Machine Learning For Stock Trading Strategies
In previous articles, we’ve defined some of the terms being thrown around lately like “machine learning” and “artificial intelligence“. These disruptive technologies will soon change the world as we know it. While some pundits predicted that we were years away from a computer that could beat a human expert at “Go”, this achievement was recently announced. If a “deep learning” program can now beat a game that has more possible moves than atoms in the known universe, then what’s stopping us from unleashing it upon the stock market and making millions?
The idea of using computers to trade stocks is hardly new. Algorithmic trading (also known as algo trading or black box trading which is a subset of algo trading) has been around for well over a decade and rapidly gaining in popularity. Here’s a look at algorithmic trading as a percentage of market volume:
If that trend continues, then this means that today upwards of 90% of trading is being conducted by computer programs. One thing to notice about algorithmic trading is that it has been moving in the direction of shorter and shorter holding times. High frequency trading (HFT) is a subset of algorithmic trading in which stocks are bought and then sold in fractions of a second. This strategy is a form of arbitrage in which the HFT algorithm spots a price discrepancy and then quickly capitalizes on it. As you would expect, HFT trading profits are becoming smaller and smaller but the volume of trades are still dominating the overall market:
Now that we know about algorithmic trading and HFT, just how does machine learning or deep learning come into play? To answer this question, the important variable to take into account is duration. While HFT and algo trading perform trades of a short duration, it becomes much more difficult to “capture alpha” when you start increasing the time frame. The reality is that some of the world’s biggest hedge funds are already all over this space and have been capturing alpha across many durations for a long time now using machine learning.
Early last year, Bridgewater Associates which has $150 billion in assets under management (AUM) started a new artificial intelligence unit led by David Ferrucci who led the development of IBM’s Watson. After working at IBM for 17 years, he was poached by Bridgewater in 2012.
Another firm called Renaissance Technologies has $65 billion in AUM and is said to have “the best physics and mathematics department in the world”. The Medallion Fund at Renaissance, run mostly for employees of the company, has one of the best records in investing history having returned +35% annualized over 20 years. The two co-CEOs of Renaissance were both hired from IBM Research in 1993 where they were working on language-recognition programs.
With $32 billion under management, Two Sigma Investments is known for using AI and machine learning as a key part of their strategy. One co-founder did his PHD in artificial intelligence at MIT and the other was an International Mathematical Olympiad Silver Medalist. Being a finance professional is not a requirement to work at this firm.
While hedge funds such as these 3 are pioneers of using machine learning for stock trading strategies, there are some startups playing in this space as well. Binatix is a deep learning trading firm that came out of stealth mode in 2014 and claims to be nicely profitable having used their strategy for well over three years. Aidyia is a Hong Kong based hedge fund launched in 2015 that trades in U.S. equities and makes all stock trades using artificial intelligence with no human intervention required. Sentient, another deep learning company we discussed before, has developed an artificial intelligence trader that was successful enough that they are consider spinning it out as a prop trading company or asset management firm.
If there’s a startup that shows promise in this space, you can bet that the 3 well established hedge funds we discussed know about it. If you had a machine learning algorithm that generated alpha, would you tell the world about it? Most likely not. But then how would you raise the capital needed to make some serious money off of your strategy? Firms like Bridgewater can be as nimble as any startup and at the same time have $150 billion in capital to play with. It’s hard to compete if you’re a startup that’s trying to get funded. If you’re looking for investors, you have to disclose what you’re doing. Word travels fast. It’s not hard to see hedge funds like Bridgewater poaching talent from AI startups that are trying to play in this space and quickly finding out what they’re up to.
For retail investors to take advantage of machine learning for stock trading, you have a couple directions to take. For ultra high net worth retail investors, you can invest your money in one of the hedge funds using AI like Bridgewater or Renaissance. For those of us who don’t have such large amounts of capital, we can wait for deep learning companies like Sentient to go public or be acquired and then invest in those vehicles. We’ll be keeping a close eye on this space because frankly, it’s just fascinating.
If you pay more than $4.95 a trade, you're paying too much. Ally Invest is one of the lowest-fee brokers around so you spend less money on transaction fees and more on stocks. With more than 30 trades a quarter it drops even lower to $3.95 a trade. Open an account and begin trading today.