Over the next year I want to try some low-intensity daily experiments to see if they have noticeable and positive effects: Things like cold showers, a gratitude journal, cutting sugar, etc. My working plan is to do one per month for a two-week stretch, to use the other two weeks of the month as the closest I can get to a control, and to track my mood and any other relevant metrics each day. Obviously it's hard to get reliable data from these, but are there ways I could increase the likelihood of extracting good information from the noise?
From my past reading, I expect that 2 weeks is far too short a time to produce noticeable results on any of the specific experiments you mentioned. You may consider a few months for each experiment, which may result in having to run a few unrelated ones concurrently if you want to fit a particular set into 2020.