The Secret to Good Pitching
It sounds too simple: win ERA and ratio and you'll win your fantasy league.  Wins and saves are pitching commodities that are overpriced.  Get a big name closer and a #1 starter and you're likely to have spent most of your allowance for pitching.  And you still have seven spots to fill on your staff.  If you spend the money to get Greg Maddux, you're likely to have fill out the rest of your staff with the Tyler Greens of the world, which negates the benefit of having Maddux in the first place.  But for the price of one Greg Maddux, you could've gotten 3 or 4 Jon Liebers.   Likewise, for the price of one Robb Nen, you could've picked up several John Rockers.
Well, that would've been nice to know 2 months ago, but right now that's not much help.  OK, here's how you make the adjustment in mid-season.  First, pare down to the miminum number of starters.  If you have 5 starters who can get you the required number of innings, go with 5.  There are two reasons why you want to dump as many starters as possible: 1) there just aren't that many really good starters out there, and 2) starters generally have higher ERAs and ratios than do relievers.  Next, pick up several relievers who have good strikeout-to-walk and hit-to-innings ratios.  It doesn't matter what their ERAs are right now.  A bad ERA for a reliever could have been the result of one particularly ugly outing.   A couple of years ago, Jesse Orosco got brutalized at Texas in one game to the tune of 9 earned runs in less than an inning.  He finished the season with an ERA of 3.40.  If you had picked him up after the Texas massacre, he gave you an ERA under 1.5 for the rest of the season.  Chances are if their ratios are good, the numbers are gonna get better.  It also doesn't matter what their roles are.  Good numbers are good numbers and if a pitcher keeps posting good numbers, the manager will find a way to get him valuable innings, eventually as either a starter or as a set-up guy/closer.
If you have 2 or 3 really good relievers, you can carry a corps of crappy starters and still be in the hunt to win pitching.  Three good relievers will give you the same ERA and ratio as a top dollar pitcher, meaning 240 innings of sub 3.00 ERA and a ratio around 1.10.  If you have one starter with an ERA of 3.25 and a ratio of 1.20, the rest of your staff only has to produce a 4.79 ERA and a ratio of 1.504 in order for you to end up with a very competitive team ERA of 4 and team ratio of 1.35.  At any time in the majors, there are 20 to 30 relievers who will produce the numbers you need.  So why don't more people pick these guys up?  Because quite often they are middle or long relievers who don't get a lot of wins or saves.  No biggie.  Like I said, good numbers are good numbers.  It doesn't matter where they come from.
Let's take it a step further.  not only are you gonna pick up good relievers, but you're gonna pick up the best possible relievers.  Look at teams where your guys have a chance to expand their roles.  In 1993 Jeff Fassero began the season as a middle/long reliever but because the Expo starting pitching didn't develop as planned he became a starter.  By the end of the season, he was one of the better starters in the NL, compiling 12 wins and an ERA of 2.29.  Cincinnati's Ron Villone appears to be on a similar path this year.  Look for pitching staffs with injury-prone starters or closers who have struggled in their save opportunities.  Currently in the NL, Greg Olson (Arz), Bob Wickman (Mil) and Rod Beck/Rick Aguilera (ChC) have struggled.  Their possible replacements are Vladimir Nunez or Byun-Hyung Kim, David Weathers or Jim Pittsley, and Terry Adams respectively.  Other possible relievers with upside are Alan Embree (SF), Joe Nathan (SF), Jerry Spradlin (SF), Steve Montgomery (Phi), Felix Heredia (ChC), Scott Sanders (ChC), Dennis Reyes (Cin), Will Cunnane (SD), Rich Croushoure (StL), Lance Painter (StL), Manny Aybar (StL), Brian Williams (Hou) and of course, Scott Elarton (Hou).