A review of "The Man Who Solved the Market" (and "the Captain")

Most systematic hedge funds are a racket; they either got lucky, or have a strategy that only works in one market regime. There's a couple of hedge funds out there who have beat expectations year after year. Ed Thorpe's Princeton Newport and its successor TGS Management is collectively one of them. Rentech is the one that lasted the longest, and is best known. They've also got the biggest bags and the principals were and are much more admirable people than the TGS principals. It helps that I actually know some of the people in the Rentech story, and people like Jim Simons are friends of a number of friends of mine, so it's something I know a little bit about on a personal level. Rentech is also remarkable for the sheer density of actual great men involved. People who accomplished great things before and after their Rentech days. Let us name some of the names: Lenny Baum, Nick Patterson, Sandor Straus, Elwyn Berlekamp, James Ax, Bob Mercer, Henry Laufer, Robert Frey, Peter Brown and Jim Simons himself. It's a lineup of machine learning, statistics and fundamental mathematics rock stars. They're all very different personalities as well: though they were welded into one of the greatest and most successful firms which ever existed. The story is sort of like the great WW-2 movie, Kelly's Heroes.

Probably the most important fact about this book is the fact that it was an unauthorized history of Rentech. As such, the people who talked were people with grievances; grievances dating back in some cases to the 1970s. I'm not sure how aware the author was of this psychological dynamic, but it was evident in the extreme to me. There are lots of specific examples I could point out where the history listed is questionable. This is valuable though, as by highlighting the social fractures, we can learn a lot about how Simons managed to weld these oddballs into a money-making machine. Imagine if Kelly's Heroes was told from the point of view of harvesting anecdotes from Oddball, Petuko, Big Joe, Willard and the German Tank Commander. That's what we have here. Actual anecdotes from Kelly (aka Jim Simons) and some of the other important characters is missing. And of course old beefs are going to seem more important to some people than they really were to everyone else at the time.

Simons always had hustle; he built a world class mathematics department at Stony Brook mostly through personal charm. He was also always a risk taker; driving a Vespa with a gang of scooter nerds from the East Coast to Columbia (oddly many of the known details of this trip are left out) before he went to grad school.

One of the amusing things about this account is the sepia toned 70s-ish of it the early days. Simons got up in the spirit of the times; being fired from the code-breaker squad for opposing the war in Vietnam, spending time on a psycologist's couch -later doing primal scream therapy and having an early marriage go spectacularly sour. James Ax was also a stereotypical man of that time; a competitive, angry, genius womanizer-misanthrope living on a boat.

The early experiences of Simons and Lenny Baum are illustrative; they started out with an actual algorithm running on a PDP-11. This in itself was a huge innovation. Baum was one of the creators of the Hidden Markov Model; a tool which has direct applicability to financial problems. I assume they were using something like this, probably looking for trending states. They had problems with it though; one must remember at the time they were inventing a lot of things. Even using data from a database in making trades was pretty innovative, let alone using decent statistical modeling in making the trades. For a while they were just winging it trading on logic and instinct, to varying degrees of success, but ultimately this wasn't a satisfactory solution for anyone. The stories are familiar to anyone who has ever tried it: discretionary trading is extremely stressful.

The next iteration, Axcom, was with James Ax and Sandor Straus. In this period the models grew more mathematically sophisticated; still using Markov ideas on Straus' rapidly growing collection of intraday data. I think Sandor Straus deserves credit as the world's first "data scientist." His account of cleaning data is probably the earliest one of performing this task. Cleaning data is the fundamental task that defines data science as a role: statisticians and economists buy clean data from somebody. The team also used a lot of Kernel Regression in this era; something I know is still an important part of Rentech and its spinoffs, but which seems to be of little interest to anybody else but me; hell I can't even get TDA people to look at it. The real breakthrough came, however, when Elwyn Berlekamp showed up, became a majority shareholder in Axcom and moved the firm into the Wells Fargo building in Berkeley. It's obvious in hindsight Berlekamp treated it as a probabilist involved in error correction codes would; developing a technique for using multiple edges in one unified trading system. Though the book doesn't say so he also probably added a rational bet-sizing system for optimizing to the geometric mean: a really sweet thing that only someone like Berlekamp would have thought of (to be fair, Thorp definitely thought of it as well). The team also narrowly avoided being caught up in a commodities broker going tits up in this era, which probably would have killed them in those days. It's good to be lucky as well as smart.

Berlekamp and Simons had a difficult long-distance relationship, as remote work wasn't a thing back then excepting for frequent phone calls. Frequent phone calls are incredibly annoying to people who are concentrating deeply. Eventually Simons bought him out and moved the rodeo to Long Island. Two important figures from the post Berlekamp days was cryptographer Nick Patterson and mathematician Henry Lauffer who were responsible for various of the innovations that we take for granted today, and a few which people would no doubt like to have access to. Robert Frey was also recruited from the Stab Art world. Another set of key hires in the 90s were Bob Mercer and Peter Brown; a couple of speech recognition specialists from IBM research (there's that Markov model stuff again), and David Magerman, a programmer also from IBM research. Taciturn saturnian Mercer and talkative mercurial Brown seemed like Castor and Pollux; opposites who meshed well together like a couple of gears, grinding out wonderful results. Brown is still CEO of the company.

Magerman, on the other hand, seemed like an asshat. He converted the company from C to C++ to make himself more valuable (a complete waste of time; 90s era C++ mostly just adds complexity for no obvious benefit over C) and blew up a live trading system by backdooring a computer. Magerman seemed to bring some computer science discipline to a company filled with sloppy-coder mathematicians and he was obviously a crucial guy who solved important problems, but the dude was a jerk. People who are good at programming are often perfectionist mindset types; meticulous people who can track down a subtle bug or manage large amounts of complexity. Unfortunately what you get with that mindset are often .... jerks. People who throw things when they don't get their way: jerks. People who think 90s era C++ was worth using, despite nobody else in the company being able to use it: jerks. People who raise hell with OSHA because the CEO is a heavy smoker: jerks. People who alienate their boss and benefactor with sperdo like "why don't you liiiike meeeee" behavior: jerks. People who have the CEO removed because he voted for the wrong political candidate: jerks.  I've known people like this throughout my career and have endeavored to always see the best in them I possibly could. Frankly his story in this book convinced me to never hire a person like this excepting as contractors. They bring bad luck, bad social interactions and you should banish them from your village. It's an astounding account in part because it must have largely been told to the author by Magerman himself.

One of the keys to its success: Rentech shared the loot. People who uncovered new alphas were important, but fixing code, cleaning data..... all received big bonuses when the company did well, which aligned everyone's incentives. Lots of work is necessary, but not so sexy, and this keeps people working on the necessary. The company, at least in the earlier days also seemed to have tremendous mission intensity; just like other type-1 organizations such as the Sidewinder era of China Lake. One of the things that didn't work so well: new employees shitting on the old employees who to their mind "didn't do anything anymore." Not sure if they ever found a way to deal with this. Probably by paying people more. One of the things which stuck out was Simons knowing what his company was worth, and taking large performance fees: Simons had after a long struggle a genuine golden-egg-laying goose, and he wasn't giving these returns away to goofballs who only wanted to pay 2 & 20. It was also amusing that many thought Rentech to be some kind of Madoff like scam; I have acquaintances who went through the interview process and thought it might have been some weird money laundry for Columbian drug dealers (he did make friends in Columbia from his early scooter trip there). Through the whole arc of Rentech, Simons had an awful lot of terrible luck in his personal life, which is really unfortunate as he seems like a genuinely nice person.

The rest of the history laid out here is boring HR drama, so I won't talk about it. It's more interesting to focus on the great years, and how they made it succeed. Big brains working together as a team, with great intensity and great rewards.

Bonus review: During one of the Simons video interviews (with James Ax's son) he also mentions a book popular back in his day called "The Captain" by Jan de Hertog. This is definitely a period piece; a sequel to a book that was a sensation under the Nazi occupation of Holland, on a young tugboat captain. Very intense, as it involved running German blockades. While it's a great read for entertainment purposes, it's also got some important leadership lessons, as he pointed out. It is a very good book for this sort of thing; being decisive, motivating very different groups of people, distracting people who need to be distracted and generally being a combination matador and stage magician. The crew in The Captain were a bunch of non-motivated odd ducks who needed to be convinced to follow the eponymous character, as well as being motivated to do a good job in general. Half of the early drama in the story was dealing with this.

Edit add: Ben Gimpert had a nice review back when the book came out, making notes of a lot of the interesting technical bits: https://blog.someben.com/2019/11/notes-on-man-who-solved-the-market-jim-simons/