Unlike other portmanteaus in the English language (think “smog” or “brunch”), the definition for “fintech” — financial technology — is evolving, due largely to the machine learning that is increasingly driving it. Refinements in the structure and application of AI algorithms and robo-advising platforms now allow firms that specialize in fintech to deliver market access to their clientele on a scale previously unavailable to all but a select few. TrueRisk Labs is one such a la carte AI firm harnessing its proprietary artificial neural networks to distribute Artificial-Intelligence-as-a-Service (AIaaS) across the financial services industry. In this way, TRL is applying its predictive analytics engine to generate trading signals in a hyper-connected world, allowing more companies, traders, and investors to have a say in redefining fintech’s future.
That approach stands to be a very impactful, given the rate of change in the field of machine learning. The stats surrounding computing illuminate the sheer scale of change that has occurred in data generation, processing speed, the physical and digital footprint of that data, and, of course, its cost. In 1980, the average price of one gigabyte was $437,500. Now that cost is less than two cents.
Since the 1990s, history has seen the world’s computers go from generating around 100 GB per day in data to roughly 50,000 GB per second now. If we then factor in the incredible growth and scale of machine intelligence, the processing power available becomes pretty impressive: 20 years after IBM’s Deep Blue beat Garry Kasparov at chess, DeepMind’s AlphaGo defeated the Go world champions Lee Sedol and Ke Jie, respectively, in 2016 and 2017. DeepMind then built another version, AlphaGo Zero, and trained it to learn by playing millions of games against itself, with no human statistical input. Within 72 hours it had learned enough to beat the earlier version of itself 100 games to zero. It took an ancient game of strategy played by tens of millions of people over centuries and became its reigning champion, synthesizing thousands of years of human knowledge in days, developing insights and unconventional strategies along the way.
Harnessing this type of machine learning has become essential in many disciplines, and the AlphaGo reference is particularly relevant with respect to fintech: processing data to achieve optimum results requires smart humans to build very smart machines, ones that are autonomous, self-learning and not bound by conventional (human) wisdom.
TRL’s CEO, David Wilson, explained the company approach to expanding AI insight like this: “Our artificial neural networks replace the way current fundamental, technical, discretionary, and quantitative managers run a portfolio. By taking advantage of the confluence of big data, AI, and data science, we’re able to reach non-linear investment conclusions and trading opportunities investors would not find on their own.”
As the size of actionable data with respect to finance grows increasingly massive, it is also becoming more varied. The analytics derived from fundamental and quantitative data analysis now share disk space with those from alternative data sets — think crowd-sourced data, from social media habits to credit card spending, or satellite imagery predicting weather patterns or geopolitical unrest. When data-driven metrics are applied correctly to fuse traditional intel with all available analytics, alpha-generation becomes more likely. And as the IoT continues being shadowed by the IoE (Internet of Everything), it will become increasingly archaic to even refer to some data as “alternative.” Data generation will continue exponentially — the market for predictive analytics alone is expected to grow from $4.56 billion last year to $12.41 billion by 2022, with the largest share composed of the Banking, Financial Services, and Insurance vertical. It is the machines extracting valuable data from all the transient and superfluous zettabytes of content that will truly give investors an edge.
There have been a few early adapters implementing AI in the recent past: Goldman Sachs, BlackRock, Bridgewater Associates and several other well-heeled names have allocated significant resources to machine learning. What has changed is the availability of deploying that same level of analysis to smaller firms and individual investors.
David continues with TRL’s mission and mode of democratizing data generated by AI: “By constantly processing the historical performance and volatility of stocks and indices and the historical correlation between different stocks, and assessing the behavior of investors and analysts, the AI learns from the numerous data sets we monitor. We tailor our custom machine learning frameworks in a way that best suits the specific needs of each client. This is true, native AI — constantly learning, constantly evolving, and constantly executing decisions in the most strategic manner possible.”
Veteran Silicon Valley entrepreneur Steve Blank once summarized the development of AI with the following analogy: “machine intelligence is not like inventing a better sword, it’s like inventing gunpowder.” This has proven exceedingly accurate in the financial services industry. The need to use data science efficiently in asset management is a given; those who deploy it most effectively will prevail in the emerging world of modern fintech.
TrueRisk Labs is a content partner of Benzinga. Trading signals from the TrueRisk AI system cover over 4000 stocks and cryptocurrencies, and are available through Benzinga.
© 2018 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.