Ok I am throwing this out there as a theory.  If anyone wants to add their two cents they are more than welcomed.  

Let us start with the idea that relievers are very inconsistent from year to year.  I think there are two reasons for this. First, most relievers only have two major league ready pitches.  Since it takes two pitches to get most major league hitters out, once a reliever loses command or confidence of one of these pitches he is toast as hitters will sit on the one pitch he can throw.  Second, reliever's workloads vary from year to year as managers ride the hot arm.  The idea is that a reliever having a good year will get 80 innings but will be less likely to throw that amount again next year just from the wear and tear of the previous year.  A reliever throwing only forty innings will be more likely to bounce back to normal stats the next year due to reduced workload.  (Remember this is a theory, I haven't looked the stats up.)

Based on these two ideas building a bullpen becomes a practice in risk management.  We have limited types of pitchers we should acquire and for very short contracts.  I think that maybe the whole logic of having closers on multi-year contracts and middle relievers on short contracts is flawed.

The first type of pitcher is one just converted to a relief role from a starting role.  This is because of the idea mentioned above that relievers have only two reliable pitches.  Most starters have three pitches they throw.  If they lose confidence in one of their pitches they still have two pitches working.  As they throw more relief innings they will eventually lose one of their pitches and focus on their best two.  Generally these guys fill up middle relief roles and make-up about fifty percent of a bullpen. (3-4 slots out of 6 or 7 bullpen pitchers)  Since they are not so far removed from starting, it might be safe to give these guys contracts ranging from two to three years, as they would still retain their three pitches.  It would also be wise to have five or six of these guys on your 40 man roster to account for injury and inconsistency.  These guys are failed starters, you should be able to pick them up for league min-1 million per.

The second type of pitchers are pitchers that have had success in the past but are coming off of low workload seasons because of general ineffectiveness.  (This is just an idea.  I usually build my fantasy staff off of guys like this and do really well.)  Of course this describes a lot of pitchers in all shapes and sizes.  Methink that it would be wise to pick up the high strike-out types and potential LOOGYs of the bunch and give them one to two year contracts at 2-3 million per.  Out of this type of pitcher you would want to find a closer, a LOOGY, and set-up types.  Again it would be wise to have a couple extras on the 40 man roster for insurance.

Let take a look at the finances. 7 guys in a pen and 12 potential candidates on a 40 man roster.  

3 set-up, LOOGY, closer types @ 2.5 mil/per = 7.5 mil
4 mid-relief types @ 750K/per = 3 mil
2 set-up, LOOGY, closer types on AAA roster @ 1 mil/per = 2 mil
3 mid-relief types on AAA roster @ 500K/per = 1.5 mil

14 million for a pen with an ample insurance policy.  Stills seams expensive, but we are talking about 10.5 million on the major league team. Oh yeah, no veteran closers required.