I concur with working with the story as that is where the main situation resides.
I often create -- types based off a concept within PL 10 for a group of PL 10s -- but the points this NPC PL 10 gets may be more than what the PCs got this means the baddie is not over powered in comparison but they have various Ace's up their sleeve to make the heroes consider whether taking down this bad guy is worth it.
I mean look at most of the "big bad guys" they can be beaten by the heroes maybe not mono-on-mono but as a team. Further you can have villains that are Immune to say the typical method used by the players but very weak to another method that is available but is rarely used.
What about a thinker type character -- they are notoriously the worst villain to face and yet they have very little in the way of "personal fire power" as their weapon is their brain and their plans -- heck the Joker took on superman and "bested" him and if that was not a physical mismatch I do not know what is.
Again it all really depends on how you choose to present the bad guy and very little to do with their actual fire-power.... Heck most often the high fire-power villains are the lackeys controlled by the boss and are often as numerous as the PCs --- further if you have a group that is focused on brute force -- present them with a team of villains that show them how finesse and teamwork are much better.
The game actually balances high level villains using probability against a party because of the way the action economy works. This is all by design. Though coming from D&D, it may feel like there is something in need of fixing.
One PL 15 villain against 6 PL 10 PC's works because the PC's will miss most of the time. They are supposed to miss. And while the villain will likely hit hard, those hard hits only come once a round, distributed across 6 PC's. If half or more of the PC's could hit every round, the villain would go down in a couple of rounds. This would be terribly anti-climactic. The probabilities of this fight are set up to allow the solo PL 15 villain to appropriately challenge the party without becoming a glass cannon, and without a TPK for the party.
I would caution strongly against changing the die away from a d20 to something like a 2d10 without changing other things as well. M&M is designed to work with the swinginess of the d20 and whiff factor is by design. It was not designed to work with a bell curve. PC's rolling 3d20 take the middle, or 2d10, will only exacerbate the whiff factor. More rolls will be clustered around the 8-12 range, and make it that much harder to hit a villain 5 PL's higher. On the opposite side, the villain themself will hit harder and for greater effect, as bonuses have a greater impact to the math of a bell curve system.
Now, if you find the whiff factor frustrating (which I certainly don't disagree with), I think there are better approaches. Use house rules to lower the PL discrepancy but grant the villain Villain Points to shake off hits. Change the action economy, etc. I'll think about it and post more later.
If you get creative you can built powers the replicate "tricks". That was kind of the point of the Luck and Talent sections of Power Profiles. Put Ranged, Indirect X, and Subtle on an attack effect and you have a "pre-set" trap. Put Subtle 2 on Teleport and the villain can always arrive at the right moment. And of course you can power stunt these things so the heroes get HP for them.
3d20 and picking the median does even out a lot of the randomness, and is great for situations where you want small differences in ranks to matter (I use it for ability checks in my games). But it's not a solution to the problem of needing really high PL villains to challenge large groups of heroes and the resulting randomness of the battles. In fact, it'll guarantee that significantly lower PL characters can never defeat higher PL ones. For that, you need some way to even out the advantage of the heroes having more actions and durability -- by giving the villains some combo of miniboss squads and minions for support, villain points, reaction attacks, traps and environmental hazards, PL-breaking defenses, and fiddling with the rules to give them extra actions.
Yeah, the single-powerful-opponent-against-team problem is difficult in almost every non-narrative focused superhero RPG, because the single opponent needs to cover so much ground defensively, and a lot of the mechanisms designed so PCs don't fold up the first time someone gets a lucky hit are particularly strong against a single opponent, even a powerful one. Its been an issue in every superhero game I've ever played.