As suspected....nice to see the Clemens closers all near the bottom.

Reardon-Russell-Aquilera.

Too bad Mike Stanton doesn't qualify...

Cool list.

I wonder what the %'s are for starters in the same situations.

Just goes to show you how miscast Gagne is as a middle reliever. Restore him to his closer role, and he will triumph once more!

Doesn't this depend, like, overwhelmingly, on the proportions of 1 vs 2 vs 3-run leads that these relievers' managers hand them?

Also, what is the difference between this and raw save percentage?

Save percentage isn't just the 9th inning.

I also wonder what explains the discrepancy between a player's career saves and leads-blown. The difference can be quite large. E.g., Gossage 229 vs. 310. Quisenberry is 178 vs 244.

What saves don't meet the criteria? Saves where the pitcher did not start the 9th, saves of > 3 runs (thus, 3+ IP?), saves for games shorter or longer than 9 innings...did Gossage and Quiz have that many of these types of saves?

I must be missing something.

Well, at least Sutter invented the splitter. I mean, otherwise, it would be a complete travesty to put him in the HOF while keeping out Quiz, right?

How do the league percentages breakdown by era (or decade)?

So, Ralph Barbieri has some justification in his hating of Greg Minton.

Fibonacci not-blown points? (Not-blown leads*not-blown percentage):

Agh, that didn't come out right....

Doesn't this depend, like, overwhelmingly, on the proportions of 1 vs 2 vs 3-run leads that these relievers' managers hand them?

Not overwhelmingly. The correlation between the percentage of leads that were 1-run leads and blown lead percentage is positive but quite small (r-squared of 0.017), and the correlation between the blown lead percentage and the percentage of 3-run leads is negative but also quite small (r-squared of 0.009). Gagne and Papelbon have a very high percentage of 1-run leads (43.6% for Gagne, 42.9% for Papelbon), as did Stu Miller (45%) and Gossage (42.5%).

The average ratio, among these pitchers, is about 39% one-run leads, 34% 2-run leads, and 27% three-run leads. FWIW, Roger McDowell had the highest percentage of 3-run leads entering the ninth; Brian Fuentes the lowest.

In addition to the reason noted in comment #5, this isn't exactly raw save percentage because many of these guys (especially the pre-1990 relievers), were starting the ninth with a lead that they didn't have when they came into the game, and wound up being the winning pitcher when they didn't blow the lead.

-- MWE

I also wonder what explains the discrepancy between a player's career saves and leads-blown. The difference can be quite large. E.g., Gossage 229 vs. 310. Quisenberry is 178 vs 244.

What saves don't meet the criteria? Saves where the pitcher did not start the 9th, saves of > 3 runs (thus, 3+ IP?), saves for games shorter or longer than 9 innings...did Gossage and Quiz have that many of these types of saves?

Saves where the game was closer before the 9th, but the lead was >3 by the time the ninth started.

The other side of the coin is the difference between career blown saves and blown leads above. The additional blown saves would be leads blown before the ninth (which hurt setup guys a lot, as they can get blown saves in situations where they're not really going to get saves), and blown saves where the guy entered after the start of the ninth.

Sutter had better defensive support in Saint Louis, I think. Did that affect his BL Pctage?

Saves where the game was closer before the 9th, but the lead was >3 by the time the ninth started.

Also saves earned in extra innings.

Sutter had better defensive support in Saint Louis, I think. Did that affect his BL Pctage?

Sutter was 13/105 (12.4%) in St. Louis, 19/125 (15.2%) in Chicago, and 10/49 (20.4%) in Atlanta. Sutter was working with smaller leads on average in St. Louis, too.

I wonder how many of Rivera's where when he pitched all or part of the 8th as well

Gagne being at the top of the list is simply a product of his high peak value, when he was exceptional, and then his rapid fall from grace where he's no longer even allowed to close games. He is no longer that same pitcher, so he'll get little chance to damage his lead.

As for the list being dominated by more recent relievers, there are a variety of reasons, chief being that relivers today are brought in to start the ninth and that's it. This greatly increases their save percentage. Hiller at #2 is interesting. Very underrated. I did a scan of his baseball reference entry and it's clear this guy was pitching two to three innings every time he came in, and he was being brought into games that were either tied or even when the Tigers were behind by a run or two to keep it close. No top-flight reliever would EVER be used like that today, but they should. He pitched before I really was watching the game, but it's clear that guy was earning his money.

I wonder how many of Rivera's where when he pitched all or part of the 8th as well

Of the 446 times Rivera had a lead to start the 9th, he came into the game before the 9th in 99 of them. 13 of his 40 blown 9th-inning leads came in games in which he entered before the 9th.

Rivera has the highest before-9 entry percentages among recent relievers, as I have noted before. Contrast that to Hoffman, who hasn't come into a game before the 9th in a save situation in several years, and Nathan, who has come into the game before the ninth and carried a lead forward into the ninth just eight times in his career.

As for the list being dominated by more recent relievers, there are a variety of reasons, chief being that relievers today are brought in to start the ninth and that's it. This greatly increases their save percentage.

This isn't save percentage, though; this is blowing a lead in the ninth when you start the ninth with one. There had been several studies (one by Dave Smith at SABR a couple of years ago) which showed that percentage not moving much at all over the years.

Hiller at #2 is interesting. Very underrated.

Yes, indeed.

No top-flight reliever would EVER be used like that today, but they should.

You'd burn them out pretty quickly in today's game if you used them like that - or you'd find yourself in situations where you had to have an alternate closer more often for those occasions where the #1 guy wasn't available.

-- MWE

Mike, what I'm questioning (and this really isn't my idea, but more from the Jamesian side) is the wisdom of saving the best arm in the bullpen for use only in the ninth inning, as has become common since Dennis Eckersley defined the art of the one-inning save. Many games are lost in the seventh and eigth innings, so it would makes sense to for a manager to bring his best arm in during those situations, including when there are multiple runners on base. It might be better to bring the "closer" in during those situations, and if he can't make it all the way through the ninth inning, then the game gets turned over to a lesser pitcher to close it out, but after the fire has been put out. Maybe what I'm suggesting a blend between the John Hiller type of reliever and the K-Rod type of reliever.

Todd Jones > The Eck. Ha ha ha.

And you could use pre-free-agency pitchers that way. They can't sulk and tank, because they want a big contract in a few years, and they also can't leave in search of saves until you've grown more (and relievers usually don't last that long, anyway).

One point that sometimes gets lost in these discussions is that allowing the other team to take the lead is much worse than allowing them to tie. Like almost twice as bad.

Of course, the list is what it is. Still, if someone wanted to take this to the next level and adjust for the size of the initial lead, I'd suggest they also take into account whether the other team pulled ahead.

Nobody seems to have asked what I thought was the most obvious question (sorry if I missed it): how does this look if you somehow "control" for pitching more than one inning. On the surface, it's no surprise that a reliever pitching his first inning does a better job of not giving up runs than one entering his 2nd or 3rd.

Another possible reason for the difference between old and new is that there was much more platooning in those days and it was, generally, much easier to gain the offensive platoon advantage in the late innings.

I could have sworn Gagne blew more than 10 leads in the first half of 2008 alone.

I could have sworn Gagne blew more than 10 leads in the first half of 2008 alone.

When starting the 9th inning?

Quisenberry, D 194 16 8.2%
Gossage, R 252 23 9.1%
Sutter, B 279 42 15.1%

I REALLY wish this list was being circulated before Sutter's induction- this IS the type of thing that might have persuaded some of the knuckleheads who voted for him.

Another possible reason for the difference between old and new is that there was much more platooning in those days and it was, generally, much easier to gain the offensive platoon advantage in the late innings.

Yes. Time and again in recent years I've ground my teeth as the Giants send up a Pedro Feliz or a Mike Matheny or a Jose Castillo against a right-handed closer, but with a 4-man or a 5-man bench, you just don't have many pinch-hitting options, let alone the capacity to full-on platoon.

Looking at Hiller ... was there something funky with the save rule in 1973? 38 saves, never above 15 in any other season, but his usage pattern doesn't appear to be any different.

Looking at Hiller ... was there something funky with the save rule in 1973? 38 saves, never above 15 in any other season, but his usage pattern doesn't appear to be any different.

Nothing funky about the rule, just random noise- a spike in Save situations- kind of like K-Rod in 2008.

Nothing funky about the rule

There is something slightly funky about the rule. In 1974 it was significantly tightened (hallelujah!) but then reverted to its previous status in '75 (boo!).

Looking at Hiller ... was there something funky with the save rule in 1973? 38 saves, never above 15 in any other season, but his usage pattern doesn't appear to be any different.

More like, something funky about how Hiller was used in 1973. Under Martin (mostly) in 1973, Hiller appeared in 17 games with a one run lead, 15 with a two run lead and 9 with a three run lead, but only 8 when his team was tied (and 13 more when the Tigers were trailing). But under Houk in 1974, Hiller was used only a total of 27 times when his team was up leading by three runs or less. (11 by 1 run, 10 by 2 runs, 6 by 3 runs) but he came in 20 times when the Tigers were tied (and ten more when they were trailing). Hard to get saves if you don't come in to protect them. Of course, he pitched 25 more innings in 1974, so you can decide for yourself which season was more valuable.

Hiller and Quiz were teh awesome back in the day, so I'm happy to see them get their due (same thing to a lesser extent for Farr, Carroll and especially Steamer Stanley).

Career saves of > 1 IP:

Rollie Fingers.....201
Rich Gossage.......193
Bruce Sutter.......188
Lee Smith..........169
Dan Quisenberry....160
Jeff Reardon.......152
Sparky Lyle........134
Mike Marshall......127
Gene Garber........127
Hoyt Wilhelm.......120
Dave Righetti......108
Mariano Rivera.....107
Ron Perranoski.....107
Doug Jones.........106
Dennis Eckersley...106
Steve Bedrosian....105
Tug McGraw.........104
Kent Tekulve.......100
Stu Miller.........100

There is something slightly funky about the rule. In 1974 it was significantly tightened (hallelujah!) but then reverted to its previous status in '75 (boo!).

Not quite true.

From 1969-1973, the official rule was that any finishing pitcher who maintained a lead got a save - IOW, if you pitched the ninth inning with a 15-0 lead, you got a save. In 1974, the rule was tightened to almost what it is today, requiring that the pitcher either come in with the tying run on base or at the plate, pitch at least one full inning with a lead of three or fewer runs, or pitch three effective innings. In 1975, the rule was loosened slightly to allow a pitcher to earn a save if the tying run was on deck when the pitcher entered.

-- MWE

Mike (or anyone), do you know what rule has been retroactively applied to earlier seasons, by MacMillan and/or Forman?

Jimenez, J 109 16 14.7%
Hernandez, R 320 49 15.3%
Jackson, M 130 20 15.4%
Borowski, J 134 21 15.7%
Orosco, J 122 20 16.4%
Baez, D 127 22 17.3%
Julio, J 107 20 18.7%

I think I speak for Indians fans everywhere when I say ouch. It's like a list of our recurring nightmares, though I must say I'm shocked to see Mike Jackson and Jesse Orosco on the worst-list and Bob Wickman in the middle of the pack.

Mike, what I'm questioning (and this really isn't my idea, but more from the Jamesian side) is the wisdom of saving the best arm in the bullpen for use only in the ninth inning, as has become common since Dennis Eckersley defined the art of the one-inning save.

*Sigh*. Eckersley DIDN'T define the art of the one-inning save - Tom Henke and Gregg Olson did. What LaRussa did with Eckersley is use him exclusively with a lead - and even that was the culmination of a trend that began with guys like Reardon, Lee Smith, and Aguilera before him. But Eckersley came into a LOT of games in the eighth inning until 1994, by which time the trend toward single-inning saves that began with Henke and Olson in 1991 was well on the way to becoming universal.

Many games are lost in the seventh and eigth innings, so it would makes sense to for a manager to bring his best arm in during those situations, including when there are multiple runners on base.

It would - if the rates of games being lost in the 7th and 8th innings were going up. But they aren't.

-- MWE

It would - if the rates of games being lost in the 7th and 8th innings were going up. But they aren't.

I'm not sure that it's appropriate to expect that the impact of a change to earlier usage patterns would be a likewise return to the older levels of success and failure - you could make arguments either way.

Todd Jones > The Eck. Ha ha ha.

More Hahaaha!!

Walt: your question has been asked/answered in The Book. Do a search for:
table 93 232

I'm not sure that it's appropriate to expect that the impact of a change to earlier usage patterns would be a likewise return to the older levels of success and failure - you could make arguments either way.

Much of what I see is speculation, from people who don't like the disruptions to the flow of the game that pitching changes cause, and who want to see fewer of them. I would certainly LIKE to see fewer pitching changes. I hate it when a team runs a succession of lesser pitchers out there while the closer sits idle waiting for a lead. But what I'm trying to understand is why usage evolved in this particular way, why usage patterns have been essentially stable for the better part of the last 10-15 years, and why there is little pressure for a change in the way that pitchers (starters and relievers) are being used. I don't think it does us a lot of good to say "teams should do things differently" without understanding how they got here in the first place and the forces that led to usage patterns changing as they have. It's not all about LaRussa's ego.

-- MWE

It's not all about LaRussa's ego.

Well, not entirely.

I understand your motivations and largely agree with your approach and stance on this issue, just thought I'd nitpick one line.

Mike,

Curious if you can break this list out into its component leads (1, 2, and 3 runs). While this is rather specialized pitching data, it really ought to be supplied as part of the B-R site (some kind of special table for pitchers, if you're reading this, Sean, under the rubric "Situational Pitching"). It should look at sixth, seventh, and eighth innings as well, if for no other reason that this would produce a more complete picture of "lead protection" over the time frame you're covering (and fill in the blanks for earlier-era relievers who came into games much earlier).

It would tricky to figure out the format for this, though, because you'd have to deal with some base-out segregation. Might work best "starting inning" vs. "others."

Is there a difference in "blown lead" percentages when a closer doesn't start the ninth (i.e. comes in with men on base)? And how has the distribution of these type of appearances changed over, say, the past 5-10-15-20 years?

And just out of curiosity (and <u>only</u> if you have it handy), what is the percentage of games where:

a) the team that scores first never trails and goes on to win;
b) the lead changes hands before the sixth inning (WPCT for overtaking team);
c) the lead changes hands in the seventh and eighth innings (same WPCT breakout);
d) the overall ninth inning "blown lead" percentage?

The "win expectancy" figures as listed in Tom T. etal for trailing when entering a ninth inning (<u>The Book</u>, p. 41) vary a good bit by how many runs a team is behind. Crudely averaging that data together (both halves of the ninth) gives teams about an 11% chance of winning in what are defined as "save situations". However, teams trailing by one run have more like a 19% chance of winning. Does the "blown lead" data here refer to all such cases, including those instances where the team for whom the reliever is pitching manages to win anyway? Again, if you have it handy, how often does this scenario happen?

I think that's enough questions for now... :-)

Here's a couple of other good links, based on retrosheet data:
win expectancy, by inning, score
win expectancy, by game state

***

It's not surprise that Mo had a disproportionate number of blown saves in games he started prior to the 9th inning: those were probably really close games to begin with.

Walt: your question has been asked/answered in The Book.

Thanks, close enough, but not quite what I asked.

What I was trying to get at was to what extent the differences in blown 9th inning leads between newer and older relievers is (potentially) due to entering the game earlier. So Joe Nathan has blown 8.5% of his 9th inning leads and, given how rarely he comes in before the 9th, he might blow only 8.4% of those when he comes in for the 9th. Gossage blew 9.1% 9th inning leads but maybe he blew only 8.2% when the 9th was his first inning of work.

But what I'm trying to understand is why usage evolved in this particular way

Maybe you could take a look at one of my pet theories. I think one reason it evolved was a reaction to the heavy platooning that was going on through much of the 70s and 80s. Lefty starter, you get the righty PH late in the game, manager counters with RHP, a batter or two later, here comes the lefty PH and the LHP.

I think the main reason it evolved was fear of injury -- reduced starter workloads led to heavier reliever workloads but not that many guys could handle 100 IP of relief a year (especially for very long). That doesn't necessarily lead to the particular pattern of usage nor the stability of that pattern of course, but it gets you to big bullpens and short relievers. And, of course, I'm not sure it's actually reduced the injuries.

Many games are lost in the seventh and eigth innings, so it would makes sense to for a manager to bring his best arm in during those situations, including when there are multiple runners on base. It might be better to bring the "closer" in during those situations

I suspect that if the new trend was to use your best relief pitcher (aka "closer") in the 7th and 8th inning.......all of a sudden there'd be a lot more blown saves in the 9th inning by lower caliber relievers.

I've suspected for a long time - ever since Bill James brought up the idea of using closers in tie games - that the main reason no one does that is that the other managers would exploit it. If you're using your closer in ties (or in tight situations earlier than the ninth), then every other team can respond in kind WHEN PLAYING YOU and then work their closers normally against other teams. That takes away the advantage for you of using the closer early, while burning your closer's arm through a whole lot of innings that other closers aren't pitching. And therefore, no one wants to be the first manager to try the new concept, so no one tries it, because someone would have to be first.

- Brock Hanke

I think the main reason it evolved was fear of injury -- reduced starter workloads led to heavier reliever workloads but not that many guys could handle 100 IP of relief a year (especially for very long). That doesn't necessarily lead to the particular pattern of usage nor the stability of that pattern of course, but it gets you to big bullpens and short relievers. And, of course, I'm not sure it's actually reduced the injuries.

Entirely agreed.

Career saves of > 1 IP:

Rollie Fingers.....201
Rich Gossage.......193
Bruce Sutter.......188
Lee Smith..........169
Dan Quisenberry....160
Jeff Reardon.......152
Sparky Lyle........134
Mike Marshall......127
Gene Garber........127
Hoyt Wilhelm.......120
Dave Righetti......108
Mariano Rivera.....107
Ron Perranoski.....107
Doug Jones.........106
Dennis Eckersley...106
Steve Bedrosian....105
Tug McGraw.........104
Kent Tekulve.......100
Stu Miller.........100

Interesting how the top three on that list are the only ones in the HoF, other than Rivera and two others (Eck and Wilhelm) who were either starters, or frequently used in game situations with no save opportunity.

I think the way to test for the theory Walt is espousing is to look at the distribution of IP for starting pitchers in individual games. The descent curve away from the 8th inning should be heaviest first (say in the late 70s), continuing on into the 90s, where the 7th inning will then show the most pronounced decline. The right search query out to be able to nail it down.

Also, has someone quantified all that platooning somewhere? It's clearly got to have declined, but by how much? Some of that roster attrition has to be attributable to the third catcher, now an almost decadent luxury. Is there a simple way to measure it (number of players with more than, say, 400 PAs)?

Regarding the blown leads and multiple inning appearances, I looked at the 2008 data for games where relievers had 1.3 IP or more and came in during the seventh or eighth innings (thus garnering a "save situation"). The difference between the seventh and eighth inning was pretty stark: of the 52 instances where pitchers came in during the seventh, 39 (or 75%) resulted in blown leads. There were nine "blown wins"--which brought the overall won-loss record for these games to 20-19. Overall ERA for this group was 4.88. In the thirteen saves (which were all 2.3 IP or more), the ERA was 1.91. Pitchers in this group ("long saves") included the following: Chan Ho Park, Salomon Torres, J.P. Howell, Kyle McClellan, Joel Pineiro, Trever Miller, Jason Grilli--and, as you'd imagine, none of these games had a score differential of less than 4 runs.

Of those who came in during the eighth and pitched at least 1.3 innings, the results and the overall quality was much, much higher. The blown lead percentage was 19% (not 75%). In those 25 "blown" situations, the team won-loss record was 10-15. Overall, of course, teams did quite well with this group, recording 106 saves with an overall W-L record of 116-15. The ERA of the pitchers who had "blown" appearances was 4.23. The pitchers who recorded multiple inning saves (including, as you'd expect, a high preponderance of closers) had an ERA of 0.67.

It's a little trickier without some advanced programming queries to get the exact percentage of ninth-inning blown leads, but a fairly reasonable estimate would be right around 14%. Relievers who successfully converted saves of exactly 1 IP had an ERA of 1.07 during 2008.

While by no means an all-inclusive or exhaiustive study, it does seem to support Tom T.'s conclusions (in <u>The Book</u>, p. 230, about using closers in the eighth inning. It also suggest that Harris's suspicion may be well-founded.

I think the way to test for the theory Walt is espousing is to look at the distribution of IP for starting pitchers in individual games.

This distribution gets skewed by spot starts and by injuries. There is a lot of churn in the back end of rotations, especially in recent years. I limit this by looking at starts by Top 3 starters, distinguished by (1) a start in the team's first five games of the season and (2) at least 20 starts for the team.

Basically, Don's correct in the timing of events, although off by an inning. Until 1975, Top 3 starters were getting into the ninth inning or later about a third of the time; since then, the percentage has declined almost every year, down to 3.4% in 2008. The descent curve away from the eighth inning started in about 1995. We haven't yet seen a descent curve for the seventh inning yet, at least among the top starters.

It's a little trickier without some advanced programming queries to get the exact percentage of ninth-inning blown leads, but a fairly reasonable estimate would be right around 14%.

I'm working on this one.

-- MWE

Thanks, Mike. Yeah, I forgot about that uptick in IP and CG that happened in the 70s as a result of the DH.

You know, it might be worth doing a "dual-track" distribution on this, one for the Top 3 and another for everyone else. I suspect that these will diverge significantly enough to make for an interesting contrast.

It would also be interesting to look at some of the individual pitchers whose careers are long enough (Clemens, Maddux, Moyer...) to see how their percentages have shifted over time, and then compare those results with other long-career pitchers from earlier in the "Retroset."

Is that the complete list of relievers with 100 such opportunities since 1954? Seems like a shorter list than I would have thought.

Is that the complete list of relievers with 100 such opportunities since 1954?

It is the complete list of relievers with 100 such opportunities in the Retrosheet event file data since 1954. Some teams (notably the Cubs, Braves, and Pirates) have a lot of games pre-1970 that don't have event data, so there may be some guys who played primarily for those teams who aren't on the list.

Yeah, I forgot about that uptick in IP and CG that happened in the 70s as a result of the DH.

It didn't appear to be as a result of the DH, surprisingly - it happened in the NL, too, and it started around 1967.

-- MWE

Active pitchers with the most saves of > 1 IP

Mariano Rivera......107
Trevor Hoffman.......55
Keith Foulke.........55
Jason Isringhausen...44
Armando Benitez......42
Billy Wagner.........35
John Smoltz..........35
Todd Jones...........33
Troy Percival........32
Derek Lowe...........30
Eric Gagne...........30
Francisco Cordero....27
Byung-Hyun Kim.......26
Tom Gordon...........24
Danys Baez...........24
Dave Weathers........22
Mike Timlin..........22
J.J. Putz............22
Jonathan Papelbon....22
B.J. Ryan............21