so I've been re-structuring my entire portfolio management infra to be scalable and more modular to sustain different model types, order types (execution engine), exchanges, etc

the problem, was that this added complexity generated a lot of new errors that usually happened overnight when it ran.

what I'm doing to mitigate that is creating a comprehensive log report db that tells me exactly what has been done at each run.

at least all the sensible stuff that matters.

and try to bring the success rate over the next months as close to 100% as possible so I don't have to wake up every night to new errors or the same errors.

also I am going to start measuring the avg duration of each run. you can see that it takes me about 2minutes to generate the order file for each of the exchanges right now.

although they run asynchronously, I know it can get a lot better.

not that it matters a ton for the low frequency models I run now, but I want to see that execution time trending down, rather than stagnant at the same timing.

track > measure > improve.