I want to share a fantastic Bloomberg Businessweek piece on the Medallion Fund by Katherine Burton.
Unless you work in finance, you may not have heard of the Medallion Fund, but if you pay attention to politics, you’ve probably heard of Robert Mercer. Mercer is often talked about as a big-money Republican donor, but that’s probably the single least interesting thing about him. For instance: Even though he sits at the head of the Medallion Fund, he is not a finance guy—he cut his teeth as a computer programmer at IBM doing speech recognition and machine translation. During that time he disappeared from his team for several months to type the conjugation of French verbs into a computer. Also, he rarely speaks out loud.
Anyway, the Medallion Fund is an investment fund open only to the 300 or so employees of Renaissance Technologies. Renaissance is a quant-driven hedge fund. It is powered not by value-investing or a belief in matching value with capital, but by mathematicians and computer programmers (as well as astrophysists and string theorists) looking to find tiny pieces of arbitrage in the global market systems. Roughly a third of its employees have doctorates.
How good are these quants? According to Burton, the Medallion Fund has made about $55 billion–that’s with a “b”–in profits over the last 28 years, with annualized returns of about 80 percent per year. As MIT’s Andrew Lo tells Businessweek, “Renaissance is the commercial version of the Manhattan project.”
Renaissance was founded in 1982 by Jim Simons, a math prodigy who had held professorships at both MIT and Harvard. He had also worked as a cyrptographer for the Institute for Defense Analyses. Simons was chairing the math department at SUNY Stony Brook in the late ’70s when he decided to try investing. He quickly abandoned thinking about supply and demand in favor of looking for more abstract patterns in market data. He brought in some of his math and crypto buddies from the IDA to help him build models. One of them Elwyn Berlekamp. “I was confident that the models would work better,” Berlekamp told Burton. “I didn’t think they would be as good as they were.”
