I am thinking of this idea and curious about initial thoughts. I have significant expertise in relevant ML and statistics (for time-series, noisy/small data etc.), as well as expertise with financial markets. I know all the challenges of starting a quant hedge fund and respect Two Sigma, Rentec, Bridgewater etc. for what they have achieved and 100s of brilliant minds they employ. My goal is not to compete with them. But my hypothesis is to target low capacity. All quant hedge funds focus on scaling to high capacity (they manage $Bs), mostly targeting stocks/futures with the highest volumes. If I focus on stocks with lower volumes (e.g. in top 5000 but not top 1000), I believe there could be sufficient signal that over time, I will be able to build an edge. If I start with prototyping, and then increase the invested amount gradually over time based on the track record of success, could be reasonable. Any relevant experiences and thoughts on this?
What’s low capacity? Do you mean market cap or trading volume? Generally all AI ML in investing is bs. It’s just a marketing. However, i believe there can be something to work with for a statistician in illiquid stock market. The volumes and amount of information is low. So straightforward ml won’t work. But there’s a lot of inefficiency. Any time you have slow transactions like in credit and real estate you have a chance but the reason for illiquidity is also the reason why it’s going to be difficult. However I’d take difficult over impossible any day. Liquid markets are impossible
As someone who has spent thousands of hours on exactly this, I’ll give my two cents 1) the difference between low cap and high cap strategies is not much. E.g. if you create a system that works on small stocks, you might as well go all the way, because it will make you tens of millions. 2) it’s a lot of work, and you won’t find out if you made it until after a lot of hours (probably thousands). Very few make it. Expected value is not to try to do this, but instead work for a high TC or start a different biz
Thanks! What happened at the end in your case?
Didn’t make it to the end. I know guys who have made it in a big way, but they also had backing and help from big names
Hi
Start with paper money :-) but yes, sounds very reasonable
What's paper money
Fake money, only used to track algorithm performance