Back to Community
CIO's blog post

Welcome Jonathan! Thanks for the blog post (http://blog.quantopian.com/the-foundation-of-algo-success/).

One fundamental thing I don't understand is that in the end, let's say you fund 10% of your 80,000 users. That's 8,000 individual algos. So, why would you worry so much about market and sector neutrality? Wouldn't you be able to achieve those objectives by dynamically weighting the large number of algos in your portfolio? I thought this was perhaps the original concept behind a crowd-sourced fund.

Also, I'm not clear on your move to $10M in capital and more trade-able stocks (e.g. https://www.quantopian.com/posts/the-tradeable500us-is-almost-here)? It seems like there is a preference for fewer algos in the fund, but at higher levels of capital per algo? We've gone from $100,000 to $1M to $10M. What next--$100M? It seems to be moving away from the crowd-sourced concept, as I had envisioned it.

I'd also be interested in your thoughts on, as you mention "the workflow followed by professional quants in addition to topics like researching and evaluating individual alphas, risk aware portfolio construction, and the practical aspects of strategy implementation." My guess is the user base consists of a considerable number of hobbyist, moonlighters, tinkerers...folks working nights, weekends, breaks when they can squeeze in a little time. So, attempting to map them onto a professional workflow may be challenging, but they may still have some good ideas and can contribute, but might not be able to pull together a full-up, $10M+ algo that you can sell to institutional investors. Just a thought that your experience in the real hedge fund world might not match one-to-one with the potentially limited focus and commitment that the crowd will bring to the table.

1 response

Even if your algo runs its stock selection every day, if you have a slow moving signal which is the same day over day and week over week, then your daily output is not independent and you cannot count days as bets.

Any changes you would recommend to the Quantopian platform to support this? For example, pipeline only operates on daily bars, so it seems inherently the time scale could be limited there. And is minute bar data gonna work, in the long run? Or will higher frequency data be required to compete with the big boys?