Join NexChange - the professional
network for the financial services
industry - and receive a free one-
year subscription to Forbes
Steven Cohen, Eying New Investors, Doubles Down on Quants — but Will It Work?
Steve Cohen, who earned a fortune during a period of time when portfolio managers in his hedge fund were going to jail for insider trading, did not make the list of top hedge earning fund executives in 2016, earning a paltry 1% on the assets he manages in his family office, Point72 Asset Management.
When the firm was managing outside money under the moniker of SAC Capital, and before he pled guilty and paid $1.8 billion in fines, times were better. Cohen, looking to find a new spark to return to the glory days of the past – and hopefully avoid the very human frailties that drive relationship-based insider trading. To accomplish this, and reduce costs, Cohen is emphasizing the development of trading robots. But can reproducing a trader’s minds through only mathematical logic work?
Steve Cohen doubles down
It was with much fanfare that Cohen rolled out his Quantitative Investment Division nearly a year ago, later poaching Bridgewater Associates head of research after that firm announced they were following a similar path.
Since then, Bridgewater has let go its tech guru, Jon Rubinstein, who it had brought in from Apple Computer to replicate the minds of Bridgewater’s great traders.
Cohen, however, appears to be doubling down on the idea of using automation to eliminate those pesky and expensive human traders. Bloomberg’s Saijel Kishan, citing unnamed sources, reported that Cohen is ramping up the effort.
Cohen has bluntly stated that the investment and trading industry is short of talent, so his move to replicate the investors comes as his firm, under a new banner, Stamford Harbor Capital, is preparing to manage client money in January 2018 when the Securities and Exchange Commission ban, leveled against his firm in 2008, expires.
Steve Cohen – he difficult point of automation for a discretionary hedge fund is found where if / then logic can’t compute
Perhaps most notable in the Bloomberg article are the factors that Steve Cohen is looking to automate – and what was not mentioned.
The Bloomberg report stated that Cohen’s quant group is seeking to “amplify the firm’s profits by using algorithms that replicated the trades of its best-performing managers.”
This can be a tricky proposition, particularly in a hedge fund where discretionary, human thoughts have been involved in evaluating trades, gathering information, sometimes through relationships, and executing those trades.
The report said that the quant team is seeking to identify position sizing traits, risk, hedging and leverage practices to identify patterns and relationships in the human’s trading style.
Perhaps most interesting – and what was left unsaid – is that those items mentioned in the article that the quants are attempting to automate are, in many cases, the repeatable, quantifiable factors. Position sizing, for instance, is often correlated to the trader’s definition of success probability. Risk management, as well, can be a linear, if-then decision-based process. If the trader sees risk ahead, they turn up the hedging dial and vice versa.
What was not mentioned in the article was the most difficult points of discretionary analysis – connecting often noncorrelated dots to develop an investment thesis.
Consider the US Federal Reserve’s decision to hike interest rates Wednesday. There has been significant discussion of replacing human-based Fed decision making with a more systematic process. It can be argued that a systematic process would not have changed market expectations for a rate hike as quickly as was recently done.
Just three short weeks ago the notion that March was much less clear, for instance. Into Fed Chair Yellen’s calculation enters, in part, the concern that US President Donald Trump and his plans for fiscal stimulus might heat up an economy that is starting to gain momentum. In large part, such a decision, based on the discretionary thoughts of Yellen and held by a majority of the Federal Open Market Committee. The fact that the independent thinking Minneapolis Fed President Neel Kashkari voted against the hike further illustrates how a discretionary analysis process produces different results.
Can Steve Cohen fire his human traders and replace them with robots? That is a question that can only be answered through testing through a wide variety of market environments – including market crash scenarios. The talent behind a human investment manager that cannot be exactly replicated by a computer is understand nuance and assigning meaning to what appear as mathematically noncorrelated events.
An automated trading system delivers consistent outputs given similar known market environment variables. Humans often don’t. This is where one real challenge in Cohen’s efforts might be found.
This article was originally published in ValueWalk.
Photo: Ivan T