Programming a price action trading BOT

I use Java. But python is fine i was only half joking about the weak typing. If you are careful it shouldnt be an issue. But with Java and even C++ you get more compile time checking.
I have studied Java when I was still in the business of IT contracting. So I am more familiar with it than Python but Python has AI (machine learning features) that I am not sure if I even have a chance to exploit but it is a nice thing to have. P.S. When I learned java from C++ I just never had a hang of it. Python seems nice, especially with Anaconda.
 
when you have emotions involved with trading it means you don't have a thoroughly thought out plan.
I was short the market with SPY put options, from these recent highs and I covered, now I am short again from the lower high, but patience is not my strong suit. I always have a plan but often I do not stick to it. I should cover between SP 2800-3000, but that is far away and not certain...
With a bot, you can test it to death and just let it go...using micros for 6 months, the take the 'M' out of the symbol...tweak it and then write a book.
 
I have studied Java when I was still in the business of IT contracting. So I am more familiar with it than Python but Python has AI (machine learning features) that I am not sure if I even have a chance to exploit but it is a nice thing to have. P.S. When I learned java from C++ I just never had a hang of it. Python seems nice, especially with Anaconda.
Ok, yeah, I forgot you wanted to play with ML.

Maybe also consider Neuroshell more so than NT or MW.
 
This is not going to be easy, but I have all the time in the world and if I do not do this, I just piss my money away scalping for ticks. My problem in trading is emotional more than picking the right trades. I quit trades too quickly, then second guess myself, I suppose these problems can occur with a well "trained" bot, as one can override trades but that is also a situation that can be programmed with stops. The gambler mentality is to do something, constant biofeedback, akin to endorphins by getting an IM on the iPhone. I am 64 and I used to be a mainframe Assembler/ Unix C programmer for nearly 20-years. I am taking an Udemy class on python and IB API's, a little dated course for $49. The instructor is Indian and he has done a lot of these courses for Oanda API etc. I have downloaded Anaconda as well. Looking at the language is not the hard part. Programming price action will be the challenge. I will distill the best buy sell setups not using anything but prior highs and lows with determination is we are in a up/down channel, breakout or trading range. Moving average is the only indicator I care to look at. Two types of orders will be used; stop for breakout and limit for trading range.
Given that you are rolling your own solution with the IB API, 90% of your time is going to be writing code to handle the mechanics of real time trading. Go with a prebuilt platform such as NinjaTrader, SigmaTrader, etc.... and you can spend 90% of your time writing strategy code. Looks like SigmaTrader now supports Python if that is what you are comfortable with.
 
Given that you are rolling your own solution with the IB API, 90% of your time is going to be writing code to handle the mechanics of real time trading. Go with a prebuilt platform such as NinjaTrader, SigmaTrader, etc.... and you can spend 90% of your time writing strategy code. Looks like SigmaTrader now supports Python if that is what you are comfortable with.
Both Ninja or Sigma is Windows only! I have tried a Win(parallel) simulator on my Mac but it is garbage.
 
If you have never used Python, then you might want to be aware that some libraries you might want to use may not be available for the version of Python you want to use. Depending on what you try to do, this may or may not be a problem. I bought a book for using Zipline (Quantopian) for creating a backtesting enviromment (Trading Evolved by Andrew Clenow) and it was a nightmare trying to get it to work. I eventually gave up. He suggested several modifications to the Zipline source to try and get it to work.

Good luck with the project.

What modifications, Zipline is up to date for python 3 it seems and the book is new as well.
 
not sure what you mean. Just partition drive and install windows there. Problem solved. Ninja will run fine there. I tried it. Even faster than windows machines.
For security and aesthetics reasons, I'm highly Windows averse! I have been using that OS for decades and I hated it, I KNPOW I can use parallels (windows 10) on my Mac (I am leasing it now) but I cannot see why a decent software maker must force windows crap on users. Decent platforms API's usually support a Mac which is a better OS for algo trading!
https://www.quantstart.com/articles/best-operating-system-for-quant-trading/
 
Back
Top