Financial firms, day-by-day are turning to machines to do the job of humans. In August 2015, a wealth management firm Charles Schwab launched a service called Schwab Intelligent portfolios. Unique as this service is, it’s not a human being that decides where to invest but an algorithm does. An algorithm is a set of lines of code programmed into a computer which can take decisions and make changes in an existing system.
Tobin McDaniel leading the Schwab Intelligent Portfolios team says – “It lowers cost for investors”.
With this service you get portfolio management at essentially no fee at all, as opposed to working with a traditional advisor fee of 1%.
A computer is quite capable of beating the world chess champion and understands voice commands on a smartphone. Many activities on Wall Street are getting automated. After 167 years of operation, the Chicago Mercantile Exchange Group closed most of its trading pits. Usual practice of traders shouting and using hand signals for buying and selling commodities has become outdated. A much quieter competitor has taken their place – the computer.
Previously trading was limited to human response time, which is about half a second or about the blink of an eye. Experts say computers boost liquidity, helping to-be buyers and sellers find another without middleman. Some algorithms read market sentiment based on news reports, earning statements and regulatory filings and then decide upon how to view a stock.
Transformation from Algorithms to Machines Which Learn
In 2014 more than 40% of new hedge funds were systematic as they used computer models for the majority of their trades.
Field of algorithmic trading is designed to react instantaneously to market changes. These algorithms search and exploit small windows of trading opportunities, measured in milliseconds and nanoseconds. The Securities and Exchange-Commission (SEC) is now looking for ways to regulate them as it does the Wall Street owing to huge number of orders placed on the US stock market using automated algorithms. Ultimately these algorithms do what they are programmed to do by humans and work at superhuman speeds to identify tiny windows of trading opportunities.
Quantitative trading which is currently practiced relies on a human being to develop a mathematical model for identifying trading opportunities. This model gets updated by hand to adapt to new markets or by changing its conditions. In the case of an artificial intelligence while the initial software is developed by humans, the AI itself develops the model and changes it over time.
Some robots developed ingest vast amounts of information which include news and social media and use its reasoning powers to recognize connections and patterns in the data. These patterns are then used for making predictions about the market which are then translated into buy and sell orders, with no direct human involvement.
Silicon Valley Fueling Artificial Intelligence
Silicon Valley has been fueling the latest advances in applying artificial intelligence to financial trading. Companies like Google have invested heavily in machine learning.
Even human emotions have patterns which are predictable. It’s a combination of tens of thousands of different predictive patterns that are identified and this is where the AI get the advantage.
Backtesting to Real World
According to industry data provider Eurekahedge, hedge funds which use AI for their investment decisions have outperformed average industry returns every year for the past 7 years. (Except 2012)
While the average masks the wide range of returns while some AI hedge funds make profits and other spectacularly fail, it’s a risky industry.
There are some AI programs out there which have a strategy to prevent financial losses while having strong emphasis on downside protection. And there are those which give really volatile returns that the average hedge fund investor would shy away.
This volatility has become very clear in each of the last 3 years, here the overall performance of global equity markets has easily outshone the average hedge fund, with or without AI.
However, as noted by a senior Wall Street banker, history has never been a good predictor of the future. This shows that historical tests do not always translate into real world success.
Latest Artificial Intelligence’s victory
Lee Se-dol Defeated by Google’s DeepMind
AI developed by Google’s DeepMind unit called AlphaGo was able to defeat Lee who resigned after hanging on the in the final period of second-reading overtime, giving him fewer than 60 seconds to carry out each move.
This was the first time the ancient Chinese game of Go had been played to a world-class level by an AI. Go’s combination of simple rules and detailed strategy has made it a major challenge for AI research.
DeepMind is a good example how an algorithm can become adaptive in nature (as mentioned above). It continually reinforces and improves the system’s ability by making it play millions of games against tweaked versions of itself. In the process it trains a policy network to help AlphaGo predict the next moves, which in turn trains a value network to ascertain and evaluate positions. AlphaGo predicts possible moves while going through permutations before selecting the one that it deems most likely to succeed.
It’s a Machine’s World
While we live in a world where we cannot prevent every problem, it will take time for regulators to spot problems and become aware of it and adopt rules for it. For preventing bigger swings, financial regulators in the US have implemented measures like single stock circuit breaker, making a pause in trading for individual stocks a requirement if the price moves 10% or more in 5 minutes period.
The success of AI funds may in fact be underestimated due to the secrecy that surrounds the business. Although a trend suggests that a robot will be running asset management sometime in the future, people still play a role in the financial system but its changing. Be it Wall Street or any other street (read self-driven cars) the machines have a huge role to play in the future and AI is the key to their success.
To learn more about implementation of Artificial Intelligence in Trading, join us for our upcoming webinar on Machine Learning.