value and opportunity

This is another book that I have been waiting for to read for some time. Jim Simons is maybe not a household name in investing, but his Renaissance Technology fund (Rentech) has clearly on of the best track record of any investment vehicle in recorded history. However, other than some other famous investors, Rentech was (and still is) so secretive that hardly anyone had a clue how he/they did it.

Plus, I have read already two other books of the author, among them “The Frackers“, a book about the fracking boom that I liked a lot.

This book sheds some light of what contributed to Rentech’s success but (Spoiler Alert) doesn’t really explain how it was possible to create 66% p.a. return (before fees) over a 30 year (!!) time period. However one caveat: This is not a compounded rate of return. Rentech always limited reinvestment and since 2009 the fund size is held constant at 10 bn USD.

Jim Simons started out as a mathematician that quickly acquired fame due to his mathematical talent. After some work as a “code breaker” in a government agency, he built up a world class math department as a professor at the University Stony Brook on the East Coast. However after some time, Simons got bored and started “playing” the markets. In 1978 he finally left the university and started his own fund at the age of 40.

Interestingly in the beginning, although he started to use some statistical models, he still “played” especially the futures markets. At first he teamed up with another mathematician called Lenny Baum and they started doing some relatively simple trend investments in currencies. These approaches didn’t work that well. They improved their methods but always had setbacks such as involuntarily cornering the potato futures market, and in the early 80s Simons actually started to do some early stage technology investing within the fund. However with other mathematician coming in, especially a guy called Sandor Strauss, Rentech began more earnestly to systematically collect huge amounts of data and to rely much more on quantitative models. But only in the late 80s, the fund finally morphed into a fully fledged quant fund and they started to create a more comprehensive model, with the final move then into equities which allowed a much bigger fund volume.

Another big push forward came when they hired two members of IBM’s speech recognition group (Mercer and Brown) which seem to have moved Rentech even one more level up, mostly with their programming skills and Machine Learning capabilities.

My personal interpretation on what made RenTech so successful is as follows:

• Simons was clearly not the lone genius, but much more like a coach who created a culture of very collaborative idea sharing and was able to assemble ultra high class talent
• The idea of using exclusively scientists with no economic background allowed them to quickly move to pure quantitative methods without the need for any economic explanations and throw traditional knowledge over board
• they focused much earlier than anyone else on gathering as much historical data as possible and later, to gather intra-day data earlier than others
• they seem to have applied very early on much more complex mathematical models, such as hidden Brownian motions plus some early versions of machine learning
• they created a holistic portfolio risk management system early on with millions of lines of code which morphed into an automated trading system, handling hundreds of thousands trades a day.
• This system seems to be able to include any kind of information and test it against all the other factors in the model in order to determine unusual and complex correlations that can be traded
• the system was not the result of a lone genius but a collaborative effort of a changing cast of highly talented scientists
• they focused early on on very short term (intra-day) strategies and on large numbers of trades to minimize the risk on any single risk factor
• they were where much aware on how fund size could influence trades and were very careful to limit the fund size in a way to be able to still make money.

What I like about the book is that the author also shows that just putting a few scientists together does not create success automatically, as seen by the spectacular failure of quant fund LTCM in 1998 whose team included several nobel prize winners, however mostly economists….

One of my favorite quants, Ed Thorpe also appears in the book, once as an almost investor into Rentech and again as a fellow quant who had to close his fund however early due to SEC issues.

Overall, according to the table at the end of the book, Rentech’s Medaillion fund so far has produced more than 100 bn trading profits which mostly accrued to employees as outside investors were kicked out of the fund many years ago, Simon alone seems to be worth more than 20 bn and former CEO Mercer has enough money to have funded Trump’s election.

Interestingly, some of Rentech’s employees questioned themselves why they were making so much money and who was paying their bill, but they didn’t seem to fully understand it and agreed that this was compensation for providing liquidity.

Overall, this is a fantastic book in my opinion and a “must read” for anyone interested in financial markets, despite the fact that obviously the “secret sauce” of Rentech’s profit has not been revealed.

The learning for any private investor might be the following: Don’t try to beat the market with short term trading in liquid securities or futures as you will pay your “taxes” to some very smart algos that know a lot more than any individual can ever know.

Maybe one final remark about Rentech’s track record: The author compares the 66% p.a. (or 39% after fees) with other track records like for instance Warren Buffett. I think this is not totally fair, as no one was able to actually compound at that rate Profits were distributed and the fund investors especially in the later years had to reinvest elsewhere, so we do not know what their actual compounded returns were but clearly lower than 39%. Buffett’s shareholders however compounded at exactly the stated returns.