Indeed, not using unit tests and instead trying to manually test all the cases sounds crazy to me!
And yes this kind of use-case is exactly where unit tests shine...
These aren't really that different. Consider the history of the earliest (non-assembly) programming languages, particularly https://en.wikipedia.org/wiki/Speedcoding , as well as the ideas expressed by Lisp.
I'm sorry. I don't want to fight here, but you have literally just said you paid Claude to do the thinking for you (except for some math), yet you're talking about this like you're some kind of scientist; or that you've done this extensive, in-depth work.
You made an AI vibe-code an app in a week and now you're impressed someone else was able to do it better?
Am I missing something? Is it maybe just your writing style that makes it come across so "from your high horse"?
It's unclear whether the author wrote all of this themselves, or if they outsourced a bunch of it to Claude. My experience with Claude was that it was terrible at writing code to do the math, even when I explained what the calculation needed to be, what the input was, and what the expected result was. It ultimately took starting a whole new project just to do the rolling window calculation, and then have that fed back in.
My biggest question for the author, if they happen to see this, is: how much manual testing validation did you do of the outputs the app produces? IE: Did you do the inputs + transformations = output calculations yourself as well, counting days on calendars, etc, to validate that the app is actually accurate? (That was the only way I developed any faith in solution I made for myself, which is way less impressive than your app). Regardless of whether you wrote the code yourself or not, a thorough test harness feels vitally important for an app like this.