Running

220 - age is clearly nonsense

It's one of the best things there is in the absence of any real world data.

In other words, if you take a formula of the form:

HR_max = x - age

Then x=220 is the best fit population wide, any other value for x (higher or lower) is a worse fit population wide.

Of course, individuals vary massively, but that's exactly the point.

If you have any other data such as "Oh, I'm 30 but I've seen my HR at 195 on my HRM" then you've got a more accurate/appropriate value to use than 220-age.

There are other formulae which are progressively better, but they're more prone to error (or just putting people off) due to the increased arithmetic required.

One formula with a better population wide fit is: 207 - (0.7 * age)
Then an even better one is (I think): 192 - (0.007 * age²)

But, yes, the best measure of HRM is to go out and do something hard and see what your HR peaks at. Then if you ever see a higher figure then that's your new HR_max.

It's also very sports dependent. Sprinting at the end of a parkrun I can only get mine to ~185bpm, but playing 5-a-side it regularly tops 200bpm.

There are second order polynomials that have a better population wide fit.