Does piece of mind count? It appears to me people that feel a need to spend time figuring out the optimum angle to get the highest production, are looking for piece of mind that they are getting the most out of their system. So the question is how accurate is it by doing all the calculations and adjusting it manually? After it's set according the calculations (If the calculations are correct) How correct are the calculations. Can you tweak some more out of it? Isn't that the reason you are spending time to do the calculations to tweak some more out of the setup? What if the calculations you did weren't accurate? What if you did everything correct in your calculations and something else in your particular setup that you didn't know was a factor? What would happen if every time you did your calculations and then tweaked it by hand you found it the same five degrees off in the same direction. Wouldn't that be cool to know. All I'm saying is calculations are great. And because there are a number of factors and those factors are changing every day what does it take to get the optimum angle to get the most production?
Ha! You are actually so right in this regard, if the geometric equations could be solved by a few hacks here, and presumably by any number of undergraduate math majors, or even bright high school students, why aren't they published anywhere? Because a) they were solved long ago, and b) they are utterly impractical. It turns out there are dozens, nay hundreds, of academic papers on optimal tilt angle being published all the time - because reality is different from geometry - and at city- and country-wide scale, a degree of tilt could mean millions of dollars in energy cost/savings.
Seems like the main real-world local issues that go beyond mere geometry are 1) weather (clouds, temps, wind, climate, etc) and 2) solar irradiation (haze, pollution, refraction, transmission), that serious scholars try to account for. As an example of the 2nd effect, one researcher found that in the past 50 years, the optimal tilt angle has decreased by 2 degrees in most mid-latitudes - I think basically saying that increased haze/pollution decreases irradiation when coming in at lower sun angles, therefore tilt higher to capture the clearer, stronger irradiation when the sun is higher overhead. That's probably why for my latitude (38 degrees), "optimal" guidance is way lower, around 30 degrees fixed per most sites.
Even PVWatts, which uses localized weather, only gives the closest weather data 1.3 miles from my house. But I'm in the foothills, so 1.3 miles west of me, the elevation rises 1500 feet, and there is signifcantly more morning clouds/fog. 1.3 east of me, gets closer to the bay, and temps can be 10 degrees different from me during summer. So if you have very localized geographic or micro-climate conditions, even the highest resolution models can be inaccurate.
So as you suggest, doing your own empirical observations, for your own personal conditions, can be important. And of course this will account for local shading and other factors. BUT - there is a challenge with this. Those of us who've tried to do empirical observations of effect of cleaning our solar panels have found this out all too well. In this case, we're not even tilting the panels, we're just trying to measure a <1-10% impact of soiling on our production. What often happens is day-to-day observations are completely overshadowed by slight variations in weather and irradiation (even if it looks like a cloudless sky), and these overwhelm the few % differences people are trying to measure.
So I think the only real benefit of the geometric models, is it gives you a rough baseline of how much is at stake for being off. And as wwhitney pointed out, even a 10 deg difference in tilt might only yield a 2% difference in output, if you're close to optimal already, so easily overwhelmed by other factors. But as far as the empirical observations go, I would only really try it if you can quickly crank the tilt, and take quick samples (like your 5-minute suggestions, not day to day) of output (presumably from inverter monitoring that updates about 1x/minute). And when I say crank, I mean crank it 10-15 degrees in each direction, not fine tune 2 degrees. Per wwhitney, 2 deg off of optimal is about 0.1% difference in output, easily ovewhelmed... see which direction yields a larger difference, that is the direction you tune towards by a degree. Then repeat experiment, until the difference is exactly the same cranking 10 degrees in both directions, means you're about centered on optimal...
As for me, if I had a tilting array, I'd just probably just take PVwatts optimal tilt for each month of the year, even being off 5 degrees makes hardly a difference (fixed optimal again is 30 deg for my location, and I ran from -15 to +15 degrees of that: