Markov Chains and Baum
A hidden Markov process is one in which the results of the chain are visible, but not the states that help explain the progression of the chain. Like receiving the number of runs scored in each inning without knowing anything about baseball.
The Baum-Welch algorithm could estimate probabilities and parameters using a little more information than the results of the hidden Markov chain.
Baum, hesitant at first, joined Simmons and applied his knowledge to trade currencies. Ax (recruited at Stony Brook) extended the models to commodities. Monemetrics was making bank!
Simons hired Hullender to program the computers. Hullender was being kicked out of Cal Tech for an unauthorized high-stakes trading operation out of his dorm room. With Hullender, Simons could make a pure system that was free of human interference.
Simons collected data from the World Bank on commodity, bond and currency prices going back to decades. He had a staffer painstakingly write down interest rate histories from the Federal Reserve office. Closing prices of major currencies were noted.
Piggy Basket Loves Potatoes
Named the “Piggy Basket”, the system mainly used momentum strategies, with a bit of correlation tests. It had $1m. The Piggy Basket made recommendations, while humans executed the trades - it was the best they could do.
Piggy Basket made bank for a few months. It then developed an appetite for potatoes, making Monemetrics almost corner the global potato market. Regulators closed down the positions, costing Monemetrics. Simons and Baum lost confidence in Piggy Basket.
Simons and Baum switched to identifying undervalued investments while reacting to market news. They even hired a French Stony Brook student to translate an obscure French financial newsletter to get ahead.
1982 - Monemetrics becomes Renaissance Technologies Corporation. Simons had a thing for tech companies.
Ever Optimistic Baum
Baum relied on intuition and instinct, and a buy-low-and-hold-forever theory. Baum racked up $43m in profits between July 1979 and March 1982 - ~2x his original stake from Simons.
In Fall 1979, they bought gold futures contracts at $250. Geopolitics jitters pushed gold prices higher. People were queuing up to sell jewelry. Baum was sitting on more than $10m of profits. Simons thought a crash immenent. Simons forced Baum to bitterly sell at $800. Gold went to $865, but came crashing to less than $500 a few months later.
With inflation sort of under control and the feds predicting decline in interest rates, Baum purchased tens of millions of US bonds. Reagan issued a lot of bonds and the economy was growing rapidly - panic selling dominated. The Yen also stopped appreciating. Baum’s positions were down 40%. At this point, Simons sold Baum’s holdings and unwound trading affiliations as per their agreement. Ultimately, Baum proved prescient - interest rates and inflation tumbled, but Baum was already doing a solo shop at Princeton.
As noted later in the book, Simons prefered automated, but boiling down to understandable equations. Whether the reasoning behind the trades was intuitive was another story altogether.