In an interview he said he finds anomalies in price action using mostly statistics and some probability theory mathematics. Of course he doesn't go much into detail, but just gives an example that trend following is an example of an anomaly, but gives no juicy secrets away. He also said he had no models the first 2 years he traded, but was extremely successful which he thinks was luck.
I wish he'd write a book ....I just finished Ed Thorpe's book, and I'd imagine a Simons book would be equally interesting