I have been live-trading, with my Fidelity IRA account, using the signals generated by the models as introduced in Forecasting and Timing Markets: A Quantitative Approach (ASIN:B0875JBWBQ ). Started from March 09 this year, I have achieved a net profit of 24.54% as of April 24, which is impressive given the market volatility induced by the COVID-19. Coincidentally, I learned from an online post that Simons's Medallion Fund also achieved an over 24% return during this same period of time. I was motivated to find out more about his Medallion Fund and thus bought this book.

I eagerly read through the entire book so that I could assess how different his quantitative approach is against the AlphaCovaria System I have been relying on as mentioned above. I am so grateful for Mr. Zuckerman who dug out so many details about how Simons's models have been built. Here is a summary of what I have learned from a quantitative trader's perspective:

(1) First, a little background. While at IDA during his earlier career, Simons and his colleagues wrote a research paper that determined that markets existed in various hidden states that could be identified with mathematical models. At IDA, they built computer models to spot "signals" hidden in the noise of the communications of the United States' enemies. This was the precursor to Simons's later persistent pursuit to testing the approach in real life.
(2) Performance-wise, Simons has been the most successful one in trading, given the performance comparisons of this list: Jim Simons (Medallion) 39.1%, George Soros (Quantum Fund) 32%, Steven Cohen (SAC) 30%, Peter Lynch (Magellan Fund)29%, Warren Buffett (Berkshire Hathaway) 20.5%, and Ray Dalio (Pure Alpha) 12%. One of the factors that Simons could succeed so much is that he is a strongly principled person with a strong belief in "Work with the smartest people you can, hopefully, smarter than you... be persistent, don't give up easily." So he is not only a great mathematician but also a great visionary and business manager.
(3) Their model dev process: By 1997, Medallion's staffers had settled on a three-step process to discover statistically significant moneymaking strategies, or what they called their trading signals: (1) Identify anomalous patterns in historic pricing data, (2) make sure the anomalies were statistically significant, consistent over time, and nonrandom , and (3) see if the identified pricing behavior could be explained in a reasonable way.
(4) Trading frequency: Medallion made between 150,000 and 300,000 trades a day, but much of that activity entailed buying or selling in small chunks to avoid impacting the market prices.
(5) Data granularity: They use five-minute bars as the ideal way to carve things up. Their data hunter Laufer's five-minute bars gave the team the ability to identify new trends, oddities, and other phenomena, or, in their parlance, nonrandom trading effects.
(6) Holding period: Medallion still held thousands of long and short positions at any time. Its holding period ranged from one or two days to one or two weeks. The fund did even faster trades, described by some as high-frequency, but many of those were for hedging purposes or to gradually build its positions. Renaissance still placed an emphasis on cleaning and collecting its data, but it had refined its risk management and other trading techniques.
(7) Their performance as measured by Sharpe ratio. 1990s, Medallion had a strong Sharpe ratio of about 2.0, double the level of the S&P 500. But adding foreign-market algorithms and improving Medallion's trading techniques sent its Sharpe soaring to about 6.0 in early 2003, about twice the ratio of the largest quant firms and a figure suggesting there was nearly no risk of the fund losing money over a whole year. No one had achieved what Simons and his team had-a portfolio as big as $5 billion delivering this kind of astonishing performance. In 2004, Medallion's Sharpe ratio even hit 7.5, a jaw-dropping figure. Medallion had recorded a Sharpe ratio of 2.5 in its most recent five-year period, suggesting that the fund's gains came with low volatility and risk.
(8) Their portfolio composition. They started with commodity, bond, and currency, but later expanded into equities, which became the major source of profits after many years of efforts.
(9) Does Simons strictly stick to their models? In general, yes, but he made calls when he saw models were malfunctioning due to extreme market conditions.
(10) How have their models worked under various market conditions? Their models are mostly neutral, which was made possible by making quick trades only to eliminate unforeseeable events. They claimed that they could make models that would work with long-term investments, but it seems that they have not done so.
(11) What is the most secret juice with their models? Medallion found itself making its largest profits during times of extreme turbulence in financial markets. They believed investors are prone to cognitive biases, the kinds that lead to panics, bubbles, booms, and busts. "We make money from reactions people have to price moves." They look for smaller, short-term opportunities-get in and get out. The gains on each trade were never huge, and the fund only got it right a bit more than half the time, but that was more than enough. "We are right 50.75 percent of the time... but we're 100 percent right 50.75 percent of the time," Mercer told a friend. "You can make billions that way."
(12) How long was their learning curve? Simons spent 12 full years searching for a successful investing formula, without much success until he and Berlekamp built a computer model capable of digesting torrents of data and selecting ideal trades, a scientific and systematic approach partly aimed at removing emotion from the investment process.
(13) Size of their computing infrastructure"‹. On page 248, it says their computer room was the size of a couple of tennis courts. I arrived at a guestimate that they might have about ~13,000 servers, computed like this: 2x78x27 (two tennis courts) x 0.6 (total area occupied by racks) / (2x4 (rack area)) x 40 (servers per rack) = 12,636. This should not be too far away from what they have.

I strongly encourage every serious quant to read through the entire book for a lot of other secret juices.

My college roommate's brother was completing his PhD and called me to ask what I thought of the offer he had to join Renaissance. I advised him that they were in all probability a fraud and he should get a real job at a real Wall St. company. Thank goodness he did not take my advice. He's done OK, and so have they, of course.

Ordered the book without asking him what he thought of it. Glad I did, if you're from my business you'll probably enjoy it a lot. It was a relief to read an employee also thought they were probably a fraud, only to sit down and audit the numbers and realize they are actually 100% genuine.

I did not buy this to find out who worked with Simons and what their background was and how they got on with each other and with their wives; and, if I'm honest, the names are so many and their quirks so mundane, that I lost track. A "cast of characters" page somewhere up front would probably have been helpful. And all the business about Mercer and Bannon I could have done without, it hardly belongs.

No, I bought this to found out how they did it.

Turns out they rode the market up: their returns have been stellar forever, but prior to 2003 it was on the kind of money on which you can do RV. Since then it's been on the kind of money where you need to be long outright, and that's what they seem to have done, or else they would not have suffered sleepless nights on the rare occasions when they did. I lived those days myself, I know what I'm talking about.

But they rode this bull market better than anybody else and in bigger size than anybody else, so kudos to them, well done!

As for the book, if it had not answered that burning question of mine, I don't think it would have been of enormous interest, I'm not too deeply interested in the private lives and habits of miserly billionaires who bring packed lunch to work. That is not to take anything away from the book, however, and the speed at which I read it indicates I probably liked it plenty more than I care to admit.

Fans of the financial markets will be fascinated by the story of an obscure academic who quits his comfortable life to toil for years as an investing also-ran but then cracks the code of the markets on to becoming the most successful investor of a generation. A child prodigy who graduated MIT by the time he was 20, Jim Simons would go to on become a US code breaker and a successful theoretical mathematician. Ultimately his risk averse, middle class lifestyle would gnaw at him. The author wittily pillories the life of an academic by reprising the joke, "What's the difference between a PhD in mathematics and a large pizza? A: A large pizza can feed a family of four."

While not destitute, Simons pined for something bigger. He felt he could figure out the market but getting there was a convoluted process. Gregory Zuckerman's analysis of the lives involved and tactics used make for an excellent read. The foundational theory would come from a 1960's paper Simons wrote calling for an unemotional approach that favored pattern fitting the different states that the market seemed to periodically enter.

The first big step was hiring Lenny Baum. Baum applied his specialty in Markhov models that use the most recent price data to make strong approximations on the future. At the same time, Simons knew he would need larger data sets so hired a Cal Tech grad named Greg Hullender and later fellow Stony Brook mathematician Sandor Strauss. After some departures, Stony Brook mathematicians Jim Ax and Henry Laufer strengthen these models and, with the help of UC-Irvine's Rene Carmona, added the concept of kernel methods (an early machine learning process that sought out complex patterns and correlations). A big breakthrough came in 1990 with Berkeley professor Elwyn Berlekamp's ability to pick up on minute oddities in the market. He was a leading force in convincing Simmons to stop worrying about why anomalies existed and to just profit from them. Such data overfitting, or trying to explain too much, is a big quantitative hurdle for many. The author has this great anecdote to show the issue:

Quant investor David Leinweber later would determine that US stock returns can be predicted with 99 percent accuracy by combining data for the annual butter production in Bangladesh, US cheese production, and the population of sheep in Bangladesh and the US.

Soon after, in 1990, Medallion would gain 55.9%. From the 1988-2008, after fee returns would be 40%.

The firm's success further evolved with the hiring of Peter Brown and Bob Mercer from IBM. Profiting on retracements, when stocks are overbought or sold, and finding patterns in unexplainable or odd patterns became their forte. Finding such quantitative peculiarities created a moat for their algorithms as other investors simply tried to but could not explain such mispricing. Mercer summed up the strategy simply, "we're right 50.75 percent of the time . . . but we're 100 percent right 50.75 percent of the time."

Interestingly, during this time, their models actually underperformed as minor bugs like the hardcoding of the SP value held them back. Once fixed, Renaissance would enhance their models to learn from correlated assets in the market, per one insider:

"This interconnectedness is hard to model and predict with accuracy, and it changes over time. RenTec has built a machine to model this interconnectedness, track its behavior over time, and bet on when prices seem out of whack according to these models."

The firm also developed basket options which allowed them to cut the downside of their trades and proved to be a more tax efficient structure.

Zuckerman does a great job bringing up other competitors. LTCM, who had a similar approach until their demise, typically double downed on their losing trades. Renaissance on the other hand tended to cut risk and used less leverage. David Shaw, a former academic and Morgan Stanley alumni, also competed in the same style. He took seed capital from Donald Sussman's Paloma Partners and grew his firm into a formidable competitor. Interestingly, one early DE Shaw programmer was Jeff Bezos. Simons' neighbor George Soros and his lieutenant, economics PHD Stanley Drukenmiller, are profiled for their analytical macro approach. And fascinating to hear that Simons lamented the steady success of equity options trader Berne Madoff (pre-scandal).

The most interesting part is the waxing and waning support Simons has for his models. The firm's success hinged on developing models whose output they could not explain. However, during the LTCM and 2008 crises, Simons decided to ignore the models and lessen his risk in order to survive. There is a great scene where while vacationing, a retired Simons calls up his money manager to see if they should hedge since the market seemed volatile. This from a man who made billions solving the market.

I would not be surprised to see if there is a movie based on this book. I like author provides some background of characters like their childhood or family which helps to better understand their life decisions. I was expecting to get some clues about algorithmic trading but I got to know probably the most interesting untold background story of the strange presidential election. I think the most engaging part of the book is the last chapters where these money funded ideological intrigues were revealed.
Unfortunately, at the end of the book, you still don't know much of the market inefficiencies identified by these quants and that have created enormous wealth for Mr. Simons and his team. But, the fact that these were kept secret for so many years and the innovative machine learning approaches were successfully developed and used in trading first time indicates Mr. Simons not only brilliant scientist but also very successful businessman and also great humanitarian (where he's channelled some of his wealth for meaningful causes like autism research, understanding origin of universe rather than cambridge analytica to replicate lies to manipulate millions on scale with the help of no-funny felonious gru and associated clowns.)

Note that I said "business" and not biographical, trading, investing, or politics. In reality, this book is all of those.

I don't understand why some reviewers complain about the book not revealing any trading secrets. What were they expecting? The master code driving all RenTec trades? I think the book gives enough details for anybody to understand what these guys were into as they developed their trading systems. I know I personally would have loved having this book in my hands 8+ years ago. It would have saved me a lot of time. That's because a book like this helps you understand how far behind you are in the process of developing a trading system (and, as a corollary, how impossible it would be for you -an individual investor- to beat these guys). We are talking about an army of PhDs using machine learning and speech recognition models to try to identify market patterns...in the 80s!!! Today, they have (real time) data on every potentially measurable thing on earth (and out of it) and they are putting that on the hands of the most talented people with the most advanced techniques! The most appealing thing of all is that they are not even high frequency traders. They are an investment firm (i.e. they are on the buy side). They don't play with the advantage of knowing how orders are coming in and trading against that. Now, that's probably the only advantage they don't have. There are a lot of very popular quantitative funds out there and their returns are not even close to those of RenTec. It seems as if this was yet another example of a the winner-takes-all situation. So, RenTec would be like the Microsoft of the trading firms.

As I'm writing this, we are in the middle of the coronavirus lockdown in the US. It's April 6th and some are saying that we are hitting the apex. Hopefully that's the case. Anyways, the reason why I mention this is because anyone who's been following the markets in the last few years know that the last couple of years have been extremely volatile compared to the previous 10 years. So, reading a book like this in that context is a extremely humbling experience. It reinforced my conviction that the most an individual investor with limited time and a full time job can do is asset management. By that I mean that it will be very difficult for a retail investor to beat the S&P 500 on an absolute basis. So, the best they can do is to try to beat it on a risk-adjusted basis (based on metrics like the Sharpe or Sortino ratios for example).

So, back to the book. Before Jim Simons started his firm, he had been working for the government as a code breaker. Apparently, the statistical/mathematical techniques used in code breaking are similar to those used in speech recognition (which is what Mercer and Brown were doing for IBM before joining RenTec). In particular, they were using Hidden Markov Models to predict sequences of data. In other words, they were using an extremely advanced form of Technical Analysis. And that was just when they were starting in the 80s.

At the beginning of this review I said that this could also be tagged as a book on politics. And that's because of the critical role that Bob Mercer played in the last presidential elections. It would not be far-fetched saying that Mercer is who made Donald Trump president of the US. The author goes into a lot of detail describing the developments that took place during the presidential campaign. It made me aware of how flawed the system could be. Mercer just happened to be a shy and nerdy scientist how found himself so rich that he could spare millions of dollars financing his libertarian hobbies. Unfortunately, he was smart enough not to fund the libertarian party but the Republican one. My final take on him is that he was just a conservative and racist person.

Finally, I found the last chapter of the book really interesting. In there, the author explains the impact that these quantitative funds are having on the money management industry and the market as a whole. He does seem to confuse an important concept though. He likens "active" managers to "traditional" managers. And he confronts active managers vs quantitative ones. As everyone knows, quantitative funds can be actively managed (RenTec's Medallion fund is actually a perfect example). The only difference is that -in these cases- the trading decisions are made by an algorithm, not a person. On the other hand, you can perfectly have traditional/discretionary managers that are quite passive (Warren Buffett). In any case, this is actually not that important since the reader can still get the main point the author is trying to make (which is that discretionary managers are nowadays outperformed by systematic/quantitative algorithmic funds). He actually gives a bunch of good examples as to how this occurs.

In short, a very enjoyable and interesting read.