Your opinions and advice please for this online combo for non-pros/non-HFT: QUANSTRAT + QUANDLE for Research in R QUANTOPIAN for System Design and Testing in Python Additionally, Quantopian is already testing IBKR execution. I am very interested in developing my Python and R chops without having to reinvent the wheel.
If you're set on using quantpian (personally I don't find it up to par), you could consider using zipline (quantopian api) and pandas. You'll be able to do the same type of vectorized backtesting using pandas alone. There's been a lot of recent progress in Python for data analysis (e.g. scikit-learn). If you need some library which isn't offered in python but in R, there's always the RPy2 module.
Good advice! I'll setup Zipline and Pandas when I get off the training wheels and try RPy2 for integration.
I've used both R and Python to build financial applications during the last year or so... R was the first language that I learned for this task during the Introduction to Financial Econometrics class in coursera... and for a while I was using it a lot... Afterwards I took several python based classes, and about 9 months ago when I decided to move to programming systems full time I started using both R and python. In any given project I would have to use python for some parts of the project... even if I was doing analysis in R, I would end up going back to python to talk to the DB, or to process large chunks of data in parallel, or if I needed to drive a web browser for crawling... and as I got better skills using pandas, numpy and scipy; I found myself going less and less to R... today I barely use R, I think that the last time I played with it was in december... and I spend my days working with Python and SQL...
Sounds like a very smooth migration. I am looking at a similar path. It seems you can be immedietely productive with R as I can surmise from some sample sessions I've done. Python would be more a generalist start for me to get programming fundamentals down. Thanks for the input.