I've discussed this with xanthan via e-mail. I think he might have a post like this over at his site but I'll put one here just in case he hasn't.

I don't know how many of you have read The Book by Tom Tango, Mitchel Lichtman and Andy Dolphin, but if you haven't, it's definitely something to pick up (if you do, please do so from the site I linked; it helps out the authors, and they in turn are donating a good deal of their royalties to Retrosheet). Basically, the three authors investigate certain baseball strategies (e.g. reliever use, platooning, the sacrifice bunt) and come to conclusions based on empirical evidence (using play-by-play data mainly from 1999-2002). The run values for this time period are slightly higher than now, but the difference is essentially negligible. The strategies presented in The Book make sense (based on the numbers) but are all a bit unorthodox. This one is especially "strange" but would make the most sense for any team that wanted to score more runs (or need to, like the Giants).

In Chapter 5, entitled "Batting (Dis)Order", Tom Tango explores various ways to optimize one's lineup. There are a bunch of little things Bochy could do to squeeze out a couple runs here and there, such as batting the pitcher 8th (it actually works! for a couple of runs a year anyway). I don't want to describe all the little things because that's a lot to write, but if you optimized your lineup like Tango showed in his constructed simulation model (that mirrors reality pretty much perfectly, as shown in The Book), the result is about 10-15 runs per season. That's only about 1-1 1/2 wins per season added, which isn't that much really. However, Tango presents an interesting question:

What if you don't ever let the pitcher hit?

According to his model, an average NL team will score 5.25 when the pitcher never hits, yet when the pitcher hits 65% of the time (which is how it is now), the team will only score 4.83 runs per game. That difference is about 68 runs per season, and, FTW!, that's 68 EXTRA RUNS PER SEASON. That's between 6-7 wins right there. Now, as Tango points out, the value of having your good (Cain), great (Lincecum), and above-average (Sanchez) pitchers on the mound is greater than substituting them for an average (at least) batter. One should, therefore, only use this strategy of never allowing the pitcher to hit when you're talking about average/below-average starters on the mound.

The Giants fit these roles pretty much perfectly, with a true 1, 2, 3 , some candidates for the 4/5 roles and a slew of good relievers.

How the "rotation" would look.

Day 1: Lincecum

Day 2: Cain

Day 3: Zito/Correia/Valdez (Tango says 3rd best reliever here, so I'm just going to throw Merkin in here, with the assumption he's healthy next season)

Day 4: Sanchez

Day 5: Correia/Zito/Valdez

The "three-headed monster" takes you through the sixth inning most days.

Relievers 1,2, and 4-6 would be used as usual, plus if any pitcher of the "three-headed" mini-rotation pitched a relatively short outing on an earlier day, he could also pitch in relief.

This setup also allows the bench to be stronger overall, with 6 bench players instead of the customary 4. Best of all, some of those guys don't even need to be able to field well in order to contribute regularly.

The only obvious downfalls here are:

a. Zito/Correia/whomever you assigned to those roles would probably be pretty upset.

b. The best bench bats wouldn't be there at the end of the games (for higher-leverage situations) for some games.

c. The third-best reliever is pretty restricted.

Not a whole lot really.

Using these strategies, along with the other ones listed in the chapter, would (in the model) add around 50 runs per season to the team's total and keep the pitching/runs against around the same level. That's about 5 wins, which is pretty damn huge.

Of course, this in itself won't help our beloved Giants become contenders. But it sure couldn't hurt.