WEBVTT - How Wall Street Is Using AI to Build ETFs 0:00:14.000 --> 0:00:17.240 Hello, and welcome to What Goes Up, a weekly markets podcast. 0:00:17.480 --> 0:00:20.160 My name is Mike Reagan. I'm a senior editor at Bloomberg, 0:00:20.560 --> 0:00:24.000 and I'm gonna higher across Acid reporter with Bloomberg and 0:00:24.160 --> 0:00:27.080 this week on the show. Well, if you've been anywhere 0:00:27.120 --> 0:00:29.640 near the Internet in the last few months, you've probably 0:00:29.640 --> 0:00:32.280 read a poem, or bits of a movie script, or 0:00:32.840 --> 0:00:35.800 maybe even some dad jokes that were written by a 0:00:35.800 --> 0:00:40.040 while a computer. Actually, the Experimental Chat Bought Chat GPT 0:00:40.200 --> 0:00:43.160 has taken the world by storm since its launched in November, 0:00:43.640 --> 0:00:46.800 sugaring a million questions about how this type of technology 0:00:46.880 --> 0:00:50.800 can disrupt various industries and fueling a fresh wave of 0:00:50.840 --> 0:00:54.680 interest in how artificial intelligence can be used by investors. 0:00:55.400 --> 0:00:57.480 We're gonna get into it with the head of research 0:00:57.520 --> 0:00:59.560 at a company that's been using AI for a few 0:00:59.640 --> 0:01:02.720 years now to pick stocks for an almost two billion 0:01:02.720 --> 0:01:07.160 dollar ETF. But uh, first, l donna I try to 0:01:07.200 --> 0:01:10.680 go to chat GPT. I've I've, like everyone else, I've 0:01:10.680 --> 0:01:12.959 fallen into the hype of this chat GPT, and I 0:01:13.000 --> 0:01:15.000 went on trying to get it to write us an 0:01:15.000 --> 0:01:17.840 intro to the podcast, but it's too busy. There are 0:01:17.880 --> 0:01:21.000 too many people using it that they just turn me down. 0:01:22.480 --> 0:01:24.920 I tried for you also, so that we can get 0:01:24.959 --> 0:01:28.440 a nice fun intro from from the robot. And it 0:01:28.520 --> 0:01:30.800 didn't work. And I even try to trick it. I said, 0:01:30.920 --> 0:01:33.839 I have a very simple request and I'm on deadline. 0:01:33.880 --> 0:01:39.440 Please can you help me out? No? Luck, No, robot 0:01:39.480 --> 0:01:41.760 doesn't like me. I wonder the robot must have a 0:01:41.800 --> 0:01:46.280 pr REP that maybe we could to complain. Yeah, it's 0:01:46.520 --> 0:01:51.280 probably a robot pr rep. I don't know. But on 0:01:51.400 --> 0:01:55.160 a pr rep who that is? Also? Yeah? Yeah, they'll 0:01:55.200 --> 0:01:59.400 just tell me to go to hell. So is that 0:01:59.640 --> 0:02:04.080 most of I'm sorry you some of them? Oh my gosh. 0:02:04.960 --> 0:02:07.040 All right, well you taught me which one. I'll get 0:02:07.040 --> 0:02:10.560 back to them. But I do think our guest is 0:02:10.560 --> 0:02:14.640 the perfect guest to unpack this, uh this topic. So 0:02:14.680 --> 0:02:16.600 why don't you bring him in? Yeah he is. It's 0:02:16.680 --> 0:02:19.640 Matt Bartolini. He's the head of Spider America's research at 0:02:19.680 --> 0:02:23.040 Stage Street Global Advisors. And Matt, thanks so much for 0:02:23.120 --> 0:02:26.120 joining us. Yeah, thank you for having me. So we'll 0:02:26.160 --> 0:02:28.440 get into one of your AI E t F s 0:02:28.440 --> 0:02:30.200 in a bit, but maybe just to start, you can 0:02:30.200 --> 0:02:33.240 give us sort of your journey into working with E 0:02:33.320 --> 0:02:36.200 t F and what you do. Yeah. Sure, So my 0:02:36.280 --> 0:02:39.799 journey in TTS is working at State Street Bank essentially 0:02:39.880 --> 0:02:42.280 since the middle of two thousand's and work my way 0:02:42.360 --> 0:02:46.200 up throughout the organization on the County team's portfolio management 0:02:46.200 --> 0:02:48.840 teams and then landed within the E t F teams 0:02:49.320 --> 0:02:51.880 helping to conduct some of the research on our products, 0:02:52.480 --> 0:02:57.040 different portfolio construction topics, investment theses, market outlooks, and market commentaries. 0:02:57.720 --> 0:02:59.760 Uh and that's really where my job is now at 0:02:59.760 --> 0:03:02.720 the of Spider America's research. You know, our job is 0:03:02.760 --> 0:03:05.560 to help makes sense of a complex world by using 0:03:05.680 --> 0:03:09.280 data driven insights, and we write market commentaries, market outlooks, 0:03:09.919 --> 0:03:14.480 provide some portfolio instruction discussions to end advisors and hopefully 0:03:14.480 --> 0:03:17.639 help them, you know, select the right investment choice for them. 0:03:17.639 --> 0:03:20.120 And if that's a Spider et F then I think 0:03:20.160 --> 0:03:22.519 that's all great, But in some cases it doesn't. And 0:03:22.600 --> 0:03:25.880 that's how we operate, trying to be fair and balanced. Yea, 0:03:26.480 --> 0:03:29.440 and uh Ai is obviously one of your areas of research. 0:03:29.520 --> 0:03:34.160 Match I'm curious, you know, this chat CHPT thing to me, 0:03:34.240 --> 0:03:35.560 and I think to a lot of people, just can't 0:03:35.600 --> 0:03:37.080 it seem to have come out of nowhere. You know, 0:03:37.120 --> 0:03:40.839 it launched in November, and you know, granted I don't 0:03:40.840 --> 0:03:42.960 follow the space that closely, but I think for a 0:03:43.000 --> 0:03:47.000 lot of people it was just sort of dumbfounding how 0:03:47.040 --> 0:03:49.080 good this thing is right at launched, you know it 0:03:49.120 --> 0:03:51.360 to me, I would have expected sort of a to 0:03:51.360 --> 0:03:54.080 see a cruder version of this that wasn't quite as impressive. 0:03:54.120 --> 0:03:57.000 But how did you see it like it was? Were 0:03:57.040 --> 0:03:59.840 you sort of as surprised as everyone else for the 0:04:00.080 --> 0:04:03.040 you're sort of research into AI? Had it led you 0:04:03.080 --> 0:04:06.000 to kind of know this type of thing was possible 0:04:06.160 --> 0:04:09.240 and in the pipeline? Yeah, So a lot of the 0:04:09.320 --> 0:04:12.800 AI work we've done is is within sort of portfolio 0:04:12.920 --> 0:04:16.120 construction and index selection on some of our funds, so 0:04:16.160 --> 0:04:18.400 we were aware of the ability to use things like 0:04:18.480 --> 0:04:22.000 natural language processing, predictive text, but also even just in 0:04:22.040 --> 0:04:24.320 our daily lives. I think some of the functions of 0:04:24.320 --> 0:04:27.719 Czech GP two we've probably just been benefiting from just 0:04:27.800 --> 0:04:31.440 in very small morsels, whether that is, you know, auto 0:04:31.720 --> 0:04:35.479 correcting your text, or the predictive text nature within your 0:04:35.480 --> 0:04:37.719 iPhone of what you might say next, like that's sort 0:04:37.760 --> 0:04:40.719 of the same idea, or even you know, when when 0:04:40.760 --> 0:04:43.360 we use the Bloomberg terminal and we asked the help desk, 0:04:43.400 --> 0:04:45.960 we sometimes get a very automated response back that's all 0:04:46.000 --> 0:04:48.400 sort of pieces of it. Uh. The first time I 0:04:48.440 --> 0:04:50.520 saw it, you know, we were sort of playing around 0:04:50.520 --> 0:04:52.680 with it of you know, write us a blog post 0:04:52.880 --> 0:04:57.279 about the benefits of ETFs and it got it probably 0:04:57.320 --> 0:04:59.800 eighty per correct, you know, how we would want to 0:05:00.000 --> 0:05:01.839 structure of the argument. And I think that's sort of 0:05:01.839 --> 0:05:04.800 where chat chypt is is that it kind of gives 0:05:04.800 --> 0:05:07.680 you about an eight. And now sort of joking with 0:05:07.880 --> 0:05:09.880 you know, some of my colleagues who have old their 0:05:09.960 --> 0:05:13.000 kids that you know, chet chypt would probably be be 0:05:13.040 --> 0:05:16.159 a B minus student if it only ever turned in 0:05:16.240 --> 0:05:19.680 its homework, because that's kind of the surface level it gets. 0:05:20.040 --> 0:05:22.320 And I have a friend who's a professor at a 0:05:22.400 --> 0:05:25.080 college and they've actually started to work on how to 0:05:25.160 --> 0:05:28.960 figure out you know, essays and reports that are written 0:05:29.440 --> 0:05:32.440 through AI. And the big thing is you've got to 0:05:32.480 --> 0:05:35.520 look at the nuance. And chat gypt doesn't really understand 0:05:35.600 --> 0:05:39.480 the complexity of nuances, particularly for topics like ets where 0:05:39.480 --> 0:05:42.560 there actually is a lot of operational nuance. You know, 0:05:43.120 --> 0:05:45.520 as a B bonus student, myself. That explains why I 0:05:45.560 --> 0:05:48.320 was so impressed. I think if you are a student 0:05:48.400 --> 0:05:50.200 right now, you could use it to help boost your 0:05:50.200 --> 0:05:52.839 grades a little bit. Yeah, well that's I think the 0:05:52.880 --> 0:05:55.080 fear of everyone has. I have a nine year old 0:05:55.080 --> 0:05:57.520 son who had to do a penguin project and he, 0:05:57.760 --> 0:06:01.200 instead of looking in a book, he yelled out to Alexa, 0:06:01.480 --> 0:06:04.359 you know how how fast you to penguin swim? And 0:06:04.360 --> 0:06:06.800 I had to tell them they can't do that. So 0:06:08.120 --> 0:06:11.080 the new reality that we're all living in, ask can 0:06:11.080 --> 0:06:16.159 penguins swim? I don't know what the the ask Alexa, No, 0:06:16.279 --> 0:06:19.400 don't ask Alexa. Well, no, my kids it's the same thing. 0:06:19.440 --> 0:06:22.320 They're sitting there doing their homework and I hear them, y'all, hey, Google, 0:06:22.360 --> 0:06:25.440 what's you know? Nine times thirty seven? And I mean, 0:06:25.480 --> 0:06:27.640 in some ways, it's just a calculator, and I think 0:06:27.880 --> 0:06:30.520 educators are gonna have to get used to it and 0:06:30.880 --> 0:06:32.960 allow it in some to some degree, I don't know, 0:06:33.000 --> 0:06:35.480 it's it's such a strange new world. But but Matt 0:06:35.520 --> 0:06:39.440 talked to us about the spider Ken show New Economy ZTF, 0:06:39.480 --> 0:06:42.960 which actually has been using AI to to pick stocks, 0:06:43.000 --> 0:06:45.880 you know, for for sort of the Layman among us 0:06:45.920 --> 0:06:50.039 South there. How how exactly does AI help in this, uh, 0:06:50.160 --> 0:06:54.400 this stock picking effort. Yeah, So the artificial intelligence behind 0:06:54.440 --> 0:06:57.240 it is natural language processing and this is run by 0:06:57.320 --> 0:07:00.040 the index provider S and P. To actually start it 0:07:00.120 --> 0:07:02.480 over the firm Ken Show there was a small startup 0:07:02.520 --> 0:07:06.320 that was incubated out of Goldman Sachs. UH SMP bought 0:07:06.360 --> 0:07:08.760 that firm and all of the I P along with it, 0:07:08.800 --> 0:07:12.520 and that's our index provider for the fund. And you 0:07:12.560 --> 0:07:16.400 know the NLP or natural natural Language process and what 0:07:16.520 --> 0:07:21.320 it does. Scans through UH perspectives and other regulatory filings 0:07:21.320 --> 0:07:24.320 from companies because you want to start with a strong source. 0:07:24.400 --> 0:07:28.040 Regulatory filings have to be quite prescriptive and if you, 0:07:28.040 --> 0:07:31.880 you know, make falsehoods about that, uh, there's penalties, right. UM. 0:07:32.080 --> 0:07:36.120 So the scans through UH regulatory documents searching for key 0:07:36.280 --> 0:07:40.760 terms to identify how these firms material operations correlate back 0:07:40.920 --> 0:07:45.480 to areas of innovation, whether it's like enterprise collaboration, clean energy, 0:07:46.040 --> 0:07:51.000 advanced transport systems, drones. So scan through all these UM 0:07:51.680 --> 0:07:55.400 regulatory documents looking for the frequency of a term used, 0:07:55.440 --> 0:07:58.800 but also the words around it, so you know if 0:07:58.840 --> 0:08:01.720 a company is saying that drow own technology is incredibly 0:08:01.760 --> 0:08:04.440 important for the future of growth of our business. That 0:08:04.520 --> 0:08:08.160 really shows some emphasis towards that type of innovation. So 0:08:08.240 --> 0:08:12.160 that would be scanned and recorded and classified appropriately into 0:08:12.280 --> 0:08:16.920 twenty five different areas of innovation, and then from their 0:08:17.080 --> 0:08:20.080 stocks are weighted and more of a modified equal weighted 0:08:20.360 --> 0:08:25.640 structure where core firms to a specific innovation are overweighted 0:08:25.640 --> 0:08:28.640 to non core firms. So basically, the you know, the 0:08:28.680 --> 0:08:31.600 way we sort of describe it is that the AI 0:08:31.640 --> 0:08:34.960 process selects the stocks and then there's a quantitative weighting 0:08:35.000 --> 0:08:38.679 methodology to weight the stocks. But the reason why we 0:08:38.720 --> 0:08:41.160 went down this path of using AI s that we 0:08:41.280 --> 0:08:45.360 wanted something forward looking, something dynamic, because you know, back 0:08:45.360 --> 0:08:48.000 in two thousand and eighteen, we understood that in the 0:08:48.040 --> 0:08:50.360 E t F world, there weren't a lot of strategies 0:08:50.360 --> 0:08:55.320 that were this forward looking innovative type paradigm. A lot 0:08:55.360 --> 0:08:58.080 of it was based on revenue, and revenue was what 0:08:58.160 --> 0:09:01.720 has already been realized that a backward looking approach, and 0:09:01.720 --> 0:09:03.960 we wanted something that was more dynamic and a ford 0:09:03.960 --> 0:09:07.040 looking approach in the AI process was able to deliver 0:09:07.120 --> 0:09:09.520 that for us. Okay, so before you tell us more 0:09:09.559 --> 0:09:12.040 about that I am interested in sort of the mechanics. 0:09:12.080 --> 0:09:16.559 So once the AI runs through and chooses these companies 0:09:16.640 --> 0:09:19.640 that it fits, that it thinks fits thinks I don't 0:09:19.640 --> 0:09:21.040 know I thinks is the right word, but that it 0:09:21.120 --> 0:09:24.880 chooses as fitting the right criteria, do you then have 0:09:24.960 --> 0:09:29.439 a human go through the results and say, okay, this 0:09:29.640 --> 0:09:32.200 actually sounds pretty good, or maybe we don't want to 0:09:32.280 --> 0:09:35.079 have X y Z company as part of this portfolio. 0:09:35.840 --> 0:09:38.640 So within the index methodology there is sort of a 0:09:38.720 --> 0:09:41.960 human control element to it, most like a quality control. 0:09:42.040 --> 0:09:45.599 So for instance, uh, if you know a company is 0:09:45.640 --> 0:09:50.600 classified as innovating within Clean Energy UM, they use the 0:09:50.760 --> 0:09:54.000 term your wind and solar are quite significantly that it 0:09:54.120 --> 0:09:57.200 said it's part of the material operations. But when it 0:09:57.240 --> 0:10:00.480 comes down to it, there's a check and balance from 0:10:00.480 --> 0:10:03.080 the index committee to say, okay, well, does firm x 0:10:03.200 --> 0:10:06.319 y Z offer a product and service in this category 0:10:06.920 --> 0:10:09.520 or they just some sort of you know, this is 0:10:09.559 --> 0:10:11.480 probably a bad term, but some sort of shell company 0:10:11.480 --> 0:10:15.080 that doesn't actually provide a product or service. They just yeah, 0:10:15.080 --> 0:10:17.200 this isn't what they do. They just say saying something 0:10:17.200 --> 0:10:19.840 that doesn't correlate back to their actual products and services. 0:10:19.880 --> 0:10:21.680 So that's where there's a little bit of a manual 0:10:21.800 --> 0:10:27.160 quality check to ensure that these firms are actually engaged 0:10:27.600 --> 0:10:29.920 in these areas of innovation and they are not just 0:10:30.120 --> 0:10:34.439 talking about it sort of you know, extemporaneously. And the 0:10:34.760 --> 0:10:38.200 other thing is too that helps in terms of you know, 0:10:38.559 --> 0:10:41.360 get let's say perfect you know, we have a champagne 0:10:41.400 --> 0:10:44.360 problem that this fund becomes a hundred and ninety billion 0:10:44.400 --> 0:10:46.760 dollars and someone wants to get into it and they 0:10:46.800 --> 0:10:49.400 just use the word drone a thousand times to game it. 0:10:49.760 --> 0:10:52.679 That helps, right, that sort of oat manual overrides sort 0:10:52.679 --> 0:10:55.640 of quality check. What I find fascinating about it something 0:10:55.679 --> 0:10:58.080 like five sixty holdings, you know, so it's not not 0:10:58.160 --> 0:11:02.520 a very concentrated fun and you know, when you're looking 0:11:02.520 --> 0:11:05.000 for innovative sort of startup type of companies, a lot 0:11:05.000 --> 0:11:10.560 of times that means really small even maybe microcap companies, uh, 0:11:10.679 --> 0:11:13.120 that you have to dig through, which are not typically 0:11:14.559 --> 0:11:17.480 very heavily followed by you know, the Wall Street analysts 0:11:17.520 --> 0:11:21.400 class just by definition, you know, if there's thousands of them, um, 0:11:21.520 --> 0:11:25.000 and this really surprised me. Uh. And you know what 0:11:25.080 --> 0:11:27.840 you say about forty eight percent of the holdings have 0:11:28.000 --> 0:11:31.920 less than ten analysts covering the stock. Is that almost 0:11:32.000 --> 0:11:35.080 a benefit for this type of strategy that it helps 0:11:35.080 --> 0:11:39.200 you sort of find these hidden gems that are maybe 0:11:39.640 --> 0:11:43.960 being completely overlooked by by the masses out there. Yeah, 0:11:44.000 --> 0:11:47.080 I mean, AI at its heart is to help increase 0:11:47.120 --> 0:11:51.440 efficient efficiencies and productivity. And what this does is allows 0:11:51.520 --> 0:11:54.199 us to cover the uncovered. So if you're using an 0:11:54.200 --> 0:11:57.880 analyst recommendations, analysts can only cover so many stocks within 0:11:57.920 --> 0:12:01.120 a given day. And there's candies some firms that are 0:12:01.200 --> 0:12:05.160 quite innovative, they're you know, performing and producing some really 0:12:05.800 --> 0:12:09.200 interesting things within our economy. You know, whether it's things 0:12:09.200 --> 0:12:13.360 within advanced healthcare like wearables that aren't really covered by 0:12:13.679 --> 0:12:16.960 Wall Street analysts because they might be smaller capitalization securities. 0:12:16.960 --> 0:12:18.760 And we sort of just know this even from like 0:12:18.800 --> 0:12:22.880 traditional finance, that the majority of analysts recommendations are in 0:12:22.920 --> 0:12:26.640 that large caps space um and then small caps and 0:12:26.640 --> 0:12:29.360 maycaps sort of do you not get as much notoriety 0:12:29.440 --> 0:12:33.120 or coverage. And AI is basically is one way to 0:12:33.160 --> 0:12:37.000 solve that problem, to give you a deeper breath of 0:12:37.040 --> 0:12:41.560 opportunities and really broaden your scope of companies that may 0:12:41.600 --> 0:12:51.240 be considered innovative. So I want to give a shout 0:12:51.280 --> 0:12:54.600 out to Katie Greifield and Sam Potter on the Cross 0:12:54.640 --> 0:12:57.400 has a team at Bloomberg, because they had this really 0:12:57.440 --> 0:13:01.040 fascinating story that said something like, we asked chat GPT 0:13:01.320 --> 0:13:04.319 to create an e t F for us, and here's 0:13:04.320 --> 0:13:06.760 the results, and and actually it had done a really 0:13:06.800 --> 0:13:10.640 good job putting something together. And you were part of 0:13:10.679 --> 0:13:13.360 this story. And Katie and I were chatting about it afterwards, 0:13:13.400 --> 0:13:17.160 and she said, Matt had all these insights into the 0:13:17.160 --> 0:13:21.199 composition aspect of because you guys have your own AI 0:13:21.280 --> 0:13:23.800 E t F. And I do wonder about that, like, 0:13:24.000 --> 0:13:27.560 is the power of the AI being able to create 0:13:27.600 --> 0:13:29.640 an E t F? Is the power? Does it lie 0:13:29.720 --> 0:13:32.079 in the sheer amount of work that it can do, 0:13:32.480 --> 0:13:35.160 whereas you might not be able to have like a 0:13:35.200 --> 0:13:39.240 team of humans combing through so many different things to 0:13:39.280 --> 0:13:41.840 the point where they get to an e t F 0:13:41.920 --> 0:13:46.040 that has five and sixty components. Yeah, it's it's all 0:13:46.080 --> 0:13:50.200 about creating efficiencies and being able to capture, you know, 0:13:51.120 --> 0:13:55.480 undiscovered or unrepresented areas within the equity markets. You know, 0:13:55.520 --> 0:13:59.800 even just looking in core portfolios, disruption happens further down 0:13:59.800 --> 0:14:03.680 that cap spectrum. And that's why using something that is 0:14:03.720 --> 0:14:08.480 able to explore data sets that are really unstructured because 0:14:08.520 --> 0:14:12.839 revenue profiles balances those more structured data sets. But using 0:14:13.360 --> 0:14:17.880 textual language processing to identify firms based on what their 0:14:17.920 --> 0:14:21.680 material operations are saying is one way to help classify 0:14:21.840 --> 0:14:24.840 them into these areas of innovation. And I think one 0:14:24.840 --> 0:14:27.600 of the things about this fund in particular is that 0:14:28.240 --> 0:14:31.560 we do understand that it is not innovation does not 0:14:31.720 --> 0:14:36.480 just benefit the pure place, is that the ecosystem around 0:14:36.520 --> 0:14:39.240 it can benefit. You know, the whole idea during the 0:14:39.640 --> 0:14:42.040 gold rush of the eighteen hundreds of it would rather 0:14:42.120 --> 0:14:44.480 mind for gold or sell the pick axes and the 0:14:44.560 --> 0:14:46.920 tents to go along with it. You probably had a 0:14:46.920 --> 0:14:48.920 pretty good business model if you're selling a lot of 0:14:48.920 --> 0:14:52.480 pick axes in the eighteen hundreds. And that's sort of 0:14:52.480 --> 0:14:55.480 the idea here is, you know, the ecosystem is also beneficial, 0:14:55.840 --> 0:14:59.080 and how do you identify that ecosystem? Uh? You know, 0:14:59.120 --> 0:15:02.080 affirm like Video for example, they make all of the 0:15:02.120 --> 0:15:06.240 sensory technology with an autonomous vehicles. That's a supplier to 0:15:06.480 --> 0:15:10.320 that ecosystem. And as autonomous vehicles take off, they're going 0:15:10.360 --> 0:15:14.360 to benefit as well, so using AI to to detect 0:15:14.440 --> 0:15:20.080 that can really help create a really targeted, but diversified 0:15:20.080 --> 0:15:23.680 portfolio of innovative stocks. And basically this e t F 0:15:23.880 --> 0:15:27.520 has many more components than it would if a team 0:15:27.560 --> 0:15:30.840 of humans was putting it together. Right, yeah, so the 0:15:31.160 --> 0:15:35.320 you know the statistic, they're over ten analysts. So let's 0:15:35.320 --> 0:15:37.320 just say we use that as an example, like we 0:15:37.400 --> 0:15:40.680 needsily at least be covered by ten ten analysts will 0:15:40.880 --> 0:15:43.440 right then and there we lose half the portfolio and 0:15:43.560 --> 0:15:47.560 tends not a big number. So if we were to 0:15:47.880 --> 0:15:50.120 take more of a human based approach to it, it 0:15:50.160 --> 0:15:54.880 would be far more concentrated with portfolio. And that's what 0:15:54.920 --> 0:15:57.240 we see with the other broad innovation e t f 0:15:57.320 --> 0:16:00.200 s out there, is that they're far more concentrated and 0:16:00.240 --> 0:16:04.000 they're also far more geared towards large cap security. So 0:16:04.040 --> 0:16:08.240 you do not get the differentiation that you would want 0:16:09.080 --> 0:16:12.560 in something that is supposed to be innovative and you know, 0:16:12.640 --> 0:16:17.640 not largely represented within core portfolio. Is right when this 0:16:17.760 --> 0:16:20.200 fund was launched, I guess it was three or four 0:16:20.280 --> 0:16:24.840 years ago. You know, growth stocks, innovative disruptive stocks were 0:16:25.280 --> 0:16:27.880 you know, the hottest things going in the market, and 0:16:27.920 --> 0:16:30.160 the funded great, you know, a few thousand and nineteen 0:16:30.200 --> 0:16:33.600 up thirty seven, two thousand twenty up sixty one percent, 0:16:34.000 --> 0:16:37.760 up about four twenty one. Then obviously last year was 0:16:38.040 --> 0:16:40.960 kind of the rug pole out from under growth and 0:16:41.000 --> 0:16:44.720 innovation down. So I'm wondering, you know, is there a 0:16:44.720 --> 0:16:48.240 way to layer AI on top of a fund like 0:16:48.320 --> 0:16:51.800 this to allow to sort of shift to a value 0:16:51.800 --> 0:16:56.440 strategy or to kind of sniff out the market cycle 0:16:56.560 --> 0:16:59.640 into what's kind of the the new hot factor to 0:16:59.720 --> 0:17:02.080 get to UM. I know that's not the goal of 0:17:02.080 --> 0:17:03.720 this fund, but I wonder if you think about that, 0:17:03.760 --> 0:17:05.439 you know, is there is there a way to not 0:17:05.480 --> 0:17:08.920 only pick the the individual stocks under a certain theme 0:17:09.000 --> 0:17:13.040 or strategy like this, but so also have that strategy 0:17:13.280 --> 0:17:17.000 sort of evolve over time and try to you know, 0:17:17.160 --> 0:17:20.040 isolate the upcoming market cycle and what's gonna what's the 0:17:20.119 --> 0:17:23.280 leadership is gonna be? Uh in case growth does have 0:17:23.480 --> 0:17:27.200 a down draft like this, So I mean, that's when 0:17:27.240 --> 0:17:30.920 that becomes just market timing, so to speak. Right, and 0:17:30.960 --> 0:17:34.600 you're now you're now doing some former factor rotation. You know, 0:17:34.640 --> 0:17:38.720 I think you could but perhaps create more of a 0:17:38.800 --> 0:17:44.080 style style neutral innovative portfolio, but that becomes much harder 0:17:44.560 --> 0:17:48.840 because then you're going to have in an optimization framework, 0:17:49.800 --> 0:17:52.359 the optimizer is gonna be working really hard to mitigate 0:17:52.440 --> 0:17:55.680 any of that small cap bias, so and then you're 0:17:55.720 --> 0:17:58.680 just gonna basically look like you know, a large cap 0:17:58.720 --> 0:18:02.600 growth tech exposure. So then it's always this trade off 0:18:02.640 --> 0:18:05.399 of like, do I want to mitigate some of these 0:18:05.440 --> 0:18:10.639 implicit style factors and get you know, sort of close 0:18:10.760 --> 0:18:16.320 up that tracking risk to traditional benchmarks, or maintain the 0:18:16.359 --> 0:18:19.440 purity of what we're trying to do of innovative exposures. 0:18:19.440 --> 0:18:21.320 So you always try to find that balance. And if 0:18:21.320 --> 0:18:25.439 you try to create more style neutral or or something 0:18:25.440 --> 0:18:30.360 that is, you know, less impacted by market cyclical factors, 0:18:31.160 --> 0:18:32.800 then you're gonna lose some of the purity of your 0:18:32.840 --> 0:18:35.879 intended focus. And I think when we are having discussion 0:18:36.000 --> 0:18:39.639 ground performance, we always just go back to attribution and 0:18:39.680 --> 0:18:43.840 we will use you know, UH fundamental risk models. And 0:18:43.880 --> 0:18:47.919 if we look at it, since inception, industry and stock 0:18:47.960 --> 0:18:51.439 selection effects relative to UH, you know, the like the 0:18:51.520 --> 0:18:55.439 SMP fIF for example, industry and stock selection effects have 0:18:55.520 --> 0:18:59.800 been positive to UH. The funds return has been a 0:19:00.040 --> 0:19:03.800 dative to performance. The industry party is interesting because there 0:19:03.840 --> 0:19:10.360 are some industries like semiconductor software, um you uh, sort 0:19:10.400 --> 0:19:14.600 of wearable technologies within healthcare, those industries are gonna be 0:19:14.600 --> 0:19:18.119 more innovative than say some firms within like staples and 0:19:18.240 --> 0:19:23.000 you're sort of consumer goods products. So industry effects byproduct 0:19:23.080 --> 0:19:26.760 of the folks of innovation. Stock selection effects is by 0:19:26.840 --> 0:19:30.320 product of the AI selection methodology and then the waiting 0:19:31.000 --> 0:19:36.560 um process. The detractors of returns have been style factors, 0:19:36.880 --> 0:19:42.159 namely higher volatility, lower quality, and high high growth you know, 0:19:42.280 --> 0:19:47.520 since inception. But those factors are are implicit because it's 0:19:47.520 --> 0:19:50.600 not what we're we're seeking to obtain. But they're also cyclical. 0:19:50.800 --> 0:19:58.760 So high volatility, low quality, high growth were being famously rewarded, 0:19:59.240 --> 0:20:03.560 uh star and through you know sort of mid right, 0:20:03.600 --> 0:20:06.280 So that was as a tail wind two returns back then. 0:20:07.160 --> 0:20:09.360 So that's how we always like to frame the performance 0:20:09.400 --> 0:20:13.760 conversation is breaking those three components out, noting that the 0:20:13.840 --> 0:20:18.040 style components are going to be cyclical and move in 0:20:18.040 --> 0:20:21.560 and out based on market directions. I'm always curious how 0:20:21.640 --> 0:20:25.280 E t F issuers decide on a theme or a 0:20:25.400 --> 0:20:28.919 topic or you know, putting an e T F together. 0:20:29.080 --> 0:20:32.320 So a couple of years ago. Was AI something that 0:20:32.400 --> 0:20:35.120 you guys when you got together, we're thinking was going 0:20:35.160 --> 0:20:37.639 to be a big deal in the coming years, or 0:20:37.720 --> 0:20:40.159 is it sort of which I think this happens a 0:20:40.200 --> 0:20:42.280 lot in the E T F space. Let's just put 0:20:42.320 --> 0:20:45.639 it out there, give it a try, and see what happens. 0:20:45.680 --> 0:20:48.800 So it's definitely not the ladder within our firm, We're 0:20:48.800 --> 0:20:51.920 definitely not that. Yeah, we're we're not gonna be like, hey, 0:20:52.040 --> 0:20:53.920 this is a hot dot, let's throw it out there 0:20:53.920 --> 0:20:56.639 and see if it works. You know, that's just not 0:20:56.720 --> 0:21:01.600 what we do UM with respect these funds. We have 0:21:01.640 --> 0:21:05.119 a pretty strong heritage within sector and industry investing, and 0:21:05.200 --> 0:21:07.680 we know that there are thematic investors out there. We 0:21:07.720 --> 0:21:10.200 see it all the time within our traditional industry suite. 0:21:10.240 --> 0:21:12.920 You know, someone that wants to play a rally and 0:21:13.000 --> 0:21:16.520 oil stocks will go by xop our Oil and Gas 0:21:17.080 --> 0:21:19.840 ETF and that's a thematic investor. And we knew that 0:21:19.880 --> 0:21:23.679 thematic investing was was going to be UM on the 0:21:23.800 --> 0:21:29.480 rise because there's some thematics like auntonomous vehicles or cybersecurity 0:21:29.560 --> 0:21:34.040 or clean energy that are hard to to to gain 0:21:34.160 --> 0:21:37.840 exposure to under a traditional GETS framework. Because some of 0:21:37.840 --> 0:21:41.720 these firms are are operate across gig sectors. You know, 0:21:41.840 --> 0:21:45.840 clean energy is a perfect example. You have firms within 0:21:45.880 --> 0:21:50.959 the legacy energy sector, the utility sector, industrial sector, technology sector, 0:21:51.200 --> 0:21:54.440 so you want to go across the sectors. So we 0:21:54.440 --> 0:21:57.040 were like, well, how do you go about doing this again? 0:21:57.119 --> 0:21:59.040 We wanted something that was forward looking. We knew that 0:21:59.080 --> 0:22:01.919 revenue was back with game. So this is how we 0:22:02.000 --> 0:22:04.199 landed on, you know, firm like Ken Show and then 0:22:04.280 --> 0:22:06.560 later obviously S and P Ken Show as a combined 0:22:06.680 --> 0:22:10.960 entity of having a really unique value proposition of using 0:22:11.400 --> 0:22:16.160 natural language processing to detect firms that are listing out 0:22:16.200 --> 0:22:23.119 these innovative UM services or innovative corporate designs as part 0:22:23.160 --> 0:22:26.840 of their material operations. UM. So that's that was really 0:22:26.880 --> 0:22:28.520 the impetus for it. And I think, you know, I 0:22:28.840 --> 0:22:32.320 sort of remember one instance. It was I think it 0:22:32.440 --> 0:22:35.680 was obviously before we launched, was probably time frame when 0:22:35.720 --> 0:22:38.919 we're really starting to kick the tires on this. The 0:22:39.000 --> 0:22:45.680 Pokemon vert augmented reality iPhone app was just really really popular. 0:22:45.680 --> 0:22:49.040 I remember playing softball and seeing a bunch of people 0:22:49.119 --> 0:22:52.000 like hanging out by the left field tree. We had 0:22:52.000 --> 0:22:54.879 no idea why, and someone put a Pokemon stay. I 0:22:54.880 --> 0:22:57.679 don't play this, so I have no idea, and I 0:22:57.800 --> 0:23:00.359 remember talking to folks and internally like you would be 0:23:00.359 --> 0:23:03.720 really interesting if we could have something that focused on 0:23:04.000 --> 0:23:07.840 these type of firms, you know, innovating within virtual reality 0:23:07.840 --> 0:23:11.400 and augmented reality coincided at the same time as we're 0:23:11.480 --> 0:23:14.280 kicking the tires on on this process. And that's kind 0:23:14.280 --> 0:23:18.040 of the idea now, owning twenty stocks and augmented reality, 0:23:18.119 --> 0:23:20.760