well, this is a really intersting article for me to read. I dont have customers who use my tools, but I did write a backtester in python. Obviously it uses dataframes, but going from OHLCV to OHLCV+indicators to OHLCV+indicators+calls to OHLCV+indicators+calls+balance calculations. thats my code a very long time (a few minutes for 4 years of 1H data) I dont know if you consider this fast or slow, but since I want to do gradient searching it is WAY too slow.

So I decided to move it to Golang. Now you have freaked me out.