To add to many other good, informative answers and comments, I would reiterate a certain maybe-negative-sounding point, that one should not base future plans on one's own research results or other achievements becoming substantially different from what they have been in the past.

Thus, the "wonderful paper" will most likely not occur, if one has not had too many of these in the past, and "betting the farm" on its occurrence is just stupid, especially if people are depending on you.

This is a special case of predicting that the near future will strongly resemble the near past...

So far as I can tell, loosely based on actual facts, but certainly on substantial direct observation, the extra-academic job market for Ph.D.'d mathematicians is growing faster than the academic job market, which these days grows at most at the rate of population growth, to match undergrad enrollments. The "problem" in academe is that the recent-years' production of Ph.D.'s has grown at a rate exceeding market growth, by a huge factor... this horribly exacerbated by the economic mess of 2007-9, etc., ... but in any case observing that the academic job market (in the U.S. for sure) is over-saturated...

Some of this is a result of lifting mandatory retirement in the U.S., but/and mandatory retirement in Europe seems to motivate some mathematicians in Europe to emigrate to the U.S. to maintain an "active" role, ...

So, unfortunately, if one gets off to a bad start in some way in academe, things are so tight that it's hard to recover... whether or not one could recover "in principle". So, if practical matters dominate, looking at actuarial stuff in the U.S., or banking/finance, could be more practically viable than trying to get postdocs or temporary teaching positions and "stage a comeback". Sadly, perhaps.