3 ms·
I had an idea to analyze markets movements for day-traders. I had the data. I had the idea. I had a anecdotal-proof that my idea was working. I was missing 1)
by HenryBemis 2y ago
I had an idea to analyze markets movements for day-traders. I had the data. I had the idea. I had a anecdotal-proof that my idea was working.
I was missing 1) the python/Julius machine that would do the work for me (yeah it would have taken less years to learn Python) and 2) the ChatGPT to help me with the code for writing the actual trading code. I couldn't test my theory in Excel (difficult to make 3D calculations on 2D tables). I was too laze to do it in Access. I could have done it in SPSS but I haven't used it since 2003, so... no.
Once "AI" came out, I knew I could offload this to a Py-AI machine, throw a some $ at it for a subscription and see if that works (and in which timeframes, instruments, etc) so I can then fine-tune and make niche predictions (e.g. Gold in Januaries, on Tuesdays, between 14:30 to 15:30 it goes up $5 with a 90% rate of success)
(yeah it would have taken less years to learn Python)
(yeah it would have taken less years to learn MQL4)