I'm a long time day trader and .NET developer, and just started using Quantopian a couple months ago to create a day trading system. Below is some feedback, questions as well as some issues I've seen thus far:
First of all, I love your system. I was not familiar with Python and your API, but was coding efficiently within a couple of weeks against your API. I will also say (...or admit) that it has been much harder to come up with an algorithm that makes money than I initially thought. When adjusting for slippage and trade fees it gets much, much harder to make money, especially when trading dozens or hundreds of trades per day. Having said that, your platform highlights this very nicely and that's exactly what is needed, and the rest is just converting my trading patterns into code logic.
Feedback, Questions and Problems:
Problem: My algo is right around 2000 lines, about 100KB right now. I seem to have bumped into a size limit and seem to get a "Error Saving Algorithm. There was a problem saving your algorithm. You don't seem logged in anymore.." message whenever I add more code. Do you guys really have a size limit on the algo code? I could keep things leaner, but I really like nicely commented code! This is kind of a biggie for me...
I'm primarily an intra-day trader coding for hundred's of trades per day and always end the day with no positions. There doesn't seem to be a way to easily keep track of gain/loss for day trade positions that I can see if you don't hold positions overnight. Is there any kind of average execution price on the order objects I'm missing and could use to calculate position gain/loss?
Feedback: I like the random Universe concept (DollarVolumeUniverse), but it would be really helpful to expand this for day trading algorithms. Very handy custom universes would be large percentage movers, and unusual volume. I know I can manually fetch this data, but it seems this would also be easy to provide as a "universe". Additionally, a 200 stock limit is a bit low if I can't efficiently filter down to applicable stocks a bit better.
Question: Pre-market data would be extremely helpful for intra-day trading (am I going against a pre-market trend, what is the volume etc.). Are there any plans to a) provide optional pre-market data for trend analysis or b) allow trading pre-market sometime down the road?
Question: I like to filter out ETF's etc. from the stocks I'm trading. Can you add some additional parameters to allow removing such types of stocks from a universe?
That's it for now. Again, great platform, great potential. Looking forward to launching my first day-trading algo in the near future.
Thanks, Beat