the player needs to see the GM's die roll, which to my knowledge is generally frowned upon,
This is not really true—it’s frowned upon by some, but standard operating procedure for others. Either way can work, and individual GMs/tables can and do make their own decision on that. The existence of abilities like this (and misfortune is not alone) suggests that Paizo generally endorses GM rolls in the open, however.
Clearly, at tables where showing the GM’s rolls is the norm, there is no problem using this ability in conjunction with this ability. So I’ll consider the question answered for such tables, and focus the rest of my answer on tables that don’t normally reveal things.
The primary reason to avoid revealing GM rolls is to prevent metagaming—if players know a roll has been made, they can start guessing why even if their characters would have no particular knowledge that any event took place. Likewise, with the number, they can start to guessing that, say, a high number means they just made a hidden saving throw and avoided some danger they didn’t know about, and a low number means they just failed a Perception check to notice some hidden danger. If players have difficulty separating player knowledge from character knowledge, hiding those rolls can have value.
The secondary reason for hiding GM rolls is that it allows for fudging rolls. Generally speaking, fudging rolls is done to avoid wrecking things for PCs when it wouldn’t improve the story for that to happen; fudging things to ruin things for PCs is generally frowned upon.
Luckily, in the case of misfortune, both of these concerns are considerably reduced: since enemies roll saving throws in response to PC actions, there is a lot less for players to learn that their characters shouldn’t. They already know the target is making a saving throw—that was the point of subjecting them to an effect that required one. They do learn, if the roll is poor but the target succeeds anyway, that the target has a much stronger save than they were perhaps expecting—but frankly that sort of thing should be getting described by the GM anyway (e.g. “the orc struggles to move its muscly bulk out of the way of the blast, but mostly manages to do so, taking only half the damage” for a difficult but successful Reflex save against fireball, compared to “the cultist laughs at your feeble attempts to sway him from his devotion, as your dominate person spell finds no purchase at all,” say).¹ And fudging rolls has a lot less justification if you’re doing it to prevent the PCs from succeeding where they otherwise would.
So I would recommend, in such cases, that revealing these rolls has few of the problems that might lead one to hiding most rolls. It would be fairly safe to reveal these rolls, even if it’s every single one. (If an enemy for some reason is forced to make a saving throw when the PCs aren’t aware of it, I think it’s fair that the oracle couldn’t use misfortune anyway.)
But if you’re really devoted to not revealing rolls, the answer is to provide enough description of the target’s attempt to save to hint at whether it was a good roll or a bad roll, so that the oracle has something to base their decisions on. Really, you should be doing this anyway, as I suggested above, but now it becomes particularly crucial.
- And, it goes beyond the scope of this question, but, really, D&D and Pathfinder in general really ought to be revealing a lot more about enemies’ capability to PCs—I have had conversations with numerous martial artists who suggest that things like BAB, Strength, overall level, and so on, should be readily apparent just from seeing someone shape up and assume a fighting stance. We can presume similar cues for spellcasters. Something to consider here—I often try to go beyond what the rules suggest for PCs recognizing the abilities of their foes. (All of this assumes, of course, that the enemy isn’t actively trying to disguise their ability; if they are, all of this goes out the window, at least if the enemy can succeed on their Bluff check.)