Today, high frequency trading is exciting, but not sexy. Traders currently use computers with blistering connections to markets to take advantage of momentary market inefficiencies and, through tremendous volumes, may work up a sizable profit. It is exciting because these high frequency trading (HFT) firms have computers which are able to spot patterns on ticker tapes and make about 70 percent of the entire market's trades on any given day. But such practices are still not sexy, because the computers by themselves are as socially perceptive as a herd of cows. The computers have the technology to trade faster than a human being ever could, and consequentially have been programmed with algorithms which make them the masters of technical arbitrage. Where computers fall short, however, is in understanding the relationships which drive humanity, and more specifically, business.
At the University, we first think of a relationship as something like a friendship, platonic or otherwise, but we do not acknowledge the complexity of interactions. There are relationships between superiors and inferiors, between enemies, between groups and organizations and internally, within ourselves. However different these relationships may seem, they all boil down to the desire to give and get attention.
The phenomenal success of Facebook and Twitter is the outcome of this desire. We are a society which is addicted to attention through "likes," "retweets" and instant, always-ready information. Perhaps it is necessary to step back in order to realize that we are constantly exuding information about our political preferences, dietary opinions, skills and mood because of our love for attention.
Besides social networking, some of this digital information driven by attention can be useful in making decisions about future stock prices. But knowing that this information is available does not automatically make it usable data. The problem with HFT computers is that they must trade based on parsed data, such as binary commands. Since they cannot fully understand relationships among human beings, they have had no way to decipher our wealth of dynamic, always-ready information. I think this can be changed.
The saving grace for the computer is that a growing number of the interactions humans have are taking place on their playing field: the Internet. What I am currently researching is the ability to take in this user information, turn it into data and automatically make arbitrage decisions with high levels of correctness. Instantaneous human decision making ability, which reflects the sentiments of a large sample of the consumer population, combined with the computers' ability to execute trades could make markets much more efficient. Wall Street's financial prosperity could dramatically improve.
Both Twitter and Facebook have application programming interfaces (APIs) which can greatly facilitate that translation between human information - such as tweets, likes and hashtags - and data such as frequency and number count. For example, on Twitter, popular topics are often denoted by hashtags which can be aggregated automatically through use of the API. It is common sense, then, that you can tell how "popular" a topic is by the number of people talking about it on Twitter. This information, which is instantly available through human expression, may not be instantly reflected in the stock market because computers have been unable to bridge information about popular human interests to a company's stock price.
I believe computers have the ability to decide what information on social media and news aggregating sites is relevant to certain companies on the stock market. Consider IBM's Watson, which, according to IBM, "can rival the greatest human contestants" on Jeopardy! and can "comb 10-Ks, prospectuses, loan performances and earnings quality while also uncovering sentiment and news not in the usual metrics..." This type of natural language processing technology is precisely what can be applied to data from social media to make decisions about what is relevant to a company's stock price, in addition to whether the information is credible and what outcome the data may have. And, most importantly, such technology could have the ability to learn from its mistakes.
Such information processing, under the umbrella of artificial intelligence, has been experimented with before, including the example of an Artificial Neural Network (NN). An NN employs bio-mimicry, the engineering of man-made items to mimic nature, to replicate the function of neurons, which help make decisions in the human brain. Like a human brain, the NN is adaptive and can learn from patterns in data. I am excited to see how this technology could be used in a HFT application.
Human decision making ability combined with the instantaneous capability of computers certainly has the potential to eliminate inefficiency in the stock market and yield a profitable result.
Andrew Kouri's column appears biweekly Thursdays in The Cavalier Daily. He can be reached at a.kouri@cavalierdaily.com.
Read More
PARTING SHOT: No man is an island
By Ava MacBlane | YesterdayRetirement, as you can tell, has been treating me well. When people ask me now if I miss it, I laugh and tell them some things I definitely don't miss. But I don't always get to tell them about the things I really do miss.
PARTING SHOT: You never know unless you try
By Lexi Baker | 2 days agoMy time at the University was shaped by the people who showed me grace and kindness and by those who had confidence in me, even when I struggled to have it in myself.
PARTING SHOT: A love letter to Cav Daily and Connection
By Ava Proehl | 2 days agoMy time here taught me to look for the genuine, human connection in every moment.
