The only problem is that you do have some relievers in the pen who may get over/under used. Your principle only works if the variance of the games works out so that there is an equal distribution of close games and 3-4 run games and blowouts. What if you play a stretch where 10-12 games in a row are close and tight? You can't use Brazoban and Gagne every game. Some mid-relievers have to step up from time to time and get big outs. Carrara was on the spot last night and was one pitch away from getting it done. He hung a pitch and Hall got it. The RRR has potential as a general rule, though.

Then, as I wrote above, you need to rest your top relievers.

Then, Carrara has to be able to get the job done.

Until then, it's not that Brazoban is a lock or Carrara is hopeless. It's just giving yourself the best possible chance.

Currently and throughout last year, Gagne was pretty much used 9th inning or later in the range of a tie score through up by 3. Using this criteria, he pitched 81.1 innings, which we should probably assume is at or near his maximum. If we extend that range from down by 3 to up by 3, and keep the maximum of 81.1 innings, we would be guaranteed to have a number of those situations where he could not pitch. Thus, my belief is that any proposed relief strategy should keep the same sized ranges of innings and runs.

In any case, the Dodgers already have guaranteed situations where he doesn't pitch. Unfortunately, those situations come in winnable games. Meanwhile, the Dodgers have also guaranteed he will work in games that are blowouts when he's over-rested, just to give him work.

That's what needs to change.

At least under the RRR, he would be used in a meaningful game every time. I don't see the flaw in that.

That pretty much invalidates any excuse he has for resting Gagne and Brazoban.

Here is the breakdown of Gagne's usage last season:

Standard Usage:

Up 2 8th - 2 (2 innings x1)
Up 3 8th - 2 (2 innings x2)
Tied 9th - 9 (2 innings x3, 3 innings x1)
Up 1 9th - 13
Up 2 9th - 9
Up 3 9th - 9
Tied 10th - 3 (2 innings x1)
Tied 11th - 2 (2 innings)
Up 1 11th - 1

Critical Situations:

Up 1 in the 8th - 3 (1 on, 2 out, 1.1 innings)
Up 2 in the 8th - 3 (1 on, 2 out, 1.1 innings)
Up 3 in the 8th (1 on, 2 out, 1.1 innings)
Up 4 in the 8th (2 on, 2 out, 1.1 innings)
Up 3 in the 9th (1 on, 2 out, .1 innings)
Up 4 in the 9th - 2 (2 on, 1 out, .2 innings)
Up 5 in the 9th (3 on, 2 out, .1 innings)

Getting Work:

Down 3 9th - 1
Down 6 8th - 1 (.1 innings)
Up 4 9th - 1
Up 5 9th - 2
Up 7 9th - 1
Up 9 9th - 1
Up 10 9th - 1

Standard usage:

Tied 9th or later = 26%
Up 1 in the 9th or later = 17%
Up 2 in the 9th = 11%
Up 3 in the 9th = 11%
8th inning up 2 or 3 (2 innings) = 9%

Critical situations = 16%
Getting work = 9%

So, if you want to add something, than you need to take something out. And taking out the "getting work" section isn't reasonable. With any strategic assignment, you're going to hit lulls, and I can't imagine it being much less than 10%.

1) Gagne should come into tie games in the 9th. There's no reason he should be reserved for the 10th or 11th.

2) Don't use Gagne when we are "Up 3 in the 9th". Instead save him for situations where we are "Down 1 in the 9th". Let Brazoban do the cherry picking for the easy save situations.

3) Do a better job defining critical situations and allocate more innings to them by not having Gagne pitch to start the 8th with a 2 or 3 run lead. Let Brazoban handle the 8th unless he gets into trouble.

What happens if you need to "just get the out" in the 9th and the out happens to be Derrek Lee and the only guy left in the bullpen is Elmer Dessens?

I know that the above scenario would be rare, but I imagine that having to just get the out more the once isn't, particularly in crucial intra-divisional games in September.

Standard Usage:

Up 3 in the 9th = 11%
Down 1 in the 8th = 12%
Tied 7th inning or later (only 9th or later if Gagne has already been used) = 20%
Up 1 in the 8th = 16%
Up 2 in the 8th = 12%

Critical situations = 20%
Getting work = 9%

Of course, no system is perfect. Yes, you will have situations where you have to get the out more than once. All I'm saying is - all I'm saying is - if Gagne is available, use him. If he needs rest, then rest him and hope for the best. But don't hold him back because you might need him later, and risk not using him when you could have.

If you have more than one "get the out" situation, then does it really matter what order you use the pitchers in? Really. Does it matter?

#14 - It's all about reciprocity. If you don't want to use Gagne down by three in the ninth, then don't use him up by three in the ninth. If the game is that remote from winning in the ninth when you're down, it must be that remorte from losing in the ninth when you're up. Are you reading what I wrote in the original post carefully?

Bob -- Here's hoping today's RDGC is 1968.

Aside from the freak year of 1974 when Mike Marshall would pitch in just about ANY tough situation, Alston was not one to use the pen a lot. His other relief stars, primarily Perranoski and Brewer didn't put in that many innings, but they often had 2 inning appearances.

As most of us remember, Lasorda in the 1980s would use a reliever until the guy's arm fell off (e.g. Tom Niedenfuer in 1985). Even when Jay Howell came by in 1988, he wasn't used in Eckersley-like style.

It seems that the acquisition of Todd Worrell, who was accustomed to being used in just save situations in St. Louis, made Lasorda change his style.

There really hasn't been any manager in baseball who has deviated from the usage patterns of relievers like Tracy has since Jack McKeon managed in Cincinnati and would bring in Williamson or Graves for a couple of innings as the situation dictated.

That was in 1999 and both Graves and Williams topped 110 IP.
But McKeon also had some rather mediocre starters in 1999 aside from Pete Harnisch.

While I like the RRR in principle, in practice, the problem is one of unilateral diarmament as it were. Baseball managers, are rather conservative now when it comes to changes. There is too much criticism if someone does something different.

I can't imagine a manager who thought he could change things around a lot, like a Paul Richards, being hired.

Bobby Valentine, who was one of the last managers to pass for an inconoclast out here, didn't use his bullpen much differently from anyone else.

Strangely, Japanese baseball for years operated using the "smokejumper" phenomenon. But when Kaz Sasaki started to have success for Yokohama, his manager would just use him in save situations. He had an insane season there in 1997. He had 38 saves and pitched in 49 games and had an ERA of 0.90. The next year. The next year he had 45 saves in 51 games and an ERA of 0.64.

Yokohama's manager was considered something of a rebel for using his ace reliever in just save situations.

Valentine is in Japan now and his time, Chiba, is having a good year. His closer, Kobayashi, has 16 saves. He's pitched in 18 games and has thrown 16 1/3 innings.

I just looked at the game for 1968 when it comes up.

BLECH!

I would argue that the reciprocity is based around how much an additional run changes your probability of winning, not your current probability of winning. By bringing in your best reliever, you're trying to maximize the chance the other team does not score, so conversely you should evaluate the value by the cost of the additional run.

Thus, if you would use a reliever up by 3 (an additional run puts you up by 2), you should also use him down by 1 (an additional run puts you down by 2).

We assume at the end of an inning V(x) = H(x), which is roughly accurate, though there is always a slight edge to the home team (about 3%).

We look at the following two situations to begin the 8th; V - H = 1 (Up-1) & V - H = 0 (Tied).

Our reliever can let in any number of runs (0, 1, 2, etc..) in the inning, but we'll just compare 0 runs and 1 run for simplicity.

IF, V - H was 1 and is still 1, H's and V's winning %s are equal to H(-1)% and V(1)% respectively. V(1)% = 1 - H(-1)% since the probabilities have to add up to 1.
IF, V - H was 1 and is now 0, H's and V's winning %s, H(0)% and V(0)% are both now equal to 50%.
Thus, the change in V's winning % is equal to V(1)% - 50% = 1 - H(-1)% - 50% = 50% - H(-1)%.

IF, V - H was 0 and is still 0, H's and V's winning %s, H(0)% and V(0)% are both equal to 50%.
IF, V - H was 0 and is now -1, H's winning % is now H(1)% and V's winning % is now V(-1)%.
Thus, the change in V's winning % is equal to 50% - V(-1)% = 50% - H(-1)%.

Thus, the additional run is equally critical in those two game situations. You can show the same analysis for V - H = 2 being equivalent to V - H = -1 and so forth.

Thank you for the post. I think this is really cool stuff.

Now I need to take a shower!

I still don't think Tracy's relief pitcher usage patterns is much different than any other manager.

Not that it's right. But I don't think he's an outlier in this area. He just expects the relievers to do better and they're not. And I don't know if he can adjust much with the personnel he has.

But, when the difference is consistently using Carrara in the most critical game situation instead of Gagne, you're likely going on DePodesta's list.

It could be worse for Tracy. He could have the Giants relievers to work with.

The Giants started a pitcher with an ERA over 5 (Lowry) and the three relievers after (Levine, Brower, and Walker) are also all over 5.

The DBacks are trying to "improve" their pen, but that so far has consisted of acquiring Matt Herges and Claudio Vargas.

However he was safe and Anderson hit a 3-run homer to give the Halos a 6-5 lead in the 7th.

1) He could have just let Duaner Sanchez pitch the entire inning. He got the first two outs. Why not let him get that 3rd out?

2) If you plan on using the "micro" managing strategy, I agree that you bring in Yhency Brazoban to get the final out of the inning.

Now that'll simplify your bullpen management.

Compare that to the very same "just get the out" situation in the top of the 8th. If your reliever fails then your offense has far fewer outs to work with to get the lead back.

So, in response to the question, is there really a difference between the two, I'd say yes. I mean, fundamentally, yes there is.

Am I being overly simplistic?

1) Gagne
2) Brazoban
3) Sanchez
4) Wunsch
5) Carrara
6) Alvarez
7) Erickson

Sanchez had gotten two outs and had only let one man on due to a curveball to the foot. I can somewhat see bringing Wunsch in to "get the out", because he may be better against lefties. But, if you're already playing "get the out" with Wunsch, what the heck are you doing bringing in Carrara?

Over.

Archfiend Russ Ortiz is pitching in the nightcap, which means Bob Melvin is going to have to be judicious in his use of the bullpen.

Watching a rain delay in an NCAA softball game is not an interesting sight.

He won't got the distance. The Dodgers relievers in 1968 weren't all that bad. But boy did those hitters stink.

You may get out of the fifth with the lead, but for the remaining innings you're playing Russian Roulette with your bullpen. One of 'em's got the bullet. And if one of them blows up in the 8th or 9th, then where are you?

But, in any event, back to the original question (#15), I still say, yes, it does matter. It may not matter as much as MLB managers try to make us believe it does, but it does matter. I know that this is painfully rudimentary, but outs are the clock of baseball and I don't think you can disregard that.

I'm all for using more effective relievers earlier in the game, but I think you still need to save a stud for the end. You can't blow all your ammo in the middle innings and hope the guys that are left will get it done.

Maybe this is what the fancy formulas are saying but I'll say it anyway. No matter how you slice it, the more quality arms you have in the pen, the better off you are.

Have fun guys/gals

Mike Lowell's OPS: .582