Finance is already the most computerized activity of all since the advent of high frequency trading, artificial intelligence and big data. Enter quantum computing in finance, and we are even closer to full robotic finance.
I remember when I started my career in the late nineties, how the stock market used to be a popular hobby. There was a grassroot equity culture in Switzerland, just like in the U.S. Taken by the internet mania, average savers turned into rookie traders dabbling in the equity market, buying stocks they were fond of and having lively discussions with neighbors about it.
A housewife would talk of her latest stock fad, and a taxi driver would share anecdotes about how a friend enticed him into the market. Everyone expected that the number of savers and other curious who turned into amateur traders would increase massively in the future, since this culture had roots in the Swiss savings culture, as a head of family would typically invest early on in shares of blue chip companies such as Ciba-Geigy or Credit Suisse.
To professionals and non professionals alike, the dotcom boom had a decisively human touch. At the time, everyone still remembered physical trading floors where voice brokers used pocket calculators. Today, these memories of a long-lost era make one sound like an old retiree talking drivel on her rocking chair.
What happened instead is that this popular trading culture was wiped out at the turn of the millennium by the dotcom crash, leaving small investors stripped and burnt. In 1996, the traditional stock exchanges of Geneva, Zurich and Basel, had given way to the fully electronic SIX exchange.
Electronic execution replaced floor traders. Soon, computers would call the shots in the markets. Technology and stricter regulation gradually pushed away small players, and high finance ushered in a sophisticated and professionalized era, while the popular culture never came back.
Twenty years later, even some traditional asset managers are being replaced by machines. Investment strategies are designed by artificial intelligence, sometimes without human intervention. Market activity is almost 100% computerized. Currently, less than 10% of stock market trading is made of fundamental single stock volumes, driven by active flesh and blood managers; the remaining 90% is made of broad index derivatives, option hedging, program trading, high frequency trading algorithms (which account for half of daily volumes in the U.S.), and largely automated passive indices (ETF).
Passive management now represents more than 40% of all equity investments (50% in the U.S.). A sign of the times: automated management has just overtaken human management in the U.S., 35% of equities are placed in funds run by computers versus 24% in funds run by humans. Quantitative funds now account for the biggest share of institutional trading (35%).
Some investment funds have turned into research and development laboratories, in order to earn excess returns, i.e. “alpha”, based on scientific methods. Their quant engineers process big data packages, including satellite images of various economic activities such as the movement of cars, shipping activity, or credit card transactions, and accelerate the data through deep learning (requiring high computational power).
This helps them create algorithms capable of predicting short term price movements. Admittedly, behavioral science is on the side of computer trading, having long legitimized its use. Because of the irrationalities humans suffer from, such as fear and greed, they tend to buy high and sell low. And they keep making the same mistakes time and again.
Therefore, the widely accepted doctrine is that you need to find ways to take emotion out of the markets, and that only machines are capable of this. The 30 year track record of quantitative trading and its more recent and prosaic version of robo advisors do not seem to warrant the claim of consistently higher returns, but confirm the superiority of mathematical trading in reducing volatility.
Nevertheless, the safety of this type of trading has never been adequately tested as of today. We know of the risks of high frequency platforms in provoking flash crashes, and we are aware that algo trading can cause huge amounts of capital to pursue similar strategies, and then dump the trade simultaneously, under the exact same signals, which is a factor of higher instability.
In currency trading, we have seen instances of mass carry trades invading one particular currency, flowing in a synchronized way, and exiting overnight, a phenomenon which tends to amplify risks. The effect on volatility is not clear. On the one hand, the increase of computerized trading has coincided with an era of overall lower and smoothed volatility, even leaving traditional day traders and market makers on the bench: more rational trading has resulted in the removal of inefficiencies in almost all markets.
A sum of trades pursued in one direction are bound to be countered by a reverse sum of trades in the opposite direction. Short selling is as impactful as long buying, and the VIX curve (S&P 500 volatility index) is testament to this overall decline in volatility.
On the other hand, we need to caution that markets have now been driven for a decade by central banks’ zero interest rate policies and measures of massive quantitative easing. This also has a tremendous effect on volatility reduction, as outcomes become way more predictable for players, who basically co-invest with the "whales" called Federal Reserve and European Central Bank.
In summary, computers have removed the notion of “market” from the markets, by reducing negotiation, human errors, inefficiencies and spreads. But so have done central banks, which are today’s second most significant market players next to computers and whose activity could mask the latter's.
The computerization of finance has even more to go, as we are now entering a new era: the era of quantum computing. The battle for quantum supremacy is engaged, with a Google processor recently performing in less than 4 minutes calculations that normal computers would take 10’000 years to complete. This capacity improves processing speed exponentially.
So it is no surprise that the first sector interested happens to be finance. Quantum skills are already very much sought for in banks, and JP Morgan is conducting experiments with IBM. According to Goldman Sachs, the financial services industry could be the first to benefit from this new revolution by bringing to the market a quantum algorithm quicker than any other sector. Reaching this superior speed of calculus could speed up computationally intensive option-pricing and risk-assessment calculations, putting the final touch to total robotic markets.
The ultimate question is how humans will keep up with the complexity. Just like in chess games, humans are now forever bested by machines in finance. Thirty years ago, the best investors were intuitive humans. Now the scientific approach to investing seems to have won. But at the end of the day, any financial algorithm is trading markets whose only underlying is human activity, with all its flaws. This is precisely why machine investing will never come close to achieving predictable or guaranteed returns.