So took me much longer to port my stock trading algorithm over to a proper automated system then I hoped...
I knew I had something impressive because I had backtested a rudementary version of my algorithm and the ROI was astronomical. Problem was the platform I was using was very basic, it couldnt autotrade for me and was almost impossible to code anything too advanced anyway. But it did provide back testing. For some time after that i used the algorithm to trade manually, checking when it told me to buy or sell and doing it. while that proved out the ROI was real it didnt really reach its full potential as I would often trade sometime after the algorithm alerted me.
Now I have everything **finally** automated took two solid months but its working. I have thrown every stock I could find at it and after backing for a year, random years every time, different stock, not **once** has the algorithm lost me a penny at the end of any year. Some stocks it never traded on because the volume was too low and apparently it knows well enough to not go in on those stocks (not entirely sure how honestly even though i wrote the damn thing), for those stocks the ROI is 0 of course. But for every other stock I am coming in multiple times over buy&hold, even when the stock goes negative overall and crashes (it knows when to short)!!!!
I am so excited how well this is working out.
#stockMarket #Quant #algotrading #algorithms #algo #programming #python
@swiley Well I switched between a few as I realized none of them supported what I needed or were so buggy they were preventing me from getting work done.
I was using quantrocket but the bugs are overwhelming and each one took up so much of my time it was one of my biggest mistakes. Now I've handrolled my own. only tools I use that could be considered a framework right now is zipline. I get my history data from polygon.
QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.