You and I sound a lot alike. Though I'm in the tech field I tend to bounce between left brain and right brain. I'm very analytical in some respects, emotional in others. (As if this thread hadn't made that sufficiently obvious...)
The plan is paying dividends already due to OOTP's fatigue model and the tendency of AI managers to use pitchers too often for too short stints, wearing them out. (I call this "Dave Roberts bullpen syndrome" -- hated how he managed his bullpen in the WS last year.)
Ironically, the first place this is coming in handy is against the Dodgers. Four game series. First game goes 15 innings. Second game goes 10. By this point both bullpens are not in great shape, but theirs is MUCH worse. And the game three starter was Stripling, who only has 35 stamina. (It's worth noting that in this universe Kershaw left, but the Dodgers still have a lot of talent.)
Anyway, once Stripling was done they had maybe 2 arms left. I still had several guys who could pitch. I actually expanded the rosters in this game to 27 and forced the AI to carry at least 8 relievers (on recommendation from someone here) because my first year I was winning games I felt I should have lost simply due to AI bullpens being worn out. The Dodgers have 9, but the extra inning games took their toll and they couldn't keep up.
I currently have 10 relievers. Overkill, but until I get a stable rotation, it's keeping me alive, and I don't have enough useful utility players for a meaningful bench anyway so 4 guys is enough there. Of those 10 relievers, 3 have 60 or 65 stamina. They are enormously useful for letting me pull my marginal starters before they get deep enough into a game to get hammered. My hook is fair, but merciless.
Even my closer has 45 stamina. I actually wanted to convert him to a starter but he only has two pitches.
Two of those long guys were starters I got last year, and at the time I was thrilled to have gotten them because even as borderline starters they replaced guys who had no business being in MLB at all. But both have been brutal so far this year. I know about sample size, but I can't keep trotting out a guy who gives up 5 or 6 runs a start.
Weirdly, I have a guy who based on his ratings should be ineffective, and yet was my best starter last year. He was signed as a free agent after an outright release by the Dodgers, so you can see what the AI thinks of him. Despite 60/35/55 stats he gave me 160+ IP of 3.88 ERA ball, which is not bad. I planned to drop him this year when his stats eroded to 55/30/55, but he keeps pitching well, so he's back in the rotation (now back to 55/35/55). I guess some guys just overperform.
Meanwhile, those two starters I pulled... both are unhappy, but both are pitching like new men.
Their combined stats as starters: 40 1/3 IP, 49 H, 18 BB, 29 K, 7.58 ERA, 1.66 WHIP
As relievers so far: 9 2/3 IP, 5 H, 3 BB, 16 K, 2.79 ERA, 0.83 WHIP
It's early, but I like what I see. And if it turns out this is just natural variation, I can always put them back in the rotation if someone else loses it I guess.
EDIT: Of course right after I post that, one of these clowns gives up two more homers IN ONE INNING! I swear, I'm just going to outright release this idiot. GRRRR. The starter also gave up 2 HR. It's nuts.
EDIT2: The next game, my team does nothing against a pitcher with 45 stuff (every time I put in a guy like that he gets destroyed). Tie game and my guys go meekly in the bottom of the 9th, my setup guy walks two guys, then the "closer" walks in the winning run and my useless offense does nothing against a mediocre middle reliever in the bottom of the 10th. I think I need to put this game away before I break something... no wonder real life managers have gray hair...