16, 2008 at 03:34 AM (#3030045)
the change in Sutter's usage patten was meaningful.


It was - but what I don't know for sure is whether Sutter's usage pattern change reflects changes in the Cubs' run-scoring pattern. Something that looks like a usage pattern change is sometimes the result of a change elsewhere - you adjust how you use relievers when you no longer have the same types of opportunities for him to pitch because your offense has gotten better (or worse).

-- MWE
   27. Don Malcolm Posted: December 16, 2008 at 05:53 AM (#3030189)
OK, let's see if I can get these tables to format properly...

I think you may be overstating the impact of Marshall's failure in 1975, Mike. He was a singular figure, and was so far above the other workloads around him that it probably was not instrumental in bringing on what we have now. Looking at five year increments where pitchers with 15+ saves have 100+ IP, we can see that the downward trend didn't reach "terminal velocity" until 1989-93:
Year           No     Rate
54-58           8      .10
59-63          15      .17
64-68          29      .29
69-73          32      .27
74-78          38      .31
79-83          32      .25
84-88          25      .19
89-93           5      .04
94-98           0      .00
99-03           2      .01
04-08           0      .00
The "rate" is number of pitchers (15+ SV) per team per league with 100+ IP. Weird-looking ratio, I know: if you multiply by ten (number of such releivers per ten teams per year) it might make more sense...

15 relievers met these criteria in 1977, which was the highest single-season incidence in history. Totals for subsequent years: 1978, 6; 1979, 7; 1980, 7; 1981, N/A; 1982, 10; 1983, 8; 1984, 8; 1985, 7; 1986, 5, 1987, 4; 1988, 1; 1989, 2; 1990, 0; 1991, 2; 1992, 1, 1993-98, 0; 1999, 2; 2000-08, 0.

So the "100+ IP ace reliever" doesn't start to go extinct until 1987-88.

As for Sutter's usage patterns: while Sutter was injured during his amazing '77 season (missed most of August), it's not clear that the Cubs really changed his usage pattern all that much in terms of strategy. His multiple inning apperances actually increased over the three years (37 in '77, 41 in both '78 and '79). Using Forman's innings matrix, we can see that the usage patterns for each year are not dramatically different:
Year    4+  1-3  tied  beh  KS%
1977    8%  63%   19%  10%  82%
1978    9%  64%   24%   3%  88%
1979    5%  73%   15%   7%  88%
KS means "Key Situations" and covers save situations and tied games.

While '79 is clearly moving in the direction of "post-modern" usage, it's not anywhere near the 90-95% values in the "1-3" range that we're seeing today.

Sutter's IP in '79 is down a bit (about 6 IP) from '77. Whitey Herzog didn't back off that workload in subsequent years: Sutter's highest IP total comes in 1984, when he set what was then the save record (45). And the '85 game logs seem to indicate that Eddie Haas may have pushed him over the edge that June when he gave Sutter three consecutive appearances in five days where he faced 10+ batters. Sutter's ERA was 5.44 from that point to the end of the year, and he was never the same after that.

Rich Lederer quotes Bill James as saying that the idea of lowering closer's workloads began in the late 70s, but it appears that it doesn't really start to happen until about more like the mid-to-late 80s. If you have other measures that support James's interpretation, Mike, I'm more than willing to be corrected.
   28. Steve Treder Posted: December 16, 2008 at 06:06 AM (#3030197)
While '79 is clearly moving in the direction of "post-modern" usage, it's not anywhere near the 90-95% values in the "1-3" range that we're seeing today.

Well, duh. In regarding Sutter's usage pattern differentiation between 1978 and '79, it would seem that whatever we're seeing today isn't particularly relevant.
   29. Steve Treder Posted: December 16, 2008 at 06:11 AM (#3030199)
From a series of articles I wrote in 2005:

Top Quintile Win Plus Save Producer, 1960-1972:

Year G IP IP/G W L Dec. Sv Sv/G Sv/T Sv ERA+
1960 68 113 1.67 11 6 16.3 21 31.5% 74.4% 159
1961 58 108 1.86 12 5 17.3 20 34.5% 56.3% 154
1962 62 99 1.60 8 6 13.5 23 36.4% 53.3% 160
1963 66 115 1.76 12 6 18.3 24 36.1% 72.0% 163
1964 69 119 1.72 11 7 17.8 25 36.6% 65.6% 161
1965 77 130 1.69 10 7 16.5 27 35.2% 63.9% 148
1966 62 99 1.60 9 4 12.5 24 38.6% 61.9% 153
1967 70 112 1.60 9 6 14.8 24 34.1% 56.5% 176
1968 65 114 1.75 10 7 16.0 19 28.7% 52.8% 160
1969 66 106 1.61 7 6 13.0 27 41.3% 67.3% 155
1970 65 99 1.53 7 6 13.0 30 46.7% 60.6% 140
1971 66 105 1.58 8 8 15.8 24 36.4% 69.1% 143
1972 64 110 1.72 9 6 14.8 31 47.9% 71.2% 166
Avg 66 110 1.67 9 6 15.3 24 37.2% 63.5% 157

Top Quintile Win Plus Save Producer, 1973-1985:

Year G IP IP/G W L Dec. Sv Sv/G Sv/T Sv ERA+
1973 67 121 1.80 9 7 15.2 28 42.1% 68.1% 163
1974 74 147 1.99 12 10 21.4 19 25.9% 80.7% 141
1975 66 113 1.71 9 6 15.8 22 32.6% 54.3% 161
1976 72 131 1.82 12 7 19.4 21 29.7% 66.0% 146
1977 71 130 1.84 10 7 17.4 30 42.2% 74.1% 205
1978 70 115 1.64 8 9 17.4 30 42.3% 69.2% 156
1979 76 130 1.72 10 8 17.6 30 39.7% 72.1% 188
1980 67 109 1.62 10 7 16.4 28 41.7% 69.0% 131
1981 72 116 1.61 9 7 15.3 33 45.2% 73.2% 181
1982 72 115 1.61 10 8 17.2 31 43.7% 69.2% 165
1983 65 110 1.70 8 6 13.8 31 47.8% 76.4% 164
1984 72 118 1.64 8 5 13.0 38 52.8% 77.2% 166
1985 71 106 1.50 7 7 14.4 34 48.5% 76.8% 152
Avg 70 120 1.71 9 7 16.5 29 41.0% 71.2% 163


Top Quintile Win Plus Save Producer, 1986-1992:

Year G IP IP/G W L Dec. Sv Sv/G Sv/T Sv ERA+
1986 68 94 1.38 8 9 16.4 36 53.2% 77.4% 141
1987 66 89 1.35 7 6 12.6 34 52.1% 73.2% 148
1988 62 78 1.26 4 4 8.6 39 63.3% 78.5% 167
1989 65 77 1.18 5 4 9.2 37 56.0% 74.4% 187
1990 65 78 1.19 5 4 9.0 44 67.1% 82.9% 255
1991 67 77 1.16 6 4 10.0 42 62.5% 87.4% 177
1992 70 83 1.19 6 6 11.2 42 59.4% 83.5% 150
Avg 66 82 1.24 6 5 11.0 39 59.0% 79.6% 175

Top Quintile Win Plus Save Producer, 1993-2004:

Year G IP IP/G W L Dec. Sv Sv/G Sv/T Sv ERA+
1993 70 78 1.12 4 4 7.5 47 66.7% 89.1% 212
1994 66 71 1.08 3 6 8.8 40 61.4% 77.6% 154
1995 63 63 0.99 3 3 6.6 41 65.4% 90.6% 180
1996 69 76 1.09 5 4 9.0 41 59.4% 86.1% 173
1997 70 75 1.07 6 3 9.0 40 56.8% 83.6% 179
1998 71 75 1.05 4 4 8.5 46 64.0% 88.4% 201
1999 67 71 1.06 4 3 7.0 42 62.6% 90.6% 192
2000 71 73 1.04 4 5 8.5 42 60.0% 90.7% 172
2001 72 75 1.04 4 6 9.5 44 61.3% 90.2% 143
2002 73 78 1.07 5 4 8.5 48 65.2% 95.6% 151
2003 70 76 1.09 3 3 6.2 45 64.0% 89.0% 264
2004 70 75 1.07 4 3 6.7 47 67.8% 89.3% 208
Avg 69 74 1.06 4 4 8.0 44 62.9% 88.4% 186
   30. Steve Treder Posted: December 16, 2008 at 06:15 AM (#3030201)
Oh, hell. The formatting is teh ####.

Hopefully the big time breakpoints of early '80s, late '80s, and early '90s are visible.
   31. Don Malcolm Posted: December 16, 2008 at 06:25 AM (#3030209)
Good data, Steve. The '73 and '74 numbers are skewed somewhat by Marshall, of course.

IP and SV/T values would tend to fix the flashpoint as being between 1985-90. Innings nosedive first, and targeted save usage follows.

What's relevant is that Sutter's usage pattern in '77 and '78 appears to be <u>below</u> the average distribution for top relievers.
   32. Steve Treder Posted: December 16, 2008 at 06:35 AM (#3030214)
What's relevant is that Sutter's usage pattern in '77 and '78 appears to be below the average distribution for top relievers.

Well, yeah. The salient issue is that Sutter got hurt and missed significant time in the second half in both '77 and (to a lesser degree missed time, but his performance tanked) '78.
   33. OCF Posted: December 16, 2008 at 06:46 AM (#3030223)
I took a look at the data for 1985, per the presentation in #16, and both leagues have a 59% concentration, with the AL having 6 70+% concentrations (out of 14), 2 at 50-69, and 6 at -50%. The NL was 3, 5, 4--with the pennant winner, the Cardinals, in the -50% group.

Just a historical reminder of the weirdness that was the 1985 Cardinals and why they show up that way:

In the 1984-85 offseason, Bruce Sutter left as a free agent, signing with the Braves. The general consensus of punditry was that the Cardinals, in 3rd place in 1984, were doomed by that to having no chance at the division. But the 1985 team had one of the all-time great defenses, with Pendelton and Ozzie on the left side of the infield and three CF-capable speedsters in the outfield. (Say what you will about replacing Lonnie Smith with Vince Coleman - it was at least a defensive upgrade.) In a HR-suppressing ballpark, Whitey took a pitch-to-contact style starting staff and urged them to challenge hitters and use that defense. The key off-season acquisition of John Tudor worked out amazingly well.

Rather than worrying about replacing Sutter as ace/closer, the team settled into designating a L/R pair of co-aces: Ken Dayley and Jeff Lahti, neither of whom had distnguished records before that year. But with the starting staff able to pitch deep into games (the defense had a lot to do with that), including an above-average number of complete games, it turned out that neither Lahti nor Dayley was worked all that hard, and both were reasonably effective.

The words "bullpen by committee" were used for that team. It turns out, as it often does, that those words were merely the name of a transitional state. Todd Worrell got called up as a rookie late in the season, and even by the end of the season, it was clear that Worrell would take over (or perhaps had already taken over) the job of closer.
   34. Don Malcolm Posted: December 16, 2008 at 07:34 AM (#3030240)
OCF: Good stuff on the Cards. I think the "committee" approach would have some affinities with what Bill James is proposing (though Bill would have you believe it's some radical strategic innovation he's pushing, which clearly seems overstated). If you can develop a relief staff in that way, you'd have depth to use the best person avaoiable for the specific situation that arises in a game in the seventh and eighth.

I would suspect that Mike's files contain examples of upwards of a dozen effective "committees" over the time frame, several of which were on post-season teams (the '69 Mets immediately come to mind, and several of Jim Leyland's Pirates teams in the early nineties featured committees).

Steve: I think what happened to Sutter in '78 was that Franks kept trying to get 3-inning saves out of him. He had seven such outings (at least 2.2 IP) through mid-August, then #8 (August 19) was the tipping point. His ERA was 1.91 on 8/15; beginning with that eighth "long save" (there was a ninth, that was equally disastrous, in September), his ERA was 7.33 for the rest of the year.

They backed off that somewhat in '79: only five such "long outings"--though there was still a five-inning stint in mid-May. Incredibly, they trotted him out there the next day with no day's rest! Amazingly, he fanned 4 guys in 1.2 IP and got the save. Maybe he was so tired the split-finger worked even better. :-)
   35. OCF Posted: December 16, 2008 at 05:49 PM (#3030525)
On another Cardinal team, the '67-'68 team of my first fandom: Joe Hoerner and Ron Willis were another L-R pair of co-aces. Only I think Hoerner, the lefty, was always a little more trusted than Willis. The rules for managing that staff included "Gibson finishes his own games." The staff had two finishers in Hoerner and Willis, a garbageman or two, and anyone else in the bullpen was a swingman who would sometimes be called on to start games. A significant part of the story of the '67 season was the emergence of such swingmen (including Briles and Hughes) into the starting rotation as injuries would require it.

Most of the "bullpen by committee" examples I can think of were short-term measures. The state seems inherently unstable, as does "co-ace." And cases in which a setup man's star shines more brightly than the closer, like Francisco Rodriguez in front of Percival in '02-'03, or Rivera in front of Wetteland in '96, also seem to be an unstable state.
   36. Mike Emeigh Posted: December 16, 2008 at 06:03 PM (#3030544)
Most of the "bullpen by committee" examples I can think of were short-term measures. The state seems inherently unstable, as does "co-ace."


They have been since time immemorial (well, since 1960, anyway). Eventually, one pitcher emerges as the star around whom everyone else in the pen revolves. There are some unusually powerful influences driving teams toward designating one go-to guy (and they were present before the one that most people usually credit, the save rule, existed).

-- MWE
   37. Don Malcolm Posted: December 16, 2008 at 10:49 PM (#3031014)
You're absolutely right, Mike: such powerful influences exist--and it's an impulse, obviously, that transcends baseball. The question that we are hoping your work can illuminate is whether the "go-to guy" can be used earlier in the game (which is what James is suggesting, with less emphasis on the "committee" concept) without somehow sacrificing his spot on the totem pole. As you allude, it's possible that such a question involves too many intangible elements to be answered empirically.

That's why I was suggesting a look back at how Doug was organizing some of the reliever data in the BBBAs, to see if it might be useful in determining the inherent "fungibility" of +!, +2, +3-run leads in the sixth, seventh, eighth innings in addition to the ninth.

The inherent "instability" is more psychological than empirical--but sometimes the causality arrows work that way, and the psychological becomes "empirical."

One other factor is the free agent era probably encourages this type of definition, because marginalizing the "non-ace" would tend to keep salaries down and create a job scramble for those relievers who never get to put on the equivalent of the "green jacket."
   38. Mike Emeigh Posted: December 16, 2008 at 11:28 PM (#3031067)
The question that we are hoping your work can illuminate is whether the "go-to guy" can be used earlier in the game (which is what James is suggesting, with less emphasis on the "committee" concept) without somehow sacrificing his spot on the totem pole.


My tentative answer to that is "no" - and it comes back to where this thread started. Relievers who work multiple innings - even in the old days, when they were often expected to work multiple innings - tended to lose a lot of leads. So if you conclude from that data point that you're going to work relievers one inning at a time, the last one is the one who is going to be in the highest-leverage situation most of the time.

-- MWE
   39. StillFlash Posted: December 17, 2008 at 01:48 AM (#3031192)
I looked back at Dave Giusti (R) and Ramon Hernandez (L).

In 1972, Giusti had 54 G, 44 GF, 22 SV. Hernandez 53 G, 31 GF, 14 SV
In 1973, Giusti had 67 G, 60 GF, 20 SV. Hernandez 59 G, 33 GF, 11 SV.

For the two seasons, Giusti had a higher pct of games that he finished, but the pct of GF that resulted in SVs was just about the same. Hernandez did setup, Giusti usually did not. But, when it came to the 9th inning, which one pitched was usually decided by the best batter matchups.
   40. Mike Emeigh Posted: December 17, 2008 at 03:03 AM (#3031246)
But, when it came to the 9th inning, which one pitched was usually decided by the best batter matchups.


Not exactly. Giusti was the clear #1 guy. This was the era of the "if you finish the game with a lead you get a save" rule (it wasn't changed until 1974), and Hernandez, rather than Giusti, usually finished out the not-so-close games and got saves, while Giusti finished the close ones. (This is still the rule today; quite often the eighth-inning guy will finish out a game when the closer wouldn't qualify for a save if he came in.)

-- MWE
   41. Don Malcolm Posted: December 17, 2008 at 09:58 AM (#3031437)
The usage patterns in the 70s, both in terms of when people came into the game, and how long they pitched, are just all over the map in comparison with today. While Giusti was "closer to being a closer than Ramon," they both got a lot of long-inning assignments mixed in (8 appearances of at least 2.1 IP in each season; Giusti had 5 in '72 and 11 in '73) and came in a lot more often in the seventh inning (27 times vs. 14 for Giusti in 72-72).Giusti came into the ninth a good bit more often (54 times vs. 29 for Ramon). They were almost identical in terms of eighth inning appearances (41 for Ramon, 40 for Giusti).

These were also the days of the swing man--Pirates basically had three of them on each of these teams (Kison in '72, injured in '73; Luke Walker both years, Bob Johnson both years, and Jim Rooker in '73). There's a sequence in '72 where Walker starts a game, gets knocked out in the fourth, and the next day throws three innings in relief. That stuff used to happen with some frequency in the 50s and 60s. If that happened today, though, the bloggers would be spilling blood from coast to coast. :-)

Still, it's a more versatile use of two guys in the bullpen than we see now. The only way we really get two guys with even fairly close save totals these days is due to injury or serious ineffectiveness.
   42. Don Malcolm Posted: December 17, 2008 at 10:02 AM (#3031439)
Sorry, that first paragraph got messed up. (Live preview doesn't quite save some of us from messing up these sentences or getting those tables to format correctly...)

Should read (Hernandez has 8 appearances of at least 2.1 IP in each year, Giusti had 5 in '72 and 11 in '73) and "Ramon came in a lot more often in the seventh inning..."
   43. StillFlash Posted: December 17, 2008 at 01:14 PM (#3031478)
Rooker, acquired in an off-season trade, started the 1973 season in the bullpen and did quite well, much better than his previous year in KC. When Steve Blass was removed from the rotation after June 1, Rooker took his place and stayed for 6+ years.

Bob Moose also moved in and out of the rotation a lot.

THere were normally 5 men in the pen. Two late inning guys, 2 swing men for the middle innings, and a mop-up guy.

Giustii twice worked 100+ innings. He was more likely to finish than Hernandez. But under Virdon, in 72-73, pct of GF that were SVs were almost identical for the two. When Murtaugh came back in 74, Hernandez SV% dropped way down. I will have to look it up, but my memory is Virdon tended to use matchups right to the end of the game, as Hernandez was exceptionally tough on lefties.
   44. Mike Emeigh Posted: December 17, 2008 at 03:45 PM (#3031572)
When Murtaugh came back in 74, Hernandez SV% dropped way down.


As I noted before, this is primarily because the save rule changed in 1974. Hernandez got saves in both 72 and 73 that would not have been saves in 74.

This is why I use leverage, not save totals, to evaluate how relievers are being used.

-- MWE
   45. Mike Emeigh Posted: December 17, 2008 at 03:57 PM (#3031579)
Expanding on my last comment:

The changes in the save rule from 1973 to 1974 to 1975 are not matched by changes in the leverage that relievers faced when they came into the game; there does not appear to be a significant change in leverage across that time frame. If you use save totals to evaluate relievers, you're going to run into problems in that time period, problems which I don't think you can properly address by manual adjustments. For the purposes of what I want to do, I need a measure that is reasonably consistent over time. Leverage - adjusted for run environment, as both Studes and Tango do - fits that bill nicely.

-- MWE
   46. Don Malcolm Posted: December 17, 2008 at 06:50 PM (#3031745)
Mike, I think that makes a lot of sense for making the discrimination--when we look at '72, using Forman's leverage calcs in the gamelogs, we see Giusti with 36 "high levarage" IP and Ramon with only 15.

I'm trying to reconcile those numbers, though, with the fact that LI is usually measured against every plate appearance (per Tom T.'s charts), and Sean's numbers seem to sum up the leverage for an appearance. Does it matter much which way we characterize it? In the splits, Sean seems to be adding it up by the plate appearance. I think that's valid as a general performance measure, but it does lack actual "result context," which an innings aproach would provide.

Looking at it from oveall appearances added up by innings, we can see that in the highest leverage appearances, Giusti has three wins, four losses, twelve saves, three blown saves, 6-25 in IS/IR, with an overall ERA of 2.00. Hernandez has six saves, two holds and a win, 0-9 IS/IR, and an ERA of 0.00.

I'd like to Sean add the innings approach to the results to augment the plate appearance approach, though I suspect for your purposes the latter is sufficient, yes?
   47. Mike Emeigh Posted: December 18, 2008 at 10:44 PM (#3033141)
Actually, I'm working on it looking at two things:

1. leverage when the pitcher comes into the game;
2. leverage at the start of the ninth inning.

Using any sort of summary of PAs is problematic because pitchers can create their own leverage by allowing baserunners, and that's not exactly what we want to evaluate :)

-- MWE
   48. Don Malcolm Posted: December 19, 2008 at 03:33 AM (#3033470)
Mike,

I made another cross-comp between Tom T.'s charts and Sean's gamelogs (Giusti's '72 season in particular) and I'm virtually certain that what Sean is showing is the LI at the point in which the reliever enters the game. He's not adding or subtracting or modifying anything WRT to the leverage, as I had first surmised.

That would mean that an IP-based summation would not be vulnerable to "self-leveraging," and that we can report the results (wins, losses, saves, holds, inherited runners) in the form I supplied in #46.

I think it would be interesting to create an output by team that shows the distribution of relief innings by leverage (high/very high, medium, low).

It would also be interesting to break those out by the inning appeared in, as there can obviously be some high leverage situations prior to the ninth.

Looking forward to your next product in the research process.
   49. Mike Emeigh Posted: December 19, 2008 at 03:40 AM (#3033478)
I think it would be interesting to create an output by team that shows the distribution of relief innings by leverage (high/very high, medium, low).


I presented some of this info at NCSSORS (at Menlo College) in October. You shoulda been there :)

-- MWE
   50. Mike Emeigh Posted: December 19, 2008 at 03:47 AM (#3033485)
The other thing I should note is that I am using a slightly different calculation for LI than does Sean. I'm using Dave Studenmund's "poor man's LI" with park adjustments (which are necessary IMO). Sean is using Tango's more accurate calculations, but I don't know whether he's using the park adjustments. LI is different in 1996 Coors Field (7.7 RPG) than in 1968 Dodger Stadium (2.60 RPG). If you don't adjust for the run-scoring environment you underestimate the leverage in low-scoring environments and overestimate it in high-scoring environments.

-- MWE
   51. Paul Wendt Posted: January 25, 2009 at 07:45 PM (#3060020)
30. Steve Treder Posted: December 16, 2008 at 12:15 AM (#3030201)
Oh, hell. The formatting is teh ####.

Hopefully the big time breakpoints of early '80s, late '80s, and early '90s are visible.


Not really.
A lot of this thread cries for graphical treatment, not merely typographical. Typographically I believe more can be done with some boldface highlights than with straightening and spacing the columns.


8.
McDaniel and Hoerner had long careers as secondary relievers, but both were pitching in high-leverage roles for much of the early part of their careers and had somewhere around 400-500 appearances as ace relievers, so it's not quite right to leave them out of the discussion even though they weren't closers for a sizeable part of their careers.

-- MWE<i>

13. Don Malcolm Posted: December 15, 2008 at 02:03 AM (#3028996)
<i>Mike,

Seems as though there was a lot more platooning of "closer-types" in the 60s/70s than from the 80s on. Since I'm not sure what the scope of your study really is, I'm just speculating on whether it's better to capture as much of the "closer-type" activity that was ambient in those years.



What is the big picture?
Why refine the "career closers" label and work on the careers of "career closers" rather than continue along lines of the table in the preface, which matches the title of the thread?

What background makes it right or wrong (#8) to include so and so?
(It does't seem to be fairness to the pitchers but something more analytical and abstract. See the points in favor of including Hoerner and Abernathy despite their careers.)

Beyond this point the thread seems concerned chiefly with the history of relief pitching rather than with the situational career or whole career performances of particular pitchers.
   52. Paul Wendt Posted: January 25, 2009 at 07:58 PM (#3060023)
I presented some of this info at NCSSORS (at Menlo College) in October. You shoulda been there :)

I knew it what it must stand for but I was wrong, not North Carolina but Northern California.
http://ncssors.wikidot.com/
Is it continuing?

You may be interested in skimming the website for the NESSIS (New England) 2007.
http://www.amstat.org/chapters/boston/nessis07.html
The organizers planned be semiannual. I don't know whether that is still expected.
   53. Paul Wendt Posted: January 26, 2009 at 05:52 PM (#3060563)
Among all such relief appearances (including those by pitchers not on the list above), the lead was blown about a quarter of the time, with a 53/47 split between being lost in the first inning worked vs being lost in a later inning.

One thing I need to do is separate this out by closers and non-closers. This query takes so frickin’ long to run against my DB that I have to do it overnight, so it might take me a couple of days to get to it.


The 53:47 split doesn't impress me, it is too dull to cut the complex knot.
Our featured pitchers may blow more leads in their first innings of work ("the 8th") because there are runners on base when they enter or because their own teams score meanwhile. They may blow more leads in their second innings ("the 9th") because there are always no outs when they begin the 9th, because they yield parts of multiple run leads in the 8th, or because they are less effective in the 9th (tired or stiff).

A useful baseline may be aggregate pitching performances in the 8th and in the 9th, by some measure such as OPS+ that does not depend on the score or the lead in runs. (Use the lead only to restrict which entire relief appearances are in the sample.)

Today managers give a lot of attention to work days and rest days. Perhaps the "work situation" at least for a closer or ace, may usefully be classified by the pattern of preceding work and rest days.

For a two-day history,
00 = two rest days
10 = rest yesterday, one inning before
02 = two innings yesterday, rest before
and so on

What I have in mind here is joint quantification of the inning-to-inning effects and the day-to-day effects. The inning-to-inning representation is simple (first and second of two innings work), the "effect" is simple (pitcher OPS+, for instance), and the day-to-day representation is complex because I recalled snippets of such talk a few times yesterday evening and overnight. Pap(elbon) is getting some work after two days off (00), or getting the day off after two days working (usually 11).

What I have in mind, in another sense, is that there is much to be learned from aggregate studies, no doubt comprehending some situations too infrequent to support any statistical knowledge of a particular pitcher.
   54. Paul Wendt Posted: January 26, 2009 at 08:04 PM (#3060697)
Here is representation of the games log for John Hiller 1973. For human convenience one line represents one month, April to September. Hyphens mark days prior to the opening game but it would be reasonable to provide a work history for very early season games by reference to exhibition games.

pitching games log, John Hiller 1973
------000010000010010000010100
0001100011001001010100101101000
100010000101000000000110012101 ; '2' means both of two games played
1110000101010000100101000110011 ; 8 days, 8 pitcher games, 8 saves
0101101100101000001010001000100
111100011010011000101000000101

That log is equivalent to 'DR' or days rest that Sean Forman provides in daily logs for pitchers (except in treatment of the first game played).

Here is a parallel version of the innings log following Mike Emeigh's count for innings: 2 innings means parts of two half-innings.

innings log, John Hiller 1973
------000010000020010000010100
0002200011006002010400102102000
300040000204000000000210012102 ; 8 days, 8 pitcher games, 8 saves
3220000303030000100109000140013
0106202100203000004050001000500
232100041010021000104000000203

That log is sufficient to derive this work/rest history for every day pitched. In the table I give the two-day and three-day innings histories (ih2, ih3) for every pitcher game during April and May only --before Hiller 1973 gets interesting but sufficient for illustration.

BF,R    IP    DR    ih2    ih3
(beginning Apr 11 after Detroit opening day Apr 7)
1,0    0.1    1    00    000
7
,0    2    2    00    000
2
,0    0.1    1    00    200
1
,0    0.1    1    00    000
5
,3    0.2    1    10    010
6
,0    1.2    2    00    000
5
,0    1.2    2    02    002
2
,0    1    1    00    000
5
,1    0.2    1    01    001
22
,*    5.1    6    00    100
5
,0     1.2    2    00    600
3
,0    0.2    1    20    020
12
,0    3.1    4    10    010
2
,0    0.1    1    00    400
10
,0    2    2    10    010
1
,0    0.1    1    02    102
6
,#    1.1    2    10    210
(ending May 28 before three days rest


BF,R - official batters faced and runs allowed
IP - official innings pitched
DR - days rest
ih2 - pitcher innings on each of the two previous days (Emeigh's innings, not official IP)
ih3 - pitcher innings on each of the three previous days


Notes that are beside the methodological point.

*
May 13, two runs in six pitcher innings, extra-innings defeat for Detroit and Hiller. ERA down from 4.15 to 3.85, never to return.

During the next eight weeks he put up strings of 10 and 11 shutout pitcher games around a 1-run, 4-inning effort. The string of 8 saves in 8 days (underline) made a second 10-game string. Between May 13 and Jul 10 he worked 22 pitcher games, 36 innings pitched (ip, official), 45 pitcher innings (pi, following Mike Emeigh), one run. ERA 1.38 after the string ended Jul 10, season final ERA 1.44.)

#
May 28, Hiller's ninth save including only three of his eight 1/3- and 2/3-inning games.

All three 1/3-inning stints were games finished, three defeats for Detroit (scores 1-3, 5-6, 5-6) and two win-saves. None of the three defeats were losses charged for Hiller.
(The short finishing efforts may include walkoff losses for Detroit and "blown ties" for Hiller. I haven't checked. Some of his longer efforts are too interesting for that, given that it is all beside the methodological point. For example see his four blown saves, all later in the season. Another day he pitched innings 2 to 10, one run, one loss.)
   55. Paul Wendt Posted: January 26, 2009 at 08:29 PM (#3060717)
BF, R, IP, and DR are published at Baseball-Reference in pitcher game logs, linked to player pages for the retrosheet era.
PI, pitcher innings is available there indirectly in the [Pitcher] variable, a codes such as '9-14f' for "innings 9 to 14, game finished".

I have always rebelled at either G or GP, the latter short for "games pitched". The meaning is pitching games, parallel to batting games and fielding games. Conventionally we use G recurrently, in pitching, batting, and fielding tables, although I have seen GP in pitching tables.

Now Mike Emeigh has argued for paying some attention to a count of pitcher innings that is parallel to pitcher games (commonly "games pitched"), where official innings pitched is not parallel in meaning.

In retrospect I would rewrite the preceding articles using the symbols and terms that I barely introduced at the end in bold.

pG - pitching or pitcher games, the number of games in which someone works as a pitcher; equivalent to fielding games at the pitcher position
pI - pitching or pitcher innings, the number of innings in which someone worked as a pitcher

IP - innings pitched, the official statistic of course

I don't like it a lot but it seems the best way to go given that IP and innings pitched are firmly established.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Sponsor

Support BBTF

donate

Thanks to
Mike Emeigh
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Demarini, Easton and TPX Baseball Bats

 

 

 

AllianceTickets.com has cheap MLB Tickets. Get all your Colorado Rockies Tickets, Seattle Mariners Tickets, San Francisco Giants Tickets and all your favorite baseball tickets here. We also carry cheap Denver Broncos Tickets, Seattle Seahawks Tickets and Denver Nuggets Tickets.

For wholesale prices on baseball gifts and equipment, check these stores out!

Baseball Autograph Signings
Baseball Card Supplies
Baseball Memorabilia
Baseball Collectibles
Baseball Equipment
Baseball Protective Gear

Page rendered in 0.5979 seconds
67 querie(s) executed