IMO vaguely is: Core is based on custom pattern recognition algo, about 3k lines of code.
I wrote a bot and gave it 25k to trade
- Thread starter _terminus_
- Start date
Since you have a profitable algo, I'd obviously define "vaguely" as "upload the full codebase."
I hope the first reaction after seeing the image was "codebase is here!
Last edited:
Both have advantages but i would not use python in performance hungry core components.
Maybe if wrapper to C like Numpy lib and not too many function calls.
I tend to add parameters that can increase test periods exponentially.
Atm its normal for test to take a week
Waiting times with python would be too long (probably like week C++ vs month+ Python)
And idk what to improve next if modifications are not tested.
So in my case development speed and execution speed are both faster with C/C++.
From: https://machinekoder.com/choosing-qt-for-python-or-qt-cpp/
Last edited:
Both have advantages but i would not use python in performance hungry core components.
Maybe if wrapper to C like Numpy lib and not too many function calls.
I usually tend to add parameters that can increase test periods exponentially.
Atm its normal for test to take a week
Waiting times with python would be too long (probably like week C++ vs month+ Python)
And idk what to improve next if modifications are not tested.
So in my case development speed and execution speed are both faster with C/C++.
A good developer is worth 200K annual. Will rather code in python and then buy 96 cores if performance is needed.
I don't think everything can benefit from parallel processing. Some things have to happen one after the other.
Correct me if I'm wrong.
IMO its always better to work around a problem vs throw money at it.
But yes developer time can cost more than hardware.
In my country avg C++ dev salery was ~2k/month.
At least 3-4 years ago when i checked statistics reported by devs.
With experience and good problem solving skill 3-5k probably negotiable but rare.
16k/month in usa? after taxes?
Last edited:
In my case it could be parallelized even between server farm and each board could use 1tb+ ram to speed task up.