A front page WSJ article on James Simon's Medallion fund, which has averaged a 45% return since its inception in 1988. Truly amazing. No one really knows how they do it, but it is rumored to employ some sort of quasi-market making, 'pairs-like' algorithm. That is, they trade a lot electronically, seeding the book and provididing liquidity (eg, limit orders to buy at the best bid or lower), and looking for abnormal movements with sympathetic stocks (eg, when IBM goes up, but Dell does not, sell IBM). This would explain why they are not really scalable ($10B Medallion has been closed to new investors for a long time).
But who knows. I hear they don't hire many economists, and instead prefer PhDs in Math, computer science, etc. This could be a slam on economists, in that we aren't as bright as these guys, or it could be our framework has been poisened by modern econometrics. However, it could be economist's are savvy enough to see the essence of the alpha, and so would be a bigger threat via leaving and starting a competitor, whereas the math PhDs are too focused on the little issues to grasp the bigger picture.
I did hear once (third hand) that in their legal dispute two ex-employees, the intellectual property at issue was quite trivial, which would suggest their alpha is merely being the fastest electronic market makers in the world. Someone has to be fastest. Considering the new co-CEOs are both computer scientists, that would make sense.
It is interesting that they started two other funds with a larger scale, hoping to make money off longer term strategies in futures and equities. These have not even outperformed the benchmarks, let alone displayed the amazing Medallion-type dominance. This highlights another reason to hide one's alpha, among the many (read Finding Alpha for more!). If your firm is in industry X, and makes money doing something very parochial within X, don't be too specific. This way, when you extend your brand, no one thinks twice about the plausibility that their new venture is totally different, and so customers are willing to give it a shot. It's possible Renaissance's alpha is in subtle price change correlations, a skill that would translate to other