Why Bad Data Ruins AI Models: David Trainer of New Constructs

Nicki Benvenuti:0:00
So we are here today with David Trainer, the uh founder and CEO of New Constructs. Uh so David, would you like to um tell us um for those who are hearing about New Constructs for the first time um how you describe what your firm does and uh what problems you solve?

David Trainer:0:14
Yeah, sure. Uh you know, we basically have created a a better mousetrap for measuring earnings and rating stocks. Uh I started this business over 20 years ago after uh a fairly long career on Wall Street. Uh I was there before, during and after the tech bubble and at Credit Swiss, which was the number one tech IPO investment banking firm. I had a front row seat to just how the sausage was made or how easy and how often willing Wall Street analysts were to misrepresent the truth about profitability valuation because they got paid for selling stock. So, it was in their best interest to say it was a good company and a good stock. And I felt, look, if these folks aren't going to be doing the work required to really understand companies because they're more interested in putting on a good face and selling the stock, then who is going to do that work? And if that work's not in place, then how are markets going to ever know the truth? And if markets don't know the truth, they're going to be manipulated and they lose their integrity. If our stock market loses its integrity, then our country suffers because I believe that honestly that the capital markets are perhaps the single biggest driver of the prosperity and growth that the United States has seen over the last 250 years.

Nicki Benvenuti:1:38
Oh, absolutely. Absolutely. It's one of the backbones of of wealth and um absolutely. Um so so tell me um you said that your machine learning u models extract data from financial um filings to build a better mousetrap for u measuring earnings. Um so walk us through what makes your earning numbers um different from what everyone else reports.

David Trainer:1:59
Well, our our numbers are going to take into account data from footnotes and disclosures that we don't think anybody else sees. And the reason that we think no one else sees them is because there are indices that Bloomberg runs based on our data and strategies based on that data that dramatically outperform the s the S&P 500. So there's there appears to be uh at least according to the market and that typically is the final arbiter or final boss on these on these kinds of things. there's alpha in in the in the data in that it can be used to consistently generate alpha and we've seen this outperformance versus the S&P 500 in the short term and the long term.

Nicki Benvenuti:2:38
Wow. So, um from what we've read and you know and um you know the um first trust new constructs core earnings leaders ETF um was up 27% last year um while the S&P managed 18%.

David Trainer:2:54
Yes, exactly. Uh yes. Um Nicki, you're referring to the ETF that tracks the Bloomberg index, the core earnings leaders index. And yes, uh that that ETF did exactly perform that well last year. 27 versus 18 is pretty dramatic outperformance. And what's what's even more interesting about this is that we're not talking about some exotic strategy. We're talking about just picking stocks that have stronger earnings quality than the market understands. And by that I mean those companies have core earnings that we measure that's higher than their reported earnings because their reported earnings are embedding a bunch of unusual gains I'm sorry unusual losses that artificially depress their earnings when the real numbers when you remove those those noise that noise are actually better than what the market thinks. And when you invest in these stocks that have better earnings it outperforms the market. And it's a very simple strategy, I think, especially in a world where people are reaching and doing all these kind of exotic strategies or they're or they're they're piling into popular names because they've been moving. And those strategies are so much more risky because they're dependent on momentum or they're dependent on some sort of broad range of events always happening. And I try to encourage people to avoid those kinds of risky strategies, especially now that there's options like what you just mentioned to invest in a strategy that's just based on a better measure of earnings. I mean, isn't that what everybody wants?

Nicki Benvenuti:4:19
[laughter] Yeah. At the end of the day, everyone wants more earnings. Absolutely. Absolutely. Um, so, um, I guess that leads into the next question, which is like everyone right now is questioning whether AI is worth it. You argue the answer depends entirely on the data feeding the model. So why does the data matter more than the model itself?

David Trainer:4:43
Uh you know this is the same the answer to this question is the same as it's always been. You know uh meet the meet the new boss same as the old boss. You know ever since there were models. There's been one immutable law about every single model. Garbage in garbage out. They're only as good as the inputs. So, you know, it doesn't matter how smart, you know, a human brain is or how big an engine is. If you feed human brain a brain bad information from inception, you know, it's not going to do well. If you put bad fuel into a big engine, it's not going to do well. Uh, this is just an elementary concept in the way the world works and it applies to models. So, these AI models that we hear about all the time now, uh, you know, they're not that good. Why? Because they're based on the internet. Internet's got bad data. And look, you don't have to take my word for it. We did this at dinner last night with a couple nights ago when my parents were over with the kids. We asked AI, "Why do you lie so much?" And it will tell you because it has bad data. It'll tell you. It's got bad data. And when you're when you're talking about consequential decisions, you don't want to make them based on AI. Nothing consequential should be based on AI because it's there there are no experts who certified any piece of AI. Until you got that a large gr a large number of experts certifying that it's correct. I just wouldn't touch it. I wouldn't touch it for anything consequential.

Nicki Benvenuti:6:10
H okay. So so the data so the data is very very important.

David Trainer:6:14
Data is is is the most important thing right? I mean the inputs are are what going to be responsible for the outputs. So if you got bad inputs, you're going to get bad outputs. There there's no magical thing about AI that's going to be able to figure out something when it doesn't know it or doesn't have it. It can't do that. It's like alchemy for information. AI can't do that. No computer can do that as far as we know. Nothing except maybe God could do that. Right? So it's impossible. So again, it all it's all about input. I mean the way everything else in the world works. I mean, it's, you know, you get in, you get out of it what you put into it. If you're not putting good stuff into it, you're not going to get good output.

Nicki Benvenuti:6:55
Absolutely. And, um, to that to that credit, I'm going to, you know, do like a little bit of a humble brag for you. So, apparently Google reportedly invested 2 million to build an agent on top of your data. So, um, what does a company see, um, like a company like Google see in new constructs? Um, that tells you you're on to something bigger.

David Trainer:7:12
Yeah, let me answer that that question with a story. Uh I I was I was good friends with a guy who who used to run professional services for Google Cloud. His name is Brad Little. We went to college together and we were actually hanging out because our kids were at a Penn State wrestling camp together and we were just talking and I was, you know, I knew he's at Google and I was like, "So Brad, you know, I know you guys are spending like10 billion dollars a month on data centers. Let me ask you a question. What kind of data are you putting in the data center?" He's like, "Oh, Dave, well, nobody's really supposed to ask us that." Well, I know. Well, I'm asking you, Brad. And he said, "Yeah, it's a good question. We, you know, you know, just like everyone else, we're putting in internet data." I said, "Well, what if you had a data set that was proven accurate for any particular genre or any particular field of study or part of the world?" He said, "Well, that would be awesome. We'd love to have a perfect data set, but they don't really exist." I said, "Well, what if I knew of one?" He said, "Tell me more about that." I said, "Well, Harvard Business School and MIT Sloan published a 70 page paper proving that we had an accurate and superior data set for core earnings, those core earnings we talked about." And he said, "That's amazing. That's exactly the special sauce we're looking for to show that our technology works well when we can marry." And this is a quote in the press release when we when we uh when we put the AI agent out there. It's called Finsites, but the press release says technology can do amazing things when it's married with good data. And so that's that's the story.

Nicki Benvenuti:8:42
Yeah.

David Trainer:8:42
AI or Google decided to invest over $2 million uh to build it and and and even more to pay us for our subject matter expertise to help them build out this AI agent based on good data. And again, it was because they wanted to demonstrate the power of the technology when it was endowed with good data.

Nicki Benvenuti:9:08
That's well that that answers exactly why Google was interested. That's fantastic. So that really validates the approach. Um and and furthermore, if Harvard Business School, Ernston Young, and Sloan all have all published work supporting your approach. So what did that outside validation confirm for you? And what did it change about how you tell your story as a um you know?

David Trainer:9:21
Yeah. Yeah. I mean, it didn't change how much, you know, how much I It didn't change how I tell my story that much because I've been telling people for a long time that our data was better. Just great to have proof. Great to have proof because I would go in and I meet with certain money managers and some money managers would say, "This is amazing. We got to have it." And someone would say, "How do you know your data is better than so- and so?" I'm like, "Well, let's do a comparison." And we pull it up and they, you know, they didn't really want to change. There's a lot of inertia on Wall Street. You just heard Sam Alman talk yesterday about there's a lot of inertia in the world to adopt AI, right? People people don't want to change and as a friend of mine once said, people hate change, especially when it's for the better. And so steal that one.

Nicki Benvenuti:10:04
Yeah.

David Trainer:10:04
There there's a lot of resistance uh to change often. Uh but when there's you from very prestigious institutions published papers saying you need to change or you're using bad data, you could be violating fiduciary duties if you're knowingly using bad data and and and making investment recommendations based on knowingly bad data uh verifiably bad data. That's going to motivate people to do things better. That's that that obviously is helpful. But it's also good to have independent third parties because of course I'm going to say my data is better. People ask me this all the time like how do you I'm like don't ask me ask Harvard Business School. Don't ask me. Ask Ernston Young. You don't want my answer. Even if I did so show you a study that I put together about how our data was better. Would you trust it? And they're always like good point. Okay. We'll take the paper.

Nicki Benvenuti:10:55
Touche. Touche. I always think that with digital marketers whenever they're like oh look this validates our approach. just like but you know that might just be you know um that you might have some bias going on there. Um so your path here is anything but typical. So um and I don't know how much of this you know you want to get into on the podcast so we can always edit. Um you got fired then a head hunter told you to go look up Wall Street at the library. So um if you're comfortable take us back to that moment and what it set in motion for you.

David Trainer:11:19
Yeah. So uh so my first job in finance was was with Arthur Anderson and they hired me to be an executive compensation consultant and it was our job to go to boards of directors and explain to them whatever you do don't pay executives based on accounting earnings because they can grow accounting earnings while running the business into the ground. We've seen this many times over Enron Worldcom you know it's happened before. Uh, and so that was where I really learned about how to find all the details in the footnotes, began to learn about that. Uh, and but that practice suffered. There were some HR issues and then they put me on the tax team. They put me on the audit team and I was really bad at tax and audit. It was so boring. I didn't I didn't I wasn't so much bad about it so much that I was bad about it.

Nicki Benvenuti:12:01
It's just that doesn't sound terribly exciting.

David Trainer:12:06
Yeah. And and I I mean, honestly, like it was cripplingly boring. Like I had no didn't know that was a thing. and and uh and so yeah, they ended up firing me and and I was devastated. I was like, "Oh my gosh, I was like 23, 24 years old, like, "Oh my gosh, what am I going to do?" And I did the only thing I could do is I called every single head hunter that they gave me gave me on the list. And uh one of them was nice enough uh I think his name was Doug Lindecker. This is like 30 years ago almost. And and he just kind of took me under his wing was like, "So what do you want to do?" And and you know, talked to him for a little bit and he said, "You should go to Wall Street." I was like, "What is Wall Street?" Uh, he goes, "You should go. It's it's a place they got investment banks and things like this." Go, "What is it? What is an investment bank?" And he and I, when I said I didn't know, he goes, "Go to the library and look it up." And so I did. And then I realized that there was a a a food analyst named Michael Mobeson who was focusing on the same concepts that I'd focused on when I was doing executive compensation consulting, which is economic earnings and return on invested capital, which are much more comprehensive measures of profitability. And uh and before I worked at Arthur Anderson, I' I've been in the management training program at Kroger. And so I'm like, food analyst, executive compensation consulting. I'm going to go get a job with Michael Mobes. Turns out Credit Swiss was hiring for at a position exactly like what I wanted to do, which was build out these economic earnings models. And Mobes was my mentor, an amazing mentor. And and and that's kind of where it all started. I came on and I figured out how to build a a template for analyzing economic earnings for all companies and all countries around the world. and we populated that template to up to I think almost 1,500 to 2,000 companies with me and my team all [snorts] manual back then and and just started doing this research and that's where I learned around the world how all the tricks were done and how where all the bodies were buried in terms of footnotes and and it was it was that that that hands-on [snorts] experience that really helped me figure out how to to build a technological solution which is what we did in New Constructs to parsing filing Wow.

Nicki Benvenuti:14:09
It sounds like it just kind of all fit together and you know in like the best entrepreneurial stories, it just seems like one thing naturally rolled into another, but it really created something of value for the marketplace.

David Trainer:14:26
In retrospect, it looks that way, but at the time there was a lot of nailbiting and handering about what the heck am I going to do and how do I do this?

Nicki Benvenuti:14:26
Uh well, the podcast we take it up 30,000. [laughter] We just pretend it was smooth.

David Trainer:14:35
Yeah. You know, that's a better story.

Nicki Benvenuti:14:39
It's a much better story.

David Trainer:14:39
But yeah, along the way it was it was scary, you know, and and it was hard and I didn't know exactly when I got to Credit Swiss how to do uh all these things. I had to learn a lot. That that's part of why Michael Mobus was such a great mentor because he really just gave me the books I needed to read and and and they had the answers and um and from there it was just just a matter of uh hard work.

Nicki Benvenuti:15:01
I love our read a lot of our listeners or readers. Uh any books you care to share that he shared with you?

David Trainer:15:01
Yeah, the you know the the cornerstone books for finance and and understanding the economics of businesses are number one creating shareholder value by Al Rapaort. Uh the quest number two the quest for value by uh how can I by Bennett Stewart I can't believe I forgot that. And then there's the McKenzie valuation book which a lot of people know about. Uh, and and those are books that I I really used all together, all three kind of they all do certain things better than others. But then from there, I, you know, I just had to go build a model. And there's really no substitute to just kind of going out there, rolling up your sleeves, and like building the model that puts it all together. uh and and and figuring out how to do that and then applying that in the real world thousands and thousands and thousands and thousands of times over and over again with a team also doing that and and and and and then cooperating with all the analysts at Credit Swiss around the world with how to do that before the tech bubble. This all happened before the tech bubble because once the tech bubble hit then nobody cared. But before the tech bubble, we were doing all that and figuring that out and kind of dealing with all the errors and issues. And that was um was awesome.

Nicki Benvenuti:16:14
And I guess that leads right into my next question. During the tech bubble, you saw the detailed analysis on Wall Street um simply stop because it didn't help um sell IPOs. So what did watching how the sausage was made um teach you about who was really doing the work?

David Trainer:16:28
That nobody was going to do the work. I think, you know, and that's when one when I started New Contracts as a business, that was like the biggest piece of push back was, hey, you know, what you're describing new contract is doing, isn't that what Wall Street does? And I was like, oh my gosh. I mean, I had no idea how I mean, and it was kind of kind of silly for me to be that naive because before I'd experienced what I experienced when I knew, I was that naive. I was naive about how how Wall Street really worked. And so, uh, for me to expect everyone else to be, you know, understand that was maybe a little, you know, a little little bit too high expectations. But there were, of course, there were um, major lawsuits and things like that that showed that Wall Street was cheating and lying that I thought everybody was aware of. There's the Spitzer settlement, which was the largest ever settlement uh, one against Wall Street was over, you know, I think it was like $1.2 billion. That was the biggest one ever. It's like back in 1999 or something when they they they got Wall Street effectively to admit that they done it wrong, that there were conflicts of interest. Uh but still people just weren't aware of it and and so yeah, it was it had to do some explaining. I think now more people tend to be tend to know about it because or talking about it maybe, but there's still a lot of people that think Wall Street's doing it the right way and Wall Street isn't. And there's a lot of very obvious evidence for that.

Nicki Benvenuti:17:58
Um, so you've compared um today's market hype to the tech bubble pointing to the money at stake in deals like the SpaceX IPO when I've talked to you previously. Um, so what patterns right now are you seeing repeat and why should investors pay attention?

David Trainer:18:10
Uh, I'm the pattern the big pattern I'm seeing right now is like everyone is doing everything they can to remain bullish on AI uh because they want these big IPOs to get done because these big IPOs bring huge paydays. SpaceX on one day, investment banking fees only on the day of the IPO, $100 million on that day only.

Nicki Benvenuti:18:27
Wow.

David Trainer:18:31
Not to mention how much money they make trading the stock. Consequently, the green shoe, which is just another payout, that was $100 million. And they're going to get probably at least that much if the anthropic and open AI IPOs go as planned. So, is it not naive to think any group of people would do whatever they could to achieve those paydays again and again? And I think we also see this in the activities of Nvidia, right? All the all the financing that Nvidia is giving all these other companies in the AI ecosystem so they can keep spending on chips. We see the big AI companies talking about hundreds of billions of dollars in AI investing and spending and purchase commitments so they can keep the party going. As long as everyone thinks everyone else is still spending, then all this keeps on happening and people make a lot of money in the interim whether or not it's real. And the problem is that these hundred million dollar paydays for the investment bankers, the paydays for the SpaceX shareholders, the SpaceX billionaires, now that money's in the bank for them, the people left holding the stock, uh, they can lose that. So, we fought we're we're calling this the final cash out. And I've got training on my website around this final cash out going on right now. how you can make money and how you can make sure you don't lose money because this I believe to be and maybe it's not the final but this is the biggest things that's happened since the tech bubble and I think it's going to be so big that it risks undermining the integrity of the market so much that investors just say forget it and these marginal retail investors that have been driving the market up for the last 20 years they they say forget it because you know what it's if I'm going to be spec speculating. It's more fun to speculate on sports or other things like I can do on poly markets or khi khi markets right if I'm going to gamble why don't I gamble in something more fun than a stock where I can lose a lot so you see that increasingly more and more now yeah I think it could be a big a big liquidity exit for people uh if they see they figure out again they're just getting screwed by Wall Street h

Nicki Benvenuti:20:51
um wow um Well, the your clients um like you know to that end like you know your clients range from some of the world's largest hedge funds to individuals managing their own retirement accounts. So um you know how are you delivering value and um you know to allude to what you just said you know preventing people um you know from damages and you know in the marketplace um you know um across such a wide um spectrum um you know from the same underlying data.

David Trainer:21:10
Yeah, we we provide our ratings uh for as little as 49 bucks a month, sometimes even less than that for certain promotions. Uh and and the way we, you know, because we own the data from the source to the rating or to the end product and we own the whole process. It's very easy for us to on one end provide very very sophisticated and granular products to our institutional clients and then all that same research that goes into our ratings stays the same. But we then we only give people the rating. We don't give them all the data they can use to reverse engineer what it is we do because there's no reason to give people that kind of stuff. Most people don't want to see thousands of balance sheets and footnotes behind what it is that we we're doing. They just want to know green is good, red is bad. And and we do that for 10,000 stocks, ETFs, and mutual funds updated every day. And by the way, Nicki, I mean, Harvard Business School also wrote a paper wrote another paper showing that our stock ratings outperform Wall Street ratings.

Nicki Benvenuti:22:11
Wow.

David Trainer:22:18
So these ratings are valuable, too. Uh there's also a Bloomberg uh live traded index based on our very attractive stocks. uh all the stocks that get that rating are in this index and that index dramatically outperforms the S&P year to date over a year last year and over five years. So, you know, this stuff works and and we make it available in smaller packages at lower prices for everybody. And that's my way of proving to the world that I'm putting my money where my mouth is when it comes to leveling the playing field and and and supporting the integrity of the capital markets. I'm not just giving this great data away to the big Wall Street insiders. I'm giving it away to everybody so they can all operate on a level playing field because that's what's required to protect the integrity of the capital markets.

Nicki Benvenuti:23:02
And for the listeners who um are interested, it's new constructs.com and then the first trust new constructs core earnings leaders ETF or FTCE.

David Trainer:23:12
Correct. Correct.

Nicki Benvenuti:23:12
Okay. Got it. Got it. Um last question. You're planning to expand globally um which is really exciting after proving the model in the toughest market in the world. Um, what does the next chapter for new constructs look like and what do you most want investors to understand before they get there?

David Trainer:23:25
Yeah. Well, we are going to go global. So, you know what what you what you mentioned uh in terms of the core earnings leaders ETF, we want to have one of those in every country of the world.

Nicki Benvenuti:23:30
Wow.

David Trainer:23:38
We want to give everybody the opportunity to invest in a a strategy that actually makes the market smarter. Most of the strategies out there make people money, not market smarter, right? Most of these strategies also they're all they're market cap weighted or they're they're equal weighted. That's the best people can do. We have another uh uh index and ETF concept called the enhanced S&P 500 where we take the same S&P 500 stocks and we weight them based on their core earnings quality as opposed to market cap. And we want those concepts to be everywhere globally. So investment decisions can be driven by the underlying profitability of the business, not some weird theme or whatever it is that people want to invest in. Let's bas let's invest based on true honest fundamentals. And that's a good way to make sure that we're feeding the right company's capital because when we feed the bad companies capital, Nicki, all we do is we give money away to from investors to the executives and to Wall Street. That's a it's it's basically a big theft griff system like the spaxs. People pilot all these spaxs and the spaxs went to zero and we lost money. You know, I mean that's that's basically a transfer of wealth from unsuspecting retail investors to Wall Street insiders. I want to stop that and I want to give investors an opportunity to be in strategies that can not only outperform but also make markets better. That's the win-win we're shooting for and that's the goal and we want to bring that to the whole world, not just to the US.

Nicki Benvenuti:25:11
That's phenomenal. Well, we're really excited. I'm excited to um um follow along in your journey. Thank you so much for um sharing all of these insights today. We're really excited. Um any um last words you'd like for our audience?

David Trainer:25:24
Yeah, absolutely. You know, look, trust that there's a better way to invest. You don't necessarily need to do what you're seeing on CNBC or or or what you know the advisor says or what your brother-in-law at a cocktail party says. There is a way to get to the truth about profitability and valuation. And if that's something that interests you, you should look us up on New Constructs. There's a ton of free insights, ton of free educational materials, tons of free reports, free training programs. I'll show you how it's done. And the reason you can trust us and the reason you always know where you stand with me is that we publish our results good or bad. The scoreboard is always on even when the home team is down and we can trace and audit every single thing we do back to the original filings. So there's 100% auditability. No one else in the world can say that. No one else has papers from Harvard Business School, Ernstston Young and MIT Sloan written about them. No one else has an ETF in Bloomberg indices based on their research that's based on superior earnings. So, it's a special unique thing. It is out there. I know a lot of people are looking for it and we have a solution for you.

Nicki Benvenuti:26:29
That's phenomenal. This is really exciting. Well, thank you so much for your time, David.

David Trainer:26:29
Thank you. Thank Thank you, Nicki. I enjoyed it.