THE BOOK--Playing The Percentages In Baseball

I agree, this is a tough issue. I was trying to due some research to see if you can give tangible credit for batters for taking more pitches (therefore helping to knock starters out of the game) by looking at their P/PA, as well as the difference in performance between starters and relievers. However, IIRC, over the past few years the average ERA for starting pitchers has been invariably around ~4.50, whereas relievers are at ~4.00. Now, I know this has a lot to do with sample sizes and favorable matchups, but if relievers are simply performing better than starters, and since a manager is able to mix and match with relievers (which he can’t do with starters) is it that much better to work the starter to get to the bullpen?

Of course, a lot of this is situational. I want to bat against Brad Penny as much as possible as compared to Ramirez, Okajima, Bard, and Papelbon...but I’d rather face Sean Green and Brian Stokes than Johan Santana.

I remember once reading that relievers must be worse than starting pitchers because, in the natural selection of baseball, everyone begins as a starter when they’re younger, and the best guys stick and the rest are moved to the pen. But maybe that’s not true. Maybe guys with less fatigue issues stay in the rotation, whereas guys that have electric stuff that fades after a few innings is moved to the pen. Also, I think a lot has to do with preconceived notations of what people think starters and relievers “look like.” Joba Chamberlain “looks like” a reliever. Jamie Moyer “looks like” a starter. I’m sure, on both a conscious and subconscious level, things like height, physical appearance, “makeup,” and even race are taken into account when managers are assigning roles to amateur pitchers.

Again, it’s an extremely interesting question.

As I recall, The Book surmised that today’s usage patterns for relievers are not optimal by any stretch. The best relievers could comfortably pitch 30-40 more innings per year without sacrificing much efficiency (if any). Of course, baseball used to operate like that.

I’d say it’s a good thing sabermetrics is dead or else I’d have to think about troublesome questions like these.

I think it’s obviously bad for the pitching team, in the long run, to go through bullpens faster (at least given modern patterns of relief pitcher usage). Yes, you can freshen up your arms by swapping out fungible relievers for guys in AAA - at least, most of the time. But there are costs to doing so - it’s not a painless transaction for the team involved. Especially if your worst non-replacement reliever doesn’t have options and your second best does, or if the guys you’re calling up are burning options and service time.

(Tangent: I’d argue that the effective replacement level for a relief pitcher for a team with a really bad starting rotation is lower, because anyone who is a competant pitcher in relief is being pressed into service in the rotation. Of course I haven’t studied this.)

The real question is whether the batting team benefits, and I’m not sure they do - obviously high pitch counts from the starter correlate highly with things that do benefit the offense, of course. But by the time you’ve worn out the opponent’s pen significantly it’s very likely that they’re moving on to the next three-game set.

I’ve always wondered about this some.

I’d love to see a team try to take as many pitches as they could against a team’s ace.  If you are facing Tim Lincecum, any of the relief pitchers that the Giants bring in are going to be inferior so the faster you can get him out of the game, the better.  I wonder how many pitches a team could force him to throw if they only tried to take pitches and foul pitches off.  He’s good and induces a ton of swinging strikes so it might be tough but if you aren’t trying to actually hit for power, I bet batters could hit foul balls for a while.  What do you think, 10 pitches per at bat?  That might be a little high but that’d get him out of the game by maybe the ~3rd inning.

Assuming Tim Lincecum is around 3 R/9 and the Giants RP is around 4 R/9, if you get him out of there after the 3rd inning (with a strategy which would yield zero runs) you’d average 2.6 R/9 which means you’d be better off seeing Lincecum for the normal length of time.

In order to make this sort of strategy worth it you need the bullpen to be pretty lousy or the starter to be super insanely good.

You hear all the time about how if a team can be patient and cause a starting pitcher to have a high pitch count, that it is a good thing (for the batting team).

=====================================

I really think that optimal batting strategy remains the same regardless of the strength of a team’s starters versus their bullpen.
With a great pitcher on the mound I don’t think it does you any favors to take more pitches than normal because you are giving even more advantage to an already excellent starter. Even if you can force him out of the game a batter or two earlier (which I don’t buy, anyway), sacrificing optimal batting strategy for the batters that do face him does not seem like a good idea.

That’s an interesting question.  Consider this: if you turn the worst team at P/PA (SF, 3.59 P/PA) into the best team (COL, 4.01) you are now “costing” the opponent .42 pitches per PA.  That’s just 10 pitches in a typical start, or under 2/3 of an inning.  The difference between the SP and 2/3 of an inning of RP is likely quite negligible.

So in my mind the key to wearing out a pitcher is not necessarily in the patience required to be an extreme P/PA batter.  The key to wearing out a pitcher is the same as the key to run scoring… not making outs.

Consider the case of 2 pitchers: Chris Carpenter & Jeff Suppan.  Both have made 21 starts, and have a very similar number of pitches per start (94.5, 92.5) and pitches per PA (3.54, 3.66).  However Carpenter and his .261 OBP-against has turned those ~2000 pitches into 145.2 innings, with 13.6 P/IP.  Suppan and his .388 OBP-against has turned those ~2000 pitches into 114.1 innings, with 17 P/IP.  That’s a difference of 1.5 innings per game.

So (duh) if you want to knock a pitcher out of the game, don’t make outs.  I think that facing additional batters is the main cause of elevated pitch counts, and clearly that’s universally a good thing (regardless of the relative talent levels of the SP/RPs) because it’s also the key to scoring runs.

IIRC, this whole concept kind of started with the late 90’s Yankees.  Those teams were patient and seemed to get to the other teams’ bullpens quicker than most.

However, the main thing alluded to earlier is that those Yankess did not make outs.  They were not intentionally trying to get the starter out of the game, but the by-product of their not making outs is that the starter’s did not last as long by IP.

Therefore their “optimal” approach to hitting resulted in starters getting out earlier in the game and the typical long reliever is not as good as a typical starter.  You really need to look at long reliever ERAs (with all normal caveats) to see the “get the starter out” advantage rather than bullpen ERA.

The converse to this is a good hitting team that does not take a lot of pitches (Angels prior to this year?).  Do they still get starters out of the game early?

Definitely an interesting question, and some research is called for.

What I see is that while many teams have decent closers and some have decent 8th inning guys, most teams have very poor choices for 6th and 7th innings. If you drive a starter out of the game by the 6th, you have a decent chance at facing a guy with a 7.39 ERA.

I guess another obvious question is whether the pitcher’s pitch count really even matter.  What is a better predictor of how long a pitcher goes in a game, pitch count or innings pitched?  Do pitchers get taken out of a game if they are doing well but have high pitch counts (they do but what extent)?

#9, if a pitcher comes out in the 5th or 6th innings and it is a close game and the manager puts in one of his worst relievers, he is an idiot. As I said, bad relievers should only be pitching in low leverage situations.  There are plenty of those.  Also, managers should not be using relievers more than one time through the order unless it is a low leverage situation.  It is one thing to talk about an optimal strategy such as getting your below average starter out of the game as soon as possible. It is another thing for a manager to f**k that up by bringing in a bad reliever (when the leverage is not low) or pitching that reliever for 3 or 4 innings and allowing him to bat…

I can think of 3 advantages to elevating the pitch count
1)A tired pitcher may be more likely to make mistakes and the earlier they are tired, the earlier a guy is tired, the less likely the opposing manager is to just pull him(the Grady Little effect)
1a)elevated pitch count(well, without lots of hits) come from guys fouling off pitches. This gives the hitters a better look at the pitchers arsenal, which may help the second time around
2)As mentioned above, most teams don’t have 3 top level relievers, so an inning or against the weaker guys is nice
3)It does seem like reliever fatigue can build up. Ie, it may help you later in the series if the first starter only lasts 5 innings.
4)The more pitching changes there are, the more likely the batting team is to run into a pitcher having a bad day.
That’s all the reasons I’ve seen given for this. Any more?

This is all a moot point.  The maxim of taking pitches, getting a high pitch count to get the starter out of the game comes from situations when aces or guys on a roll are on the mound. 
Now, certainly, if you’re taking to task that this tenet applies with scrub starters on the mound, I’m with you.  HOWEVER, I believe what started as sound baseball logic (against aces or on-a-rolls) got overutilized by the Tim McCarvers of the media world.  Real baseball people would never try to work a count to get Russ Ortiz (circa ‘09) out of the game.

#11, MGL,

You may be underestimating how many managers go simply by set bullpen roles, defining bullpen roles by inning. If that is a strategy employed by most managers, it may be an optimal strategy to raise pitch counts so that you get the designated crappy 6th inning guy.

This doesn’t work in the playoffs, perhaps, but probably works pretty well in regular season.

I posted this on fangraphs the other day: Why not do away with starters completely?

The pitcher in question was Phil Hughes, and the observed result was that when he was moved to the bullpen he went from blah starter to knockout reliever.  He throws a lot harder from the bullpen and he only faces batters once per appearance.  Seems like this is a pretty common story.

A team could take some retread starters and convert them to good relievers.  It could also mix in some normal relievers and some replacement relievers.  It could then give each member of a large pool (hmm, let’s say 10 or 11) of these “relievers” a once-through-the-lineup every few days.  Starting in the first inning.  No pitcher sees a batter twice in an outing.  Each guy goes short innings at max effort every time out.  Bonus: in the NL, you get more chances to pinch hit for your pitchers early in the game.

Relievers have a higher performance baseline than starters.  This could really work.

Yes, we talk about a similar strategy in The Book.

#11, you’re absolutely right that it’s a bad strategy, but I see it all the time. Managers just won’t bring in their regular 9th inning pitcher in the 8th, won’t bring in their 8th inning guys in the 7th, or their 7th inning guy in the 6th. Or their LOOGY in the 6th, or for more than 1 or 2 batters anytime. And that means when they need a guy in the 6th, it’s a bum, time and time again.

#15 My dad suggested something similar, then later that year LaRussa said he was going to a 3 man rotation with pitchers expected to go no more than 3 innings.  The experiment didn’t last long.  I don’t know if he gave it enough time, or if he really had a roster built for it.

#14, I am somewhat assuming that most managers are not going to automatically bring in their worst relievers in the 6th inning of a close game.  There are plenty of late inning blowouts to do that.

That being said, even a mediocre reliever is better than an average starter who is working on his 3rd or 4th time through the order.  Don’t be fooled by the high ERA’s you see from a few of every team’s bullpen pitchers.  Small sample size and selective sampling.  For example, put 8 average relievers (true ERA of around 4.00 or so) in everyone’s bullpen, and it is almost guaranteed that 1-2 per team will have horrible ERA’s at mid season.