The misery index is pretty simple. It could be even simpler, because changes in the unemployment rate have similar predictive validity.
Of course no model will reliably predict the winner of close elections, and those that purport to are fraudulent. Incumbent share of the national two-party vote is a better metric to target, but obviously of less interest given that four of the five past elections had small popular vote margins and in half of those the EV went the other way.
That said, in some elections a poor economy all but predetermines the outcome (e.g. 1992, 2008). There are others in which its role is much less clear (e.g. 2016).
As of April, my expectation is that 2020 will belong to the former category. If this bears out, Trump will not just lose but lose badly.
The question of course is; to what extent will voters give Trump (and other world leaders up for re-election in the next 12 months or so) a pass on the economy due to the pandemic?
I'd suggest a lot of people will give their leaders a break, at least inititially, due to the unique cause of this recession, and only start looking for blood later when the recovery takes a while to materialize. (e.g. I think Justin Trudeau's minority government should be a lot more worried about a 2022 election than a spring 2021 one). Whether that's enough to save Trump's bacon is another story, but at minimum I'd suggest he'll do a lot better than what we'd expect for an incumbent president in a "normal" double digit unemployment election.
