Hi all,
I'm about to start writing my thesis in university and I am hoping to use the open source technologies on Quantopian for the bulk of my quantitive analysis. I've spent quite some time experimenting with the different features but had a few questions that I was hoping to get clarified.
I've already made a very simplistic long/short momentum algorithm but I (purposefully) didn't address position concentration constraints in it. I was wondering how, roughly speaking, does the algorithm allocate capital when the objective (using the optimizer) is set to MaximizeAlpha (which is just the historical return over a year) and the pipeline returns the top 250 winners and losers (so 500 stocks)? Is there some kind of "rule of thumb" that describes the capital allocation process of the order_optimal_portfolio() call in the absence of position concentration constraints? (the DollarNeutral constraint is also present, if that matters)
What is the easiest way to get some t-stats and standard errors on the coefficients of the HML, SML and UMD factors AND/OR the Quantopian Risk Model factors, using Quantopian's online jupyter notebooks? A related question: Isn't there likely a multicollinearity issue with the Quantopian Risk Model using 11 sector factors (which are probably quite correlated) instead of an overall market factor?
Is there any way I can download the return series of the portfolio/algorithm so that I can do further analysis with STATA locally on my computer? (A lot of econometric techniques I intend to use are probably very possible with Python but I think it will be easier for me to do it with STATA because of my familiarity with the program)
Thank you all for your time.
