Hedge funds and narrow artificial intelligence

Yesterday, the Wall Street Journal published an article about Rebellion Research, a hedge fund that uses machine learning algorithms for trading decisions. Here is an excerpt:

“With artificial intelligence, programmers don’t just set up computers to make decisions in response to certain inputs. They attempt to enable the systems to learn from decisions, and adapt. Most investors trying the approach are using “machine learning,” a branch of artificial intelligence in which a computer program analyzes huge chunks of data and makes predictions about the future.”

There is also a short video about Rebellion Research.

Of course, the article says nothing about the methods used, but there were some interesting teasers. Rebellion only trades stocks, doesn’t use leverage, and doesn’t short stocks. Also, their typical holding period is 4 months and the longest holding period was 2 years. Their system is dynamic, meaning that it changes strategy with changing market conditions. All of this is very intriguing. When one hears of a fund using narrow artificial intelligence (the reasoning behind my use of the adjective “narrow” will form the basis of future posts) , one assumes that the fund is really using such methods to find pockets of market liquidity in a high frequency trading operation.

Here is Rebellion’s self-description from their website:

“Rebellion Research is a quantitative asset management firm that applies machine learning technology to generate a diversified, risk-averse portfolio of securities. Rebellion’s Artificial Intelligence evolves as market environments change, dynamically shifting in and out of sectors, countries, and investment styles.”

The WSJ article mentioned Cerebellum Capital, another hedge fund using machine learning algorithms. Here is their self-description:

“Cerebellum Capital is a hedge fund management firm whose investment programs are continuously designed, executed, and improved by a software system based on techniques from statistical machine learning.

The system is responsible for constantly creating its own new models for how the markets will move, testing those models, refining them, and learning trading strategies that take advantage of these predictive models. The system is provided with a wide variety of traditional and non-traditional, publicly available and licensed data streams as inputs to its model creation and improvement process. Cerebellum’s software system learning optimizes for a proprietary mix of expected return maximization, risk/volatility reduction across the portfolio, and portfolio independence from major markets when they trend downward. Cerebellum’s architecture for continuous improvement, self-diagnosis, and fault tolerance is based on a collective 30 years research in the area of statistical machine learning applied to real world, mission critical time-series problems.”

Here is a short article written by Astro Teller, one of the co-founders of Cerebellum.

For a complete non sequitur, Spencer Greenberg of Rebellion is a grandson of Detroit Tigers hall of fame 1st baseman Hank Greenberg.

Add to FacebookAdd to DiggAdd to Del.icio.usAdd to StumbleuponAdd to RedditAdd to BlinklistAdd to TwitterAdd to TechnoratiAdd to FurlAdd to Newsvine

This entry was posted in Narrow Artificial Intelligence and tagged , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s