The Glooko app helps simplify diabetes management by making data more easily available to patients and their care providers.
Stock predictions made by machine learning are being deployed by a select group of hedge funds that are betting that the technology used to make facial recognition systems can also beat human investors in the market.
Computers have been used in the stock market for decades to outrun human traders because of their ability to make thousands of trades a second. More recently, algorithmic trading has programmed computers to buy or sell stocks the instant certain criteria is met, such as when a stock suddenly becomes cheaper in one market than in another — a trade known as arbitrage.
Software That Learns to Improve Itself
Machine learning, an offshoot of studies into artificial intelligence, takes the stock trading process a giant step forward. Pouring over millions of data points from newspapers to TV shows, these AI programs actually learn and improve their stock predictions without human interaction.
According to Live Science, one recent academic study said it was now possible for computers to accurately predict whether stock prices will rise or fall based solely on whether there’s an increase in Google searches for financial terms such as “debt.” The idea is that investors get nervous before selling stocks and increase their Google searches of financial topics as a result.
These complex software packages, which were developed to help translate foreign languages and recognize faces in photographs, now are capable of searching for weather reports, car traffic in cities and tweets about pop music to help decide whether to buy or sell certain stocks.
Finding Work-Life Balance in the Financial World
View this infographic to learn how to use your smartphone to work smarter — not harder. Download Now
Mimicking Evolution and the Brain’s Neural Networks
A number of hedge funds have been set up that use only technology to make their trades. They include Sentient Technologies, a Silicon Valley-based fund headed by AI scientisk Babak Hodjat; Aidiya, a Hong Kong-based hedge fund headed by machine learning pioneer Ben Goertzel; and a fund still in “stealth mode” headed by Shaunak Khire, whose Emma computer system demonstrated that it could write financial news almost as well as seasoned journalists.
Although these funds closely guard their proprietary methods of trading, they involve two well-established facets of artificial intelligence: genetic programs and deep learning. Genetic software tries to mimic human evolution, but on a vastly faster scale, simulating millions of strategies using historic stock price data to test the theory, constantly refining the winner in a Darwinian competition for the best. While human evolution took two million years, these software giants accomplish the same evolutionary “mutations” in a matter of seconds.
Deep learning, on the other hand, is based on recent research into how the human brain works, employing many layers of neural networks to make connections with each other. A recent research study from the University of Freiburg, for example, found that deep learning could predict stock prices after a company issues a press release on financial information with about 5 percent more accuracy than the market.
Hurdles the Prediction Software Faces
None of the hedge funds using the new technology have released their results to the public, so it’s impossible to know whether these strategies work yet. One problem they face is that stock trading is not what economists call frictionless: There is a cost every time a stock is traded, and stocks don’t have one fixed price to buyers and sellers, but rather a spread between bid and offer, which can make multiple buy-and-sell orders expensive. Additionally, once it’s known that a particular program is successful, others would rush to duplicate it, rendering such trades unprofitable.
Another potential problem is the possible effects of so-called “black swan” events, or rare financial events that are completely unforeseen, such as the 2008 financial crisis. In the past, these types of events have derailed some leading hedge funds that relied heavily on algorithmic trading. Traders recall that the immensely profitable Long-Term Capital Management, which had two Nobel Prize-winning economists on its board, lost $4 billion in a matter of weeks in 1998 when Russia unexpectedly defaulted on its debt.
Some of the hedge funds say they have a human trader overseeing the computers who has the ability to halt trading if the programs go haywire, but others don’t.
The technology is still being refined and slowly integrated into the investing process at a number of firms. While the software can think for itself, humans still need to set the proper parameters to guide the machines toward a profitable outcome.
Technology and industry trends are shaping the next era of finance. Check out our complete line of finance industry solutions to stay ahead of the competition.