/biz/ - Business & Finance

>>9648133
(typed on Microsoft word, look 15 mins)

This would be step one. I’m sure its been done before. If I could just see a working program that can accomplish this first step I’d be able to build on it easily. I just CAN’T get it started.
Here it is:
The idea is to take the standard candle stick color code:
Letter= ( abs(open-close)/abs(high-low) > 0.5 )? (open < close ? A : C) : B
And evaluate charts so the datasets are simply strings of texts that look like “ABCCABCABCACCABCCABAABCAC…”

The machine learning algorithm looks at the datasets that have been reduced to the three letters.
Then the ML algo makes predictions on what the next letter will be on a real time chart.

For example the filtered 10 min chart for GoPro looks like
“ABCCABCABCACCABCCABAABCAC”
and the ML algo out put looks like:
A= 10%
B=30%
C=60%

does this make any sense? I can explain what happens next once the ML program can do this if you're interested.

>>8148009
you're stepping into the real dark arts, anon, risk analysis math is harder than functional analysis