Bob Melvin should never have had a decision to make at all. In the 8th inning of Saturday's game, A's leading 2-1, Eric O'Flaherty got a comebacker that should have been a routine inning ending DP. However, Marcus Semien botched it but good, missing the bag at 2B before firing a poor throw to 1B that Mark Canha could not handle. Suddenly, instead of "inning over" it's 2 on, 1 out.
O'Flaherty regrouped and retired Robinson Cano on a bouncer to 1B that advanced the runners to 2B and 3B with two outs, with Nelson Cruz due up and Kyle Seager to follow. There were multiple good moves to be made here and unfortunately Melvin made none of them.
The important thing to recognize is that with a one run lead, two outs and runners at 2B and 3B, you are in a spot where any hit will leave you trailing and any out will leave you in front. So this is all about "base hit management" -- you want to face the batter whom you can reduce to the llowest batting average and you want your lowest BAA guy on the mound. Other oft essential considerations, such as OBP and slugging, take a back seat in this particular situation.
The best move, in my view, would have been the simplest: Leave O'Flaherty in and let him pitch to Cruz. Platoons be damned, because O'Flaherty is still excellent against RH batters holding them to a career .266/.323/.345. As importantly, O'Flaherty was in the game and dealing, having just gotten the comebacker and then retired the dangerous Cano. And as dangerous as Cruz can be, his career .267 BA is nothing special.
The next best move, given that this was almost certain to be the highest leverage situation in the game, would have been to go to the A's anointed "best reliever" who, in Sean Doolittle's absence, is Tyler Clippard. Calling for a 4-out save from Clippard would have been perfectly reasonable, especially since it would have offered a RH-RHP matchup for Clippard in the 8th.
The next best move after that would have been to IBB Cruz and go after Seager, only because O'Flaherty has held opposing LHs to a career line of .201/.242/.263. It's a rare case where the on deck hitter's OBP was arguably lower than the current batter's BA -- that is, if you look only at O'Flaherty's splits and not at the fact that Seager is an above average LH hitter (which is why I would in fact elect to pitch to Cruz).
The worst move is the one Melvin selected: To go to Otero. Otero is a good pitcher and due to his versatility -- he can go 2 IP, he can pitch in back to back games, he can get you a key DP -- he is a valuable member of the bullpen. But even at his best, Otero gives up his fair share of hits because he is a ground ball pitcher highly subject to the variance of batted ball luck.
For his career Otero has allowed 142 hits in 139 IP, not surprising since Otero is a pitch-to-contact sinkerballer with a 5.2 career K/9 IP. Otero gets better as you zoom out -- he can get into, but also out, of trouble -- and he gets worse as you zoom in. That is, he is not the guy you want to put in if you have to get the next batter out, and he's a godsend if you need a bunch of quality innings throughout a week or a month.
So then why did Melvin remove O'Flaherty, an excellent reliever who was dealing, in favor of Otero? "Righties should face righties and lefties should face lefties." Well, actually no. I mean, sometimes. Why he didn't have Clippard up is another matter, but I'm not familiar enough with Clippard to know how long he can usually go how many days in a row.
What I am familiar with is the tendency of managers, nowadays, to over manage the platoons and to give short shrift to the notion that a reliever in the game who is throwing great is often a better bet than "what's behind door #2," and that some relievers are better against opposite-hand batters than their peers are against anyone. And most of all, that you need to know when you need strikes, when you need ground balls, when you need to suppress HRs, when you need to suppress any kind of hit -- and that your decision needs to match the need, not just the hand.
I love Melvin and I hope he manages the A's for the next 10 years, and in general I do think he manages the bullpen well. But today I think he pretty clearly made a bad call.