Posted on

Is Text Analysis key to Renaissance’s Success?

Jim Simons is the greatest moneymaker in modern financial history, and no other investor – Warren Buffett, Peter Lynch, Ray Dalio, Steve Cohen, or George Soros – can touch his record. 

His firm has earned profits of more than $100 billion, and between 1994 and 2004, its signature fund, The Medallion Fund, averaged 70 per cent annual return.

Medallion’s returns don’t seem to correlate with known factors and the only thing most people get to know is that the strategy is “statistical arbitrage”.

People are confounded by the fact that the proliferation of other quantitative hedge funds in recent years hasn’t caused Medallion’s performance to deteriorate.

Last year, there was a very readable book about Jim Simons: On the man who solved the markets – How Jim Simons Launched the Quant Revolution, Penguin 2019.

The book doesn’t reveal any detail about Renaissance’s advantage and people seem even more confused than before:

When finance professor Bradford Cornell first saw the annual investment returns of Renaissance Technologies’ Medallion fund, he was “dumbfounded”.

In his recent paper, Medallion Fund: The Ultimate Counterexample, UCLA 2020, he writes: 

“The performance of Renaissance Technologies’ Medallion fund provides the ultimate counterexample to the hypothesis of market efficiency. Over the period from the start of trading in 1988 to 2018, $100 invested in Medallion would have grown to $398.7 million, representing a compound return of 63.3%. Returns of this magnitude over such an extended period far outstrip anything reported in the academic literature. Furthermore, during the entire 31-year period, Medallion never had a negative return despite the dot.com crash and the financial crisis. Despite this remarkable performance, the fund’s market beta and factor loadings were all negative, so that Medallion’s performance cannot be interpreted as a premium for risk bearing. To date, there is no adequate rational market explanation for this performance.”

In the book Outside Insight: Navigating a World Drowning in Data, Penguin 2017, it has been suggested that key to Renaissance’s success is its ability to analyze text, and that the information advantage, used for trading, is created through real-time analysis of large data sets of text.

In 1993, Renaissance Capital hired two prominent computational linguists, Peter Brown and Robert Mercer.

When Jim Simons stepped down from the daily running of the business in 2009, he appointed Brown and Mercer to run the company jointly. 

Could one of the main factors to Renaissance’s remarkable results be that it systematically leverages the wealth of information on the Internet?

“The open Internet is one of the largest data sets of text. It is technically very hard to mine and therefore poorly utilized. Renaissance, with its unique team of world-class scientists, can solve this problem better than anyone else, and – by extracting insights that nobody else is able to find – they create an information advantage.”

The profile of the key people running Renaissance “is an indication that key to Renaissance’s secret recipe is its ability to analyze text, and that its information advantage, used for trading, is created through real-time analysis of large data sets of text.”

Sounds about right. Could be. Only time will tell. 

Meanwhile, the rest of us can read up about the use of alternative data and text analysis in the fund industry.

Posted on

The case for Alternative Data

AI-based strategies in the investment process are outperforming other investment strategies. Language technology enables significant alpha creation, as proven by the most successful hedge funds. Strategic initiatives are moving toward advanced analytics of both structured data (e.g. price data) and unstructured data (e.g. language data), in combination – to be used in mainstream asset management. On the back of Advanced Analytics, Machine Learning, Computational Linguistics, and advanced qual-to-quant tools, Alternative Data is growing rapidly and is predicted to be a major game-changer in the investment industry as a source for alpha creation. Compared with high-frequency trading, the edge in some Alternative Data is continuous in a complex system, and the Alternative Data has – therefore – much higher potential and more durable value (i.e. it is not an arbitrage). A major challenge for the industry is the lack of robust, consistent and long historical time-series needed to train machine learning models on how to create alpha from language data: There are rafts of new language technologies, investment and strategic initiatives to build trading models and leverage alternative data, but there is little or no training data. Firms are beginning to set up targets and they begin to accumulate data in multi-year projects, with the understanding that it takes several years to accumulate enough data to build predictive models that are robust enough for practical investment applications.