I am looking at a new trade idea, that I am already manually trading, that if I decide to build it out for real, will take me a few weeks.

theres a few steps i consider before making a decision to sink significant time into a project.
its quite a significant restructure and im not sure on the ROI of the task just yet. its seems juicy but have no way to tell until i do some volume.
first thing I am going to do it to do a separate simpler version and model out the effect. its quite a lot of data and theres things i can do first rather than trying to incorporate into my main infra right off the bat. it will take me 1-2 days to get this stage first.
next, if i can model something more appropriately than what I am doing now manually, I'll design a simple execution algo, or a simple csv file i can just plug into something and execute, nothing fancy, can be built out in a morning. the idea here is can the most simple thing actually capture the effect I am observing here. what are the quirks. what sucks about it. etc etc.
after 1-3 months of gathering data, and if it makes me an attractive return, ill start the steps to integrate it with the main infra and go from there.
I try to never sink in too much time on things I am not sure are worth it. rather start with the simplest possible version and then prove it first and work on stuff that is already yielding higher ROI or that are more likely to do so.