5 Ways Companies are Transforming Their Businesses with Machine Learning
At the intersection of machine learning and natural language processing sits Amenity Analytics, a relative newcomer to the text analytics industry. Amenity's founders see the information gushing from today's knowledge economy as a massive, untapped resource of important data for companies across a spectrum of industries.
The brainchild of a former portfolio manager and a machine learning academic, Amenity Analytics has automated the process of sifting through complex documents and narrative content to find and visualize meaningful data sets. For businesses that rely heavily upon the consumption and analysis of complex documents and texts, natural language processing (NLP) tools can save man-hours on a massive scale and deliver powerful insights to decision makers.
We spoke with Nate Storch, co-founder and CEO of Amenity Analytics, about how companies can stop drowning in information and start surfing it instead.
NLP is the branch of computer science that turns complex, unstructured data—specifically text written in human language—into structured datasets that can then be processed, visualized and analyzed.
I was an investment portfolio manager in my previous career, a job that required constant scanning of SEC filings, earnings call transcripts, broker research and news for nuggets of information relevant to our investments. Amenity Analytics was born out of finding a solution to information overload, which was one of my biggest problems in financial analysis.
I first learned about natural language processing when I was researching an investment and met my future business partner, Professor Ronen Feldman. Prof. Feldman is an early pioneer in NLP who coined the phrase "text mining" back in the '90s.
I saw immediately that NLP technologies could be a game-changer for me in investment analysis, not only bringing enormous time efficiencies to scanning complex financial documents but also creating data sets that could be visualized and analyzed to uncover trends and outliers. I saw potential applications of NLP to many other industries as well, and that's why Prof. Feldman and I founded Amenity Analytics.
We created a platform that gives people the ability to take a document and apply artificial intelligence to mimic the perspective of an expert. Forensic accountant, top financial analyst, an expert in ESG—all of those models are there, and countless others can be created. Amenity's NLP platform can quickly highlight, extract, and visualize data so users can spot patterns and outliers or establish causality—correlating stock prices with suspicious behavior, for example.
When you think about it, it's ridiculous that humans are still consuming text in much the same way we did 100 years ago. At Amenity, we are challenging our clients to reassess a process very core to their operations: reading complex financial documents. Here are five case studies that illustrate the innovative ways our clients are applying Amenity's NLP tools to solve problems, answer questions, and revolutionize the way they consume information.
1. Analyze quarterly earnings conference call transcripts to uncover deception or looming issues.
One of the spaces we're most focused on is my old industry—investment services. Amenity's platform is designed to address the needs of financial analysts and portfolio managers on a fundamental level, in a way that is going to be highly disruptive for the industry.
We are working with some of the world's largest and most sophisticated hedge funds. Our platform empowers their discretionary investors—the people doing true fundamental analysis—by providing additional insights and datasets into what's going on at their targeted companies.
For example, Amenity's software can process earnings conference call transcripts to flag companies that may be in trouble—before simmering issues impact earnings results. We've embedded the language analysis techniques used by CIA interrogators into our platform to capture and highlight clusters of language that might indicate uncertainty, doubt or even dishonesty, such as clichés and detour statements. A sudden spike in this type of language used by a CEO during a company's earnings call is often a reliable indicator that something is amiss.
Our platform can also process documents through a niche investment lens. For example, we built processing models from the perspective of an Environmental, Social and Governance (ESG) expert. If an analyst wants to quickly find pertinent ESG information related to a potential investment in a certain document, in the news, or in a 10-K or in an earnings call transcript, they can enter our platform and use an ESG model to highlight their documents from the ESG perspective.
2. Make better underwriting decisions, faster.
Starr Insurance Companies has customized Amenity's platform to assist in underwriting Directors and Officers (D&O) Insurance. Underwriters are often tasked with coming to a decision very quickly around whether they want to underwrite a company's D&O risk or not. And in order to make that decision, they have to process and understand information that's included in a stack of documents that can be a foot high.
Starr Insurance turned to us for help, first off, to create a model that analyzes those documents from the perspective of their top underwriters, those who consistently have the best feel for the data points that indicate whether a company is an acceptable risk or an unacceptable risk. We've been able to define and codify those data points, so the model can lead the underwriter directly to the relevant text in those documents that reveals that critical information.
So now, when one of their underwriters sits down to start analyzing a company, he has a checklist and nearly all the information that he is responsible for knowing is right there in front of him, already highlighted. Not only is he more efficient, but he's also more accountable because the information that he's responsible for knowing has now been defined and is very much measurable.
As the company continues to acquire and track relevant information, Starr's underwriters can add it to their predictive models to identify where and how those data points correlate to risk or to opportunity. As machine learning improves underwriting models, it drives smarter decision-making. For example, an underwriter could review the historical record of various types of litigation to identify how they correlate to claims and problematic situations. Underwriters can score this data so they can identify specific risks for the models to flag so it can extract related commentary.
Another insurance client recently tasked us with tracking diseases and epidemics. The company was trying to understand the path that an epidemic was taking by analyzing how people were talking about it—first in the news, second in social media and finally within the academic and scientific communities. We created a tracking model that brought together all three views to help the insurer separate true emerging epidemics from controlled diseases that were merely getting overblown in the news media or going viral in social media.
3. Track sentiment on political and economic issues.
Amenity was recently approached by a large investment bank to monitor news and conversations related to trade wars and flag changes in the tone of policy makers. For this project, we conducted a dynamic analysis of social media conversations and news statements regarding trade and tariff issues that were being made by major players in political and economic arenas, including Donald Trump, Steve Mnuchin, Robert Lighthizer and Wilbur Ross.
The trick in monitoring social media—as well as monitoring increasingly fragmented traditional media outlets and blogs—is cutting through noise to find data that is truly relevant to the specific topic. We identified and scored trade-related terms in key regions of the world, looking at equity, currencies, credit, and commodities. We were able to pull from unlimited sources and social media, cut the noise and false positives, and flag statements that were indicative of shifts in rhetoric or perspective.
4. Establish causality of market moves.
NLP is a useful tool for teasing out underlying factors of sudden movements in price or volume of trades in commodities, stocks, or currencies. Amenity was recently challenged by a client to sift through over 5 million news articles to discern what made the dollar move during a specific range of days. We were able to program our models and deliver an answer to our client in less than a week, via a dashboard of visualized data which enabled them to pinpoint the cause of each move.
We've created a similar dynamic analytic system for another client, centered around monitoring conversations related to cryptocurrencies. It processes discussions on cryptocoins to flag what is moving markets in each of them. We input Twitter's raw API, blogs and chatrooms that discuss cryptocurrency, and scraped various news sources such as Dow Jones and Reuters. We created a taxonomy of weighted topics to score meaningful data, so the system can pinpoint news and conversations that are directly connected to changes in coin prices as well as to buy and sell recommendations.
5. Prioritize documents for review.
In fact, Nasdaq partnered with Amenity to harness NLP technology for its regulatory compliance program. Today, Nasdaq's analysts spend about 60% of their workday reviewing the more than 48,000 SEC filings submitted by Nasdaq companies each year. While their compliance program effectively evaluates securities for compliance with quantitative requirements (e.g., equity), it has limited ability to facilitate the qualitative elements of an analyst's review (e.g., equity offerings, investigations). Nasdaq believes that our technology can improve efficiency and save time by helping them prioritize the filings that require manual review without compromising the integrity of the market.
One of the easiest ways for public companies to experiment with this technology is to use it to systematically analyze the earnings call transcripts of their competitors. Boards and executive leaders can use NLP technology to see at a glance what's going on in their industries with an incredible level of depth and granularity. They can monitor what all the companies in their space are saying about the competitive environment—not just in their specific industry, but up and down the supply chain as well.
Amenity has recently launched a new cloud-based software solution called Viewer, which enables any interested party to simply log on through Amenity's website and rapidly process earnings call transcripts and uncover real-time, actionable insights that can only be gleaned by analyzing massive numbers of complex documents at scale. Soon we will be adding SEC Filings, news, and research to Viewer.
Nathaniel "Nate" Storch is the Co-Founder and CEO of Amenity Analytics, a New York and Israel based provider of NLP analytical solutions and software. Nate is an experienced financial industry executive with nearly two decades as an investor and business builder. Before co-founding Amenity Analytics, he served as Managing Partner of Pilgrim Hill Capital, where he provided equity-related capital and strategic advice to small and mid-cap companies. Previously, Nathaniel served as Partner and Senior Portfolio Manager at Swieca Holdings/Talpion Fund Management, Partner and Head of Equities at One East Partners, and Head of Merger Arbitrage Research & Event-Driven Healthcare at Highbridge Capital Management.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.