In none of the presentations of Pascal's Wager have I seen it applied to other entities, concepts, etc. that are found among the world's religions. For example, not believing in god (or not behaving in a way consistent with that belief) is not the same thing as believing in the devil (or acting in accordance with that belief).
Pursuing Pascal's approach seriously, each entity or concept requires its own wager -- the devil, demons capable of possession, and so on. You might be able to group some of these into larger, more catch-all wagers like one about "evil beings that can influence human outcomes here and now as well as in the afterlife." But there will still be a list of wagers, one for each independent cluster of beliefs. The list will be finite and probably not too long -- no more than a dozen clusters, I wouldn't think.
Each wager follows the same logic as the one about god and heaven. If the thing doesn't exist, you're out some finite amount if you believed in it, compared to the person who did not. However, if the thing does exist, the non-believer gets nothing (maybe they are even punished) while you enjoy infinite benefits.
For example, if witches and sorcerers don't exist, then over a lifetime you've used up a certain amount of time, energy, and resources in order to protect yourself from them, to redirect their magic back at them, etc. Since beliefs cost you nothing, this comes down to all the little and large behaviors you work into your life. Adding it all up, it's probably no more than all the nuisances that we go through sorting every recyclable item into this and that category, hauling the bin out to the curb, lugging it back, paying taxes to run the program, the opportunity costs of the machinery, labor, etc., that powers the program, and so on.
We perform these rituals largely without thought or investigation into whether it really has the effect on the world that we believe it does, and that experts tell us it has. So a witch protection routine isn't so burdensome by modern standards.
As for the potential upside if witches do exist, well then the routine could save your life or that of a family member or friend or ally. Aside from overall health status, there's the stuff you own or the resources you depend on, whether plant, animal, or artifact -- they'll be more bountiful and secure if you can keep the witches away. And then there is the greater reproductive success you'll have if witches have a harder time making you infertile, stealing your children, causing abortions, etc. Not to mention the greater sense of happiness or satisfaction that you'll enjoy as a result of all this. Most people couldn't put a finite price on all of these benefits, and when we look at how such benefits would help the person out in Darwinian fitness terms, they do appear, if not infinitely large, then close enough to it, especially compared to the puny potential losses.
If you don't behave in a way that says you believe in witches, then you're going to miss all of that upside -- indeed, since you're so sure they don't exist, you'll be their easiest target and they'll exploit you first, hardest, and for the longest.
It's odd that modern people, even if they don't accept Pascal's wager, at least treat it as a respectable approach when it comes to god and heaven, but would certainly keel over laughing if they heard someone apply it to witches, possession by spirits, divination, etc. In those cases, it would look too backward and primitive, and professing intellectual beliefs is mostly about showing how sophisticated you think you are, so it would be a no-go. Of course, maybe beliefs and behaviors about witches are among the most widely held and practiced across the globe because they're closer to the unknowable truth than are other religious beliefs and acts.
The rationalist skeptic will foolishly object that you don't need to believe in witches to avoid the real harms of those styled as such. They'd say there's a perfectly naturalistic basis for what appear to be acts of witchcraft -- a once-a-century drought is what killed your crops, a newly introduced pathogen is what wiped out your livestock, and it was a sociopathic serial killer who went on a murderous rampage, not a witch acting from afar. A person can avoid these naturalistic sources of danger without having to dress it up in religious garb.
Why is this objection foolish? Because people who are complacent, overly optimistic, and not paranoid enough in one domain of life are that way in all others as well. I mean, c'mon, a once-a-century bust of the housing market couldn't wipe out the global economy! A brand new "killer germ" -- yeah right, how many times have we heard that one before? An epidemic of serial killers -- please, that's just the powers that be trying to stoke your fears in order to brainwash you into obedience and consumption so you won't be a threat to their power.
So, someone who quickly dismisses the probability times the impact of a bizarre event of supernatural origin is also going to dismiss the contribution of a bizarre event of natural origin. They won't perform the same degree of protection from "witches" as the believer in witches, only stripping away the supernatural mumbo-jumbo and replacing it with more rational and naturalistic ideas. They have a domain-general contempt for what they see as unfounded paranoia about the bizarre, no matter the intellectual framework that supports the paranoia.
Again it's not hard to imagine how easily this overly complacent mindset would be weeded out by natural selection. It seems to be on the rise in industrialized countries, but that's not even 10 generations in most of them, thus hardly a proven stable equilibrium. For all we know this super-sanguine sensibility is going to lead us right over a cliff -- ah c'mon, we've taken 1000 steps in this direction so far, how could the next step do us any harm?