Interview with James Simon

Jim Simons of RenTech

Very interesting interview with James Simon of Renaissance Technologies, what might be the most successful hedge fund to date. It is also one of the most secretive. This is an hour long and quite a treat. (the good stuff is between minute 25-45)

Normally I would have just posted this on LargeCapLinks as an interesting video to watch when you’re bored but there was too much good stuff in here:

No one knows exactly what the hell they do. They boast 5% management fee and 44% of profits and still average 35% to investors in Medallion funds so whatever it is, it must be good? He outlined what may be their overarching strategy:

1) Machine Learning – This is the algo that picks what to buy and what to sell. Machine learning basically takes a bunch of time-series data you think are related to the price of the security you’re trading. You tell the machine the optimal times to buy & sell, and then let it take what it’s ‘learned’ and apply it to future data. So for example, if I think the price of the S&P is some function of USD, Interest Rates, Oil, and DRYS stock price (inside joke there), I would feed the time series in, along with the S&P optimal buy/sell signals that I create (bottom and top ticks naturally), and train it on a portion of the data. Then feed it new data and see if it can continues to time buy/sell decisions optimally. This is structured machine learning.

It’s then a process of feeding the learning algo the “right” data, testing it, optimizing it, and repeating the process. DRYS stock price might not have a huge amount of relevance in predicting the forward returns of the S&P, shocker. Maybe we should try time of day, phase of the moon, weather. And actually that might not be too far off, in More Money than God, Robert Mercer provided the following:

In one simple example, the brain trust discovered that fine morning weather in a city tended to predict an upward movement in its stock exchange. By buying on bright days at breakfast time and selling a bit later, Medallion could come out ahead – except that the effect was too small to overcome transaction costs, which is why Renaissance allowed this signal to be public.

It used to be said that if a team from the original National Football League won, the market would head upward. As a matter of statistics, this relationship might hold; but as a matter of common sense, it is a meaningless coincidence.

“Some signals that make no intuitive sense do indeed work.” Indeed, it is the non intuitive signals that often prove the most lucrative for Renaissance. “The signals that we have been trading without interruption for fifteen years make no sense,” Mercer explains. “Otherwise someone else would have found them.”


2) Market impact and costs are very important. This part of the interview may have been light on details but the tone and inflection signaled to me that this is almost as important or perhaps more so than signal generation. In my experience it’s actually quite easy to create a highly profitable strategy in backtest, just assume commissions and slippage are 0. In fact, this paper the argument is that most returns in financial literature are garbage due to incorrect assumptions on market fills.

I wrote an article some time back on the tax ‘arbitrage’ employed by RenTech which certainly fits the theme. And further in the interview he discusses about general business and alluding to the focus on revenues over cost. Costs people! Costs!!

3) Minimize volatility. I would guess this means portfolio volatility? Perhaps optimizing strategies for reduced vol over sharpe ratio? Creating minimum variance portfolios are all the hype these days. He also says this is one of the areas with some sophisticated math. Fwiw, the python library BT has a really nice weighing algo to optimize position size based on inverse of vol and mean-variance optimization.

4) Limiting AUM to the sweet spot, Medallion has about $3.3Billion so they are certainly no Bridgewater.

5) Building the right team. Hire smart people, give them lots of freedom, make them talk to everyone else, let them share profits as partners, and provide them with the best infrastructure and data you can. Sounds very Steve Jobs? As Simon talks more about it, you can basically tell that he didn’t even come up with most of the strategies. In fact, many people attribute the early success to another employee, Elwyn Berlekamp. If interested, another decent book to check out is The Quants. He does say he has a good personality to ‘manage’ smart people and pick talent. Also very interesting in that he had no idea of the outcome of his hires, they had little to no financial experience, nor did he expect to. When everyone shares in the profits, you’re glad to help out.

In unrelated news, I’ve transfered the blog from to a new hosting running software. Yes, I didn’t know there was a difference for awhile! It’s a good way to learn some CSS/PHP so I’ll be playing around with the themes, layouts, etc. If anyone is curious, installing and transferring the blog was much simpler then I thought using bluehost.

Please let me know if you see anything you like/hate/errors/omissions/etc. Hopefully to all those subscribed, you’re still getting these!



Be the first to comment on "Interview with James Simon"

Leave a comment

Your email address will not be published.