It's mathematically rational, but analytically difficult to overcome. That's the voodoo. We know why it's hard to pin down anything about relievers, but that doesn't help in figuring out ways to overcome those limitations with repeatably successful results.
Those issues are what makes them misfit toys. If they weren't they'd be starters. But the have some hiccup or hitch in their skills. Maybe they lack durability. Or stamina. Or a pitch (or two). Or command. or have an arsenal that is quirky at first but can't survive a second trip through the lineup. Guys that slot into relief roles and never come out even though they could be starters are more often the exception rather than the rule (because most teams would obviously prefer a valuable asset be in the more valuable role of starting vs coming out of the pen for like 1/2 or less the # of innings.
Yes, but with starters this is a bad thing because if a guy has, say, 100 innings of value before he gets figured out or breaks down then as a starter that's less than a year whereas a decent pen arm can carve out at least a couple of seasons before his clock strikes midnight. So that's how the bullpen market works: There are too many jobs out there in bullpens around baseball to fill them with guys who are talented and stable, so you make due with guys who are performing and you worry about whether or not it will last after the fact. So accepting that has to be built into the approach you take to pen management. You have to live with the fact that you're choosing guys operating on a clock you can't see with a deadline you can't reliably predict. And that's the voodoo.
Fair enough. though the fact that he couldn't stay healthy is part of the voodoo as well. Because you can't predict with consistency whether or not a guy with recurrent health issues will get over them to become a dependable player or if he'll be the next Dustin McGowan.
Will you though?
Last year's top 10 reliever FA contracts (by total value):
1) Hendriks (White Sox) - all-star year. worth it.
2) Blake Treinen (Dodgers) -Great year. worth it.
3) Trevor May (Mets) - Solid year by results, peripherals suggest possible issues
4) Pedro Baez (Astros) - Got hurt. pitched <5 innings in MLB, plus <10 more in AAA.
5) Trevor Rosenthal (A's) - didn't pitch at all last season
7) Brad Hand (Nationals) - We saw how this worked. He was a gas can. And he wasn't even that good before the deal from the Nats.
8) Ken Giles (Mariners) - didn't pitch at all last season
9) Archie Bradley (Phillies) - was passable, but not great.
10) Kirby Yates (Blue Jays) - didn't pitch at all last season.
The same pattern happened in 2020 as well. and if I keep going I imagine it's much the same working back through the years. Yeah there are some signings that work out, but on balance you appear as likely or more likely to come to regret a big reliever signing than you are to love the results.
Obviously you're going to have to pay for the biggest and best guys. The problem is that there's little to stop those relievers from being nearly as volatile as the lesser ones. So you might be better off betting on a bunch of tier 2 or 3 guys giving you one or two breakouts than you are spending the money on one massive marquee talent who screws you over if he gets hurt or loses his mojo.