Really interesting research, thank you for posting about it. Have you thought about giving a bit more structure your process of prioritizing your ideas for testing? As you’ve said, you can’t try everything at once, and you are trying to intervene on a complex system that could be sensitive to many different interventions and combinations of interventions. Therefore, you need to develop a reasoned way to prioritize your tests. The way I formulate this problem is: How do I pick out especially promising ideas against a background of many plausible ideas?
For this, I like using natural experiments. That is, once you have a way of measuring your fatigue that you trust and a bunch of plausibly connected observations that you think may somehow be connected, you can carefully look at any days of especially high or low fatigue and ask: what was different on those days or the days just prior? Note that this is not (yet) any kind of formal intervention or trial. And it will of course miss complex and intermittent causal influences. But it’s good for giving some structure to a search for really big effects that may be repeatable.