Paper trading is back up. I took the off time to fix the bugs I had referred to earlier. The whole system is getting really robust IMO. Is this week going to be the "hands off week"? We're away on vacation this week in PA, so maybe it's a blessing in disguise: I can't touch it!
Returned from vacation, nice and relaxed. I love vacation. For about 34 years of my life I never took a vacation, and now I take multiple vacations a year. Something I learned at the hedge fund. Boss man let us use his vacation spots for free. It was incredible. I miss that place. Now... How did "hands off" week go? Well, it wasn't hands off to begin with as something was screwy till about Wednesday, so it was only hands off for Thursday and Friday. The screwy bit (issues now fixed) made me miss some entries which meant that the only actual entry last week was a loss of about $600. Being AFK, I had a chance to take stock of what I have, which is: - A custom system for trading with IB based on backtrader - A dashboard to monitor health of algo(s) - A "stop all trading" approach - A system that self-heals when things don't appear to be working as they should I've additionally set up the system to bring itself up and down a half hour after/before trading starts/ends. Technically, I could leave it up and nothing would happen, but this seems neater. So what's going to happen this week? Only one point of focus: matching backtesting entries/exits with paper trading. I believe that the self-healing properties of the system should make it possible to have decent insight into the matching or lack thereof of paper trading with backtesting.
@nooby_mcnoob it sounds like that you have your software up to a point where you can let it run unsupervised. And that it only will require manual intervention on those truly unexpected exceptional cases. And then? Then you can go on holiday as often as you like. Hopefully the software will generate enough money to pay all your holiday expenses, while you’re away from home.
I should mention though: it doesn't run perfect unsupervised. For example, when "self-healing", it loses knowledge of open trades/positions which impacts my visibility into things. Fortunately, stop losses are already entered by that point by design so I won't lose more money than I expected, but this is one thing I need to fix. Just a thought: is running unsupervised a bad goal? The main reason I do this is because of my experience with selling software unsupervised: I make money while doing anything but working. The only thing I screwed up there was managing the support burden.
It sounds like you have a trade off to make between how much supervision is required versus the complexity of the software. The more exceptional cases the software can handle by itself the less you need to manually interfere. But the more complex the software will get. I don’t think that anyone else but you can decide how much software to implement and where you place the border/limit/envelope (choose your preferred terminology).
Indeed. Right now the complexity is very manageable. The only thing I want to sort out is losing the current positions when detecting issues. For example, here are the number of potential data issues detected in the last 24 hours. Yes, I could sort them out, but I'd rather be robust to any type of issue.
That’s wishful thinking. Honestly though, it becomes a giant leap of faith in your broker. All exits should rest on their infrastructure, and always have a source of mobile/remote access if unattended. Don’t even think about resting them on your alley mechanic brainchild