New Constructs: Machine Learning to Save Time and Reduce Risk

Hi, this is David Trainer, CEO of New Constructs. When I started New Constructs, the pain point I was trying to solve was how to read annual reports and quarterly reports with real diligence. As a former auditor, executive compensation consultant, Wall Street analyst, hedge fund manager, and now running a research firm, I know exactly how the sausage is made. And I know that if you skip a page in an annual report, you might be blown up because you don't know what you don't know. And unfortunately, these reports are really long and complicated, but there is no substitute for doing the diligence of going through these filings. So back in 2003, I created what may have been the first ever machine learning environment. It was very simple. It was a combination of the filing, the database, and a simple parsing tool. And the idea is that the parsing tool would make things much faster and quicker. Humans were never typing anything in. Simple tools allowed the machine to go from a blob of text to a blob of words. In addition, we had a machine environment so it was able to track every single parsing action, not just for financial statements but for the footnotes. And over time, this meant that the machine could be taught to recognize patterns. For example, if a human treated a certain line item the same way each of the 17,000 times or more it was reviewed by one of our experts, Well, then we didn't need a human to analyze that line item anymore. Also, in a machine environment, you can create all kinds of checks and balances. We call them data checks. We've got thousands of them. Every time we made a mistake over the last dozen years, we created a data check to make sure that that mistake gets flagged or gets prevented or gets fixed automatically so that we're not doing the same thing over and over. The idea is for machines to do the grunt work, the monotonous work, the repetitive work, so humans can focus on the tougher tasks. Like finding new accounting loopholes. And we've run 120,000+ filings through this machine environment. So like the picture shows, an expert going through filings, and again, 120,000 filings. That means we've created the largest library of human expert-validated parsing instructions in the world. As you can imagine, going through that many filings has created a lot of learning for the machine to do. And once the machine just got a little bit smart, smart enough to parse Just a little bit, it could start really learning, not just following our instructions, but learning because it could parse filings and compare its performance to the filings that a human had parsed, grade itself, fix itself. And what's resulted is not just in a better data collection technology, but an entire research platform for serving our clients. We went to go get the best data in the world because we wanted to have the best models in the world. And we download directly from the SEC. We have a proprietary extraction system. Unique data classification protocols based on our expertise in accounting and finance. And that's the best database in the world, we believe, for the companies we cover. Modeling engine sits on top of that and it's agnostic. We can calculate anything you want. We can create reports. Main thing is that we own all of our own data so we can give our clients whatever they want. We're not in the business of selling religion. We're in the business of providing them an efficiency tool. And we can deliver that data via API, via our website. via PDF reports, anything you want. It's important to know that when we started New Constructs, it was to provide multi-billion-dollar institutional money managers with high-quality cash flow and valuation models. They relied on us to do the diligence they did not have time to do. We had to go through the filings and make sure there was nothing there to blow them up, give them a model that did all the things they would want to be able to do. And that's how New Constructs started. But we knew in order to do that work with real integrity, We had to create our own data collection system, and that's why we created the collection system, and that's why our models are so good. That's why we have the clients we do and have earned the recognition that we have earned across the industry. Note that we've got major financial firms, we've got a Big 4 accounting firm on the consulting side as a client, major wealth management firms. And if you follow us at all, you've seen us in Forbes, Barron's, Wall Street Journal, MarketWatch, etc. We're a thought leader on a lot of topics— finance, accounting, forensic accounting, as well as the new DOL fiduciary rule. And when you compare new constructs to the competition, what you see is what sets us— that really sets us apart is our due diligence and our transparency. We want you to know how much work we do. We link all of our filings into our models. So our institutional clients that see our models, they can look under the hood. They can see an adjustment we're making to the income statement, the balance sheet, or the discounted cash flow model and link that back to the original filing. And we want our clients to know how much work we do. And we also want you to challenge our competitors to provide as much transparency as we do. We don't believe they can. If they would, wouldn't they show you? Also, keep in mind that we're entirely independent. We don't make any money from the companies we cover or take any banking or trading fees. We provide coverage on stocks, ETFs, and mutual funds, and the economic earnings models that show all the details behind all of our ratings. Our system is easy to use. It's a website. Throw in the tickers and get an answer. Red is bad. Green is good. Note that we cover mutual funds, ETFs, stocks side by side because the models are the same. A mutual fund or an ETF is simply a collection of stocks. You can analyze cash flows and valuation on a group of companies just as you can an individual company. If you want more detail about the ratings, we'll show you. We'll show you. We break it down into quality of earnings and valuation and all the elements within that. And you can see here you can add tickers to the portfolios, see the model, download it all into Excel, or download a PDF report. And of course, we've got screening tools as well to take advantage of the huge amount of coverage we have on stocks, ETFs, and mutual funds. Really think about this as value investing at scale. And if you're not worried about high-quality data, let me leave you with this: we work directly with FASB, we've worked with regulators in the past, and let me tell you, They're not able to go through the footnotes for you. More importantly, as a member of the Investor Advisory Committee, I know that disclosures are getting more complex. Businesses are getting more dynamic. Filings are already 200+ pages on average. They're going to get longer. I helped write the new rule for bringing off-balance sheet debt back on balance sheet, and it's a nice improvement, but it also required additional disclosures, as do Many of the new accounting rules that are coming down the pike. And there are more new accounting rules or changes to rules coming down the pike in the next couple of years than have happened in the prior 15 years. So diligence is more important than ever. We recommend to our clients, to all investors, give your clients the best technology has to offer. Leverage the latest technology for research on stocks, ETFs, and mutual funds. Your clients deserve it. Whether you're an institutional money manager, whether you're an advisor, or whether you're an individual investor, everyone deserves the best in technology.