In his mid-to-late 30s, Simons grew frustrated with mathematics. He and collaborator Jeff Cheeger were stuck on a problem for two years with no progress. "It's okay to work for a long time if you feel you're getting somewhere, but we got absolutely nowhere," he recalled.
Meanwhile, he had some family money to invest and found it interesting. He started a small trading operation with a few partners, initially using no mathematical models - just "fundamental" trading like everyone else. For the first two years, they were "extremely successful" - which Simons attributed entirely to luck.
But analyzing price patterns, he noticed structure in the data. There appeared to be anomalies - patterns that could be exploited. He began to see trading as a code-breaking problem, similar to his NSA work. Could he build algorithms to predict market movements?
He started hiring mathematicians and building models. The firm would eventually become Renaissance Technologies, though it would take years of development before models completely replaced human judgment.