Jim Simons founded the hedge fund Renaissance Technologies, and its Medallion Fund has been the highest returning fund of the twenty-first century. This book tells the story of Simons’ early life, academic career, limited success with his earlier investments. It also explains how Renaissance found its edge with quantitative analytics, market-neutral strategies, and short holding periods for its investments.
Early on, the company’s analysts faced a classic explore-exploit trade-off. The simple strategies they were using were so lucrative that putting time and effort into more complex models seemed like too much work:
Building formulas was difficult and time-consuming, and the gains figured to be steady but never spectacular. By contrast, quickly digesting the office’s news ticker, sutdying newspaper articles, and analyzing geopolitical events seemed exciting and far more profitable. (p. 61)
Over time the company began to collect its own price data and hire programmers and mathematicians to analyze it and build predictive models. An early insight they had was that their own trading could push prices against them, driving away their advantage. This phenomenon is known as “slippage,” and makes it difficult to trade profitably even if your predictions are correct (p. 150). In response they began to make more trades but hold for shorter time periods.
[B]uying and selling infrequently magnifies the consequences of each move. Mess up a couple times, and your portfolio could be doomed. Make a lot of trades, however, and each individual move is less important, reducing a portfoliio’s overall risk. (p. 108)
This is not the same as today’s high-frequency trading of the kind described in Michael Lewis’s Flash Boys which is more of an arbitrage strategy than a prediction of price movements (p. 222-3).
Shorter trading periods also meant that more data was available to them. There are only a bit over 100 non-overlapping time periods for annual price data, but orders of magnitude more daily and intra-day prices available for modeling (p. 246).
One notable difference between Simons’ team and other quantitative modelers is that the former were more interested in predictive accuracy rather than understanding causal relationships (p. 111). This is not to say that they would accept any signal the model picked up on, though. They still required it to have some degree of plausibility:
“Volume divided by price change three days earlier, yes, we’d include that,” says a Renaissance executive. “But not something nonsensical, like the outperformance of stock tickets starting with the letter A.” (p. 204)
The company’s employees faced two challenges once their models became so successful. One was the feeling that they weren’t really doing anything, since the trades were made without human intervention. The second was the psychological challenge of trusting the models when markets turned against them:
The goal of quants like Simons was to avoid relying on emotions and gut instinct. Yet, that’s exactly what Simons was doing after a few difficult weeks in the market….
Simons’s phone call [to Ashvin Chhabra, asking if he should switch to a short-selling strategy] is a stark reminder of how difficult it can be to turn decision-making over to computers, algorithms, and models — even, at times, for the inventors of these very approaches.
Even for the man who “solved” quantitative investing, human reactions can run counter to better judgment.