advisoranalyst.com Open in urlscan Pro
192.124.249.3 Public Scan

Back to summary

Submitted URL:
https://sendy.advisoranalyst.com/l/zmo0rd9aWOwzN892dqRDz3oA/qujndU4nAzshC4BTW8HLFA/sbra5CDym7FIetVcGF2G7g
Effective URL:
https://advisoranalyst.com/2023/04/12/chatgpt-ai-systemic-risk-the-case-for-dividends-feat-sri-iyer.html/
Submission: On April 27 via api (April 27th 2023, 3:30:43 am UTC) from CH — Scanned from DE

Form analysis
2 forms found in the DOM

GET https://advisoranalyst.com/

<form role="search" method="get" class="cs-search__nav-form" action="https://advisoranalyst.com/">
  <div class="cs-search__group">
    <button class="cs-search__submit">
      <i class="cs-icon cs-icon-search"></i>
    </button>
    <input data-swpparentel=".cs-header .cs-search-live-result" required="" class="cs-search__input" data-swplive="true" type="search" value="" name="s" placeholder="Enter your search topic">
    <button class="cs-search__close">
      <i class="cs-icon cs-icon-x"></i>
    </button>
  </div>
</form>

POST

<form method="post" class="subscription">
  <input type="hidden" name="list_id" value="d1ef406aa8">
  <div class="pk-input-group">
    <input type="text" name="USER" class="user form-control" placeholder="Enter your name">
    <input type="text" name="EMAIL" class="email form-control" placeholder="Enter your email">
    <button class="pk-subscribe-submit" type="submit">Subscribe</button>
  </div>
  <input type="hidden" name="_wp_http_referer" value="/2023/04/12/chatgpt-ai-systemic-risk-the-case-for-dividends-feat-sri-iyer.html/">
</form>

Text Content

* Markets
* Watch
* Podcast
* ETFs
* ESG
* Spotlight
* Appointment Notices
* US Stocks
* Canadian Market
* Economy
* CE
* AA on Twitter
* AA on Instagram
* AA on Linkedin
* About

Investment and market analysis, commentary, outlook, and practice resources for
financial industry professionals.

* Home
* Markets
* Commentary
* Alternative Investments
* A - J
* AllianceBernstein
* Brooke Thackray
* David Rosenberg
* Don Vialoux
* Fidelity Investments
* Frank Holmes, US Funds
* Guy Haselmann
* Invesco Canada
* Jeffrey Gundlach
* Jeffrey Saut, Raymond James
* Jeremy Grantham, GMO
* L - Z
* Liz Ann Sonders
* LPL Financial
* Mawer Investment Management
* Newfound Research
* Purpose Investments
* Ray Dalio
* Resolve Asset Management
* Russ Koesterich, Blackrock
* William Smead
* Som Seif, Purpose Investments
* Warren Buffett
* WealthTrack
* Stocks
* SIA CHARTS
* Canada Section
* Canadian Market
* New Alternatives
* US Stocks
* Europe
* Global Investing
* Emerging Markets
* China
* BRIC
* ETFs
* Earnings
* Value Investing
* Passive Investing
* Fixed Income
* Bond Market
* Bonds
* Corporate Debt
* Credit Markets
* EM Bonds
* High Yield
* Alternative Investments
* Spotlight
* Appointments
* ETFs
* ESG
* Video
* Podcast
* About
* Advertise
* Privacy Policy
* Terms of Use
* Advisor
* CE

THE LATEST

NOW WHAT?

SAVITA SUBRAMANIAN: RECESSION, SHMECESSION. OWN STOCKS OVER BONDS, CYCLICALS
OVER DEFENSIVES

TS LOMBARD: DOUBLE TROUBLE AHEAD

JPM: NARROWEST STOCK LEADERSHIP, EXTREME CONCENTRATION, DEFENSIVE FACTOR
CROWDING UNDERWAY

* `
* Zweig
* Zuce
* Zorba The Greek
* Zone News
* Zone Exposure
* Zone 7
* Zoloft
* zlotnikov
* ZLB.TO

CHATGPT, AI, SYSTEMIC RISK & THE CASE FOR DIVIDENDS

Total
0
Shares
0
0
0
0

1:06:27

Read the transcript

[00:00:00] Speaker 1: This is the Insight is Capital Podcast.

[00:00:07] Speaker 2: The views and opinions expressed in this podcast are those
of the individual guests and do not necessarily reflect the official policy or
position of advisoranalyst.com or of our guests. This broadcast is meant to be
for informational purposes only. Nothing discussed in this broadcast is intended
to be considered as advice.

[00:00:21] Pierre Daillie: Welcome back. I’m Pierre Daillie, Managing Editor at
advisoranalyst.com. My very special guest today is Sri Iyer, Lead Portfolio
Manager and Managing Director and Head of i3 Investments at Guardian Capital LP.
Sri, welcome back. It’s great to see you again and catch up with you at this
pivotal moment in the economy and monetary policy and markets. What’s new with
you? And how are you?

[00:00:48] Sri Iyer: Pierre, again, glad to be back. Well, it looks like every
time we talk to each other, something big is happening in the market. So-

[00:00:55] Pierre Daillie: For sure.

[00:00:56] Sri Iyer: … [inaudible 00:00:57] own good, and I think since we last
spoke, a lot of the things we discuss have come to fruition. So, it’s kind of a
very good point of time to kind of reconnect with you and try to get you some
ideas as to what we’re thinking and what we’ve seen where things are going. So,
really looking forward to our chat today.

[00:01:13] Pierre Daillie: Yeah, um, me as well. Very excited, because I, you
know, I think we’re, you know, as I mentioned, we’re in this, we’re at this,
what seems to be like a turning point or a pivotal moment in markets and, you
know, this meltdown of SVB. It’s really highlighted, you know, the, the, the,
the story that people tell that the Fed is almost always responsible for
breaking the economy, once, you know, we, we, once it, it engages in a rate
hiking regime.

[00:01:41] Sri Iyer: Mm-hmm.

[00:01:42] Pierre Daillie: And that’s exactly, that’s exactly what we’re
starting to see the signs of, right?

[00:01:46] Sri Iyer: Yeah, I, I would say the average person sees a bipolar
response in the market right now by the Fed. Um, my take is the Fed, in my
opinion, actually is having a very measured response to the market cycle. Um,
the real risk of market happen in the blockchain Bitcoin side. If there was no
Bitcoin, things could be much, much worse and things could have precipitated
much more, because a lot of the risk capital went into that side of the market-

[00:02:21] Pierre Daillie: Right.

[00:02:22] Sri Iyer: … um, response. So, with the unwinding of the Bitcoin
phase, and you saw the meltdown in that phase with, with, with a blow up in a
couple of exchanges on that side. The market got a taste of what, uh, de-risking
really looks like. Despite that, as we say, don’t fight the Fed is the common
adage on the statement, um, I think the Fed is still committed to reducing
inflation and I don’t think so the market should construe this bailout however
ad hoc it has been to be a signal towards any kind of pivot to say that there’s
going to be a backstop to a measured response to the market.

So, about three months ago, the standard response to the market was the
subconscious bailout was going to come up from a pivot. Now, with persisted
inflation just because CPI just came out today and you saw that the inflation
has been persisted across virtually every category you can imagine-

[00:03:25] Pierre Daillie: Right.

[00:03:25] Sri Iyer: … other than used car, used car sales is down about 2%, big
D. So, what you’re seeing here is persistent inflation we should want the Feds
targeting. What the Fed is saying on one side is that we’re going to keep going
at that target at any cost, but the definition of costs is not a bailout for
some kind of failing institution, but more about protecting the average investor
because of a contagion response to consumption and the consumer in general. So,
I don’t think so the Fed has taken a bucket to a tsunami yet. That’s my two
cents. I think this response to SVB and anything else is actually a good
decision by the Fed because you want to take the air out of the balloon in a
measured way as much as possible and not pop it.

[00:04:28] Pierre Daillie: Yeah.

[00:04:28] Sri Iyer: So, the narrative of a hard landing versus a soft landing
and all of those kinds of, um, conversations in my world are not really
relevant. What I’m trying to say here is that do not assume the Fed is in a
bipolar mentality here. It’s quite determined and measured in keeping interest
rates high and the fallout of high interest rates for longer, high inflation for
longer is a compounding effect on the economy, and so the Feds got a much more
longer horizon than the average human being as right now, and we’re gonna
understand the Fed doesn’t look at it from a day-to-day basis or a
month-to-month basis, but looks at it as a structural change in the way free
money in the market needs to be taken out. Then I think the action that the Fed
did is quite measured, in my opinion. I could be contrarian-

[00:05:19] Pierre Daillie: Yeah. I, I, I, I… I mean, my, my thoughts on it is
you, you really have to applaud the speed with which they reacted to this matter
and so let’s, let’s, let’s take a pause on that part of the conversation,
because I want to change gears just for a moment. Um, I’ve been dying to ask you
about your take on ChatGPT and generative AI in general. I mean, when, when that
happened, actually, you were one of the first people that I thought of, because
we’ve had so many conversations about, um, you know, surrounding artificial
intelligence. You must have been beside yourself when, when AI took center stage
in the tech race this year. What’s, what’s your take on it?

[00:06:05] Sri Iyer: Oh, it’s, it’s… In my opinion, it’s very real. For us,
we’re really excited in the sense. Obviously, ChatGPT came out in November, and
it took the world by storm, right? It’s, it’s, I mean, it’s, it’s the, it’s the,
what I call it, the poster child for artificial intelligence for all the wrong
reasons. I’ll put it that way, um, but it’s basically a LLM model, large
language model, and the biggest thing that ChatGPT did is to shake up the world
beyond the abstract sense of what AI is, is because suddenly the average person
is actually touching AI in a very, very cognitive way. So, the evolution of what
you call humans and bots merging, ChatGPT is the first instance of that, which
could be kind of overt versus being covert.

[00:07:01] Pierre Daillie: Right.

[00:07:02] Sri Iyer: So, even an average person, my, my, my dad, who’s 80 years
old, was asking ChatGPT a question, which is an 80-year-old when who was around
when there was no computers is now chatting to the bot and trying to get an
answer. So, you could tell how profound his response is. And so if you look at
it from a philosophical sense, the 1800s is all about mechanization, water and
steam. Then you saw the 1900s, where you saw mass production, electric power,
and turbines and assembly lines, and in 2000, you saw computers, automation,
electronics all emerge. 2000’s not that far away. Then it was like, 2010 you saw
Internet of Things, networking and machine learning, which is what we got into
in 2017, 2018, and now you’re looking at ChatGPT, regenerative AI, human bot
reaction, collaborative AI, cognitive AI, prescriptive AI.

So, it’s going at a pace that guys like us even can’t keep up with it. So, as
someone who implement artificial intelligence in investment decision making, um,
I think the emergence of regenerative AI, which is more prescriptive predictive
type of model that is using natural language processing, as a, as a core base,
um, is quite profound. We, we use NLP data for new sentiment analytics, and-

[00:08:28] Pierre Daillie: Right.

[00:08:28] Sri Iyer: … uh, gauging the tone of a news article and aggregating
the tone of the market. We do all of that, but regenerative AI is, is, is, is,
is kind of the next, next level of I would say, uh, inferencing, for lack of
words, and let me just kind of, um, simplify it in the sense that most machine
learning the guys like what we do. By the way, I don’t profess any knowledge in
the ChatGPT space. I’m just giving you a context of what data scientists like us
do. Most data-

[00:08:57] Pierre Daillie: But you do you understand… I mean, but Sri, you do
understand how artificial intelligence works.

[00:09:01] Sri Iyer: Oh, very much so. Very much so. So, if you look at what we
do with machine learning and AI, uh, deep learning and all aspects is a lot of
the thing is about processing data, curating data, what you call feature
engineering data, feature selection, um, and that’s where 90% Of all the work
happens. Once you train the models, and then you give it today’s information, it
can infer-

[00:09:26] Pierre Daillie: Right.

[00:09:27] Sri Iyer: … a solution from what we have trained it with. ChatGPT
have made this whole thing fully upside down. It’s crazy. Now, you go to see
society use ChatGPT to infer, while they speed up processes, the number of GPUs
being used and TPUs is being used. GPUs are graphical processing units. TPUs are
tensor processing units. The size that this has scaled out to now, you’re going
to see an upside down response in the world where 70 to 80% of AI is not going
to be used now’s going to be used for inferring, and only 30% is going to be
used for training.

[00:10:04] Pierre Daillie: Wow.

[00:10:05] Sri Iyer: So, that’s a monumental juxtapositioning in a matter of two
years. So, guys like us look like a carpet has been pulled out of our feet right
now to figure out where’s the trade-off between compiling and inferring a
training and inferring? So, the ratio of training data and inferring solutions
has been put upside down in the world of ChatGPT. So, it’s open source. You got
programming languages like PyTorch now that are coming out with a new generation
that teenagers can use now where it’s going to be a common language that you can
use for AI training across any processor, whether it’s the GPU, CPU, TPU.

So, you have got to have a common language of training data and inferring
knowledge. So, it’s been a little bit of a dramatic shift in the world of
Democratization of artificial intelligence. That innate feeling is what is
creating such a big buzz in ChatGPT right now.

[00:11:03] Pierre Daillie: Well, sure, I think I think a lot of people are
looking at ChatGPT, you know, especially in the content creation world, um, you
know, to, to make them better writers, provide you know, like, like, if you’re
writing an essay, you can ask ChatGPT for an outline for that essay. You know,
you can, you can… you know. It, it, it certainly, uh, helps as a creation, you
know, as a creativity tool as well, because it, it, you know, it has the ability
to go into areas of thought that you might not have thought about yourself if
you’re using it.

But I, I think, I think it’s interesting what you said though about how, you
know, people are using, you know, u… Content is being created and inferred using
ChatGPT and then it’s being published, but does that mean, and, and so, but I
think what you meant was that because I heard, I, I did hear some, um, rumblings
about this in, in interviews that I’ve watched about it, but I think one of the
points was that, was that if the future if, if in the, as things progress,
obviously, in the future ChatGPT or the AI behind Chat, the ChatGPT is training
itself on, on, on bad information. Then that bad information will trickle into
new information down the road, right?

[00:12:25] Sri Iyer: Absolutely. Training data is everything. If you, if you
look at the source data for ChatGPT, it’s the digital universe that’s been
around for 30 to 50 years now, and we can all make our own personal judgment as
to what the digital universe looks like. So, ChatGPT learns from the narrative
of digital data and digital data always has a narrative-

[00:12:47] Pierre Daillie: Yeah.

[00:12:47] Sri Iyer: … because news is always intertwined with opinions and
facts.

[00:12:51] Pierre Daillie: So, Sri, sorry, does that mean that, does that mean,
ultimately, that ChatGPT will wind up training itself?

[00:12:59] Sri Iyer: ChatGPT already trained itself.

[00:13:02] Pierre Daillie: [laughs]. Yeah.

[00:13:02] Sri Iyer: So, that’s what regenerative AI is, in the sense-

[00:13:05] Pierre Daillie: Yeah.

[00:13:05] Sri Iyer: … it is using its data-

[00:13:06] Pierre Daillie: Re… Yeah. Regenerative. That’s right. I mean, it’s
generative, which is, which is from scratch, but then regenerative is where that
stuff is being recycled.

[00:13:15] Sri Iyer: Absolutely.

[00:13:15] Pierre Daillie: Right?

[00:13:16] Sri Iyer: So, it, it is detecting false positives through a feedback
loop of humans. So, when it gives you a question, when you give it a question,
and it gives you an answer, and if you respond to ChatGPT that your enter is
biased. That’s when you get the true answer from ChatGPT, not the first answer.
The second answer from ChatGPT is much more accurate than the first answer
because it is regenerative. It, it, it… Once you question its bias, it will go
back and recheck to see it has bias or not, and if it has bias, it will come
back and respond to you saying that, yes, that could be bias, but so that gives
you that caveat, so that you as a human being can be intelligent enough to not
take it at face value, but use it as an opinion piece or information response
that you can then make your own opinion upon.

[00:14:03] Pierre Daillie: Interesting. That’s very, very interesting.

[00:14:06] Sri Iyer: Versus just, just reading the first thing it sends tells
you. Reading the first thing it tells you while it is quite accurate I don’t
think so involves a degree of intelligence. Probing ChatGPT to get a more
detailed response is where the real learning comes in and ChatGPT appreciates it
because your feedback loop is what it’s using to learn and make it better.

[00:14:27] Pierre Daillie: Right, but what… I mean, if you, if you just consider
the simple adage of garbage in, garbage out. Um. You gotta be, you gotta be
careful.

[00:14:38] Sri Iyer: Yes.

[00:14:38] Pierre Daillie: You do have to question the results you have, but the
fun… But that’s interesting that you can actually question the AI issue itself.

[00:14:45] Sri Iyer: You sure can. You sure can.

[00:14:46] Pierre Daillie: In it, in its… Right?

[00:14:47] Sri Iyer: Yes.

[00:14:48] Pierre Daillie: And so, so users are going to have to become
increasingly careful about where the information is coming from, and, and I
mean, it, it, it’s almost like… Well, I don’t want to say It’s… It’s in a way,
it’s human-like. If all, if all we’re caught in is a feedback loop, you know, of
bad information like, for example fake news. There are entire cohorts of
civilization that are, that are, you know, caught up in the fake news, thinking
that that’s reality or because they’re visiting the same sites every day.
They’re visiting the same, they’re visiting the same sources. They’re, you know,
they’re seeing it in their feed.

[00:15:28] Sri Iyer: Well, it’s on, on both sides, right? It’s both on the left
and the right and that’s where the bias kicks in, right? When, when the
narrative is only right or only left, that’s where the problem is and ChatGPT,
uh, right now will be biased on the narrative in the market per se, as to
whether much more about left is being written know much more of our right’s
being written. It doesn’t have that cognitive intelligence yet to say that I
want to stay unbiased unless you probe further, but I would say that we’re
missing the big point here.

This naive conversation about the bias that ChatGPT bring completely takes away
the revolutionary aspect of ChatGPT. Where ChatGPT really comes into handy or
comes… It’s going to be a dramatic changes is in software development. You’re
going to have software development where code is going to be written for
training data. That means semiconductors will be getting more and more powerful.
It’s going to be used where you’re going to have such massive big data because
of ChatGPT. It’s going to change data centers where storage is going to become a
massive, massive growth area in the market. Where, where’re you going to store
all this data? It’s got… It… When people say it’s in the cloud. It’s not really
in the cloud. It’s in hardware, sitting somewhere.

cybersecurity, right? ChatGPT can now generate lifelike emails with content to
do cyber crimes and cyber attacks. It can write codes and viruses codes that you
can implant somebody and it has all that because virus code is available on the
World Wide Web. So, it can learn from that and build your code to destroy
something.

[00:17:09] Pierre Daillie: Right.

[00:17:09] Sri Iyer: You know, you’re looking at, basically, search engines in
the future. Um, right now, we are doing some basics like Google searches, but
tomorrow, it’s going to replicate the Google where you’re going to have a chat
bot. So, you’re not going to really type anything. You’re going to chat with
ChatGPT or Google/Bing, or BERT or anything like that. You’re just going to have
a conversation and ask a question, and it’s going to give you an answer. Not
only, not in fact in typing. It might even give you a voice response to, to…
Boy, Space Odyssey 2000 comes to mind.

[00:17:42] Pierre Daillie: [laughs]. Yeah.

[00:17:44] Sri Iyer: It’s fiction becoming reality in front of our eyes in our
conversation. Another area’s communication, your space, right? Uh, writing. It
can write a full textbook on any topic now.

[00:17:54] Pierre Daillie: Right.

[00:17:54] Sri Iyer: Um, media, content generation, news articles, social media
posts, um, marketing, story writing, script writing of movies. All of this can
be done by ChatGPT in a very precise, concise, closed form solution. If you as a
domain expert can give it some basic parameters and tell it exactly what you
need. Then you’re looking at music, right? You’re looking at generate melodies,
critique music. It could become a critic. It can even write music if you tell it
that these are the topics or these are the kinds of parameters. The legal, big
highly disruptive ChatGPT, you’re gonna see it can write legal documents. It
could write dissertations. It could even pass board exams for legal exams. So,
you’re looking at a disruptor response there and I’m not stopping there.

[00:18:44] Pierre Daillie: Wow.

[00:18:44] Sri Iyer: Look at healthcare.

[00:18:45] Pierre Daillie: Yeah.

[00:18:46] Sri Iyer: Healthcare, telemedicine, you’re talking to a doctor.
Tomorrow you’re going to be talking to a bot, giving you prescription
medication. You’re going to look at it as a remote monitoring system for a
doctor. The doctor will know through the chat bot is if this guy taking his
medicine or not and it’ll give a prescriptive response to the patient.
Pharmaceuticals, drug screening, pre-drug, pre-discovery, drug discovery,
process analysis, probability of success in drug discovery. All that is going to
go through these kinds of frameworks.

You saw what happened to banks, right? Predictive Analysis, credit default,
consumer default. All these things could be used in a more regenerative sense on
a predictive basis based on what’s being written or what could be solved and
aerospace engineering, fuel optimization. Unfortunately, defense contractors
could use it for very bad things, ChatGPT.

[00:19:37] Pierre Daillie: Yeah.

[00:19:37] Sri Iyer: So, you’re looking at much, much more that ChatGPT or
regenerative AI can impact than just the common very myopic response of, uh,
politics or left, right opinion or anything like that. That’s, in my opinion, a
very, very small component of the ecosystem that ChatGPT exposes human
civilization to. Again, this is my opinion.

[00:20:02] Pierre Daillie: Amazing, amazing, truly amazing. I, I, I… Thanks for
sharing that. I hadn’t even… Uh. You know, I haven’t even gone that far in, in
my, my own reading, as far as, you know, all the, that, that laundry list of,
of, um, both, I think both blessings and, and, and dangers. Um, you know, you
can see where we’re we could, we could use or our children could use ChatGPT to
stand on the shoulders of giants, but what do they do with it once they’re
there? And what, what what’s, what, what are the, what are the ranges of
outcomes that are possible, um, from that? Because you know it’s going to be
both used for a very productive means and on the other side, it’s also going to
be used for, for probably destructive means as well.

[00:20:59] Sri Iyer: Like anything else in society-

[00:21:01] Pierre Daillie: Yeah.

[00:21:01] Sri Iyer: … over a thousand years. So, as humans, we can’t question
the future nor should we live in the past, nor should we live in the moment. We
just have to accept inevitable when it happens. If you do that, then you will
not see confrontation, but if you always are in all, all defense up in front of
you, and you always see everything as a threat, then you’re going to confront
things. Rather than confront things, if you accept the inevitable, I think
society will evolve much more. You’re going to have some bad things, but you’re
going to have much, much more good things come out and bad things, but there’s
never been a situation that any evolution in civilization always resulted only
in good.

There’s always been bad with good, but that’s, that’s nature and that’s the way
evolution works. So, I would say accepting the inevitability in the world of
artificial intelligence should be the baseline norm that will help civilizations
and people evolve in a better way. In our world, in our investment industry,
artificial Intelligence is still nascent, but guys like us-

[00:22:13] Pierre Daillie: Not, not, not, no… [laughs]. You know what? Now you
know why I was dying to ask you. I was dying to ask, like specifically, I was
dying to ask you about, about ChatGPT about AI because, because-

[00:22:26] Sri Iyer: I think AI’s very overrating. It’s very liberating for me
and my team especially. It’s because investment decisions are now being made
with a lot more objectivity and a lot more sense of inevitability rather than
just prognosticating the future and having a lot of error. So, the acceptance
rate of decision making exponentially grows if you accept technology and the
evolution of what’s going on in the market right now.

[00:22:52] Pierre Daillie: Before we get to talking about investing, uh, I think
it’s only fitting to introduce you to those listening who don’t know you. So,
bear with me.

[00:23:04] Sri Iyer: No problem.

[00:23:06] Pierre Daillie: Sri joined. Now you’ll understand why I wanted to
talk to Sri about AI. Sri joined Guardian Capital LP in 2001. As well over 25
years of experience managing quantitative investments in risk management, he was
instrumental in the development and implementation of Guardian Capital’s
proprietary systematic strategies, which subsequently led to the creation of the
i3 Investments team. Sri’s expertise in guiding the overall development and
implementation of systematic strategies for the firm is unparalleled. Sri and
his team manage multi-billion dollar investment portfolios using a combination
of AI based predictive models to make investment decisions.

Sri has through his leadership pioneered the application of machine learning and
deep learning models at Guardian Capital LP since 2018 and continues to research
and innovate in the space of feature engineering, selection, while managing an
all star team of data scientists. Sri, in my opinion, is one of the most
experienced investment professionals in systematic investment strategy in the
business. If Sri wasn’t in investment management, I suspect in an alternate
reality, he would have his own tech startup. So, [laughs], so Sri, let’s, let’s
set the table for our conversation about, uh, investing, um, because your, your
area of specialty is in the, uh, is in dividend investing. Um, how can investing
in… By the way, was that, was that a fitting introduction?

[00:24:48] Sri Iyer: That was a… It’s a humbling introduction, I would say. It’s
funny you mentioned dividends in the sense. Obviously, we don’t do only
dividends. We do growth dividends we recently launched a hedge fund to, but
you’re right. The relevancy of how you present, it was amazing. I mean, it’s
like my whole career of 35 years just flashed in front of my eyes very quickly.
So, thank you. It was-

[00:25:15] Pierre Daillie: Well, you’re welcome, Sri. I think one of the things
that that we’ve, that I love about you is that you’re, you’re very, you’re, you
are humble. You’re modest. You’re humble, especially given the things that
you’re doing. Um.

[00:25:30] Sri Iyer: Totally. I manage a team of all-star data scientists. They
humble me every day.

[00:25:34] Pierre Daillie: Well, and you’re so-

[00:25:35] Sri Iyer: [inaudible 00:25:37] in the team I can imagine.

[00:25:37] Pierre Daillie: Yeah, you’re so quick to, you’re always you’re always
very quick to, to call it a team effort, which is also admir- admirable, so, um,
I’m sure your team loves it as well. Um, so, let, let, let’s get to the
discussion about investing because I think, I think in the context of what we’ve
talked about at the very beginning about SVB, about, you know, the… some of the
changes that are taking place in the economy, uh, whether or not, um, you know,
whether or not rate, you know, the rate hiking regime is going to continue or
pause or well, and we’ll get to that. How can investing, like it’s a turbulent
period. I mean, this past year has been turbulent for investors. Um, how can
investing in dividend paying stocks help investors defend their portfolios in a
turbulent market?

[00:26:33] Sri Iyer: I’ll keep it brief some of these answers because the
practicality of your questions are very profound here. So, to answer your
question, how do dividend paying stock help investors in turbulent markets, what
comes to mind is low volatility, downside capture, cashflow visibility, and most
important, opportunity to improve yield at cost. These are some basic responses
that could be tagged and achieved by dividend investing during turbulent
markets. So, clearly, when markets are undecided in their direction or the
magnitude of the variance of the direction and path is violated, people look for
some kind of downside protection, some kind of a lighthouse, as to how to moor
their boats do, and I think dividend investing is that lighthouse.

Now, one of the biggest things that people do not perceive is the opportunity to
improve yield at cost. That is, if you’re buying a security that pays a coupon
at X price, and over the years, you have held that security, then whatever the
coupon is, today, whatever the payout on that stock is today, if you divide it
by the original price of the stock, your coupon is significantly higher than
taking what it’s paying today by the price it’s today. So, when you see a
turbulent market, and you see weakness in the stock market where the baby gets
thrown out with the bathwater, it is a great opportunity to buy dividend-based
portfolios, because you can lower your cost basis and increase your yield at
cost. That is true wealth building for the long term.

[00:28:24] Pierre Daillie: Amazing. That’s a great point. I think, I think, you
know, we saw, we saw on Monday of this week, and today’s March 14. So, we’re
talking about March 13, ’23. We saw the regional bank stocks in the U.S. take
basically take it, you know, get a kicking because of the, the uncertainty
surrounding the Silicon Valley Bank meltdown and, um, maybe it’s not a meltdown,
but collapse. And so it’s the, it’s the old cockroach analogy, right?

[00:29:05] Sri Iyer: Mm-hmm.

[00:29:06] Pierre Daillie: There’s never just, there’s never just one, right?
And, and so, so some people have pointed to Janet Yellen basically saying that,
that we’re going to do whatever it takes to guarantee all depositors their,
their deposits. So, suddenly the $250,000 FDIC limit went out the window and now
what’s, what’s been implied through that statement is that 100% of deposits will
be guaranteed by the, by the, you know, at the U.S. government, whether it’s the
Fed or the treasury, I’m not sure. The Treasury, I guess.

Um, so, uh, you can see why… I mean, I can see why like in a in a turbulent
period, just to circle back to your answer, investors would immediately start
looking for companies that have, you know, wider moats, companies that have
steady or growing dividends and they wouldn’t, they would also, at the same time
look for companies whose moats have shrunk and say, "This company might cut its
dividend," right? But, and that those are qualitative measures of businesses,
but so it’s very interesting. The same way depositors were feared that they were
they would pull their money out of lower tier banks, and, and put them into tier
one banks yesterday. I don’t know if the… You know?

There’s, you know, there’s been throughout the day, there’s been a string of
interviews with, with regional bank CEOs, um, you know, and they’re not all
suffering, you know, runs on, on their, on their deposits, but you can see that
same mentality happens in the market where where investors, um, default back to
the highest quality companies that they can… I mean, depend on how you define
quality, of course, but dividend paying stocks obviously have always held a very
dear position in the market. So, why is it important? I mean, I don’t know if I
asked. I’m answering the question that I’m about to ask with, with my chatter,
but why is it important to focus on finding consistent companies with proven
track records of growing their dividends year over year? And did I, did I… I
think I might have just answered the question I just did.

[00:31:34] Sri Iyer: You did, but let me give you a slightly different context
to that in the sense, what dividend growth does is it signal secure viability of
a company’s business. That’s all it does. It gives you the secular viability
tag. Consistent cash flow personifies a consistent company. So, when somebody
says management is consistent or a company’s consistent, it has to be reflected
in the cash flow. It allows for a clean evaluation. When I’m valuing companies
to own through our big data DCF models, if I have secular cash flow and clean
cash flow, I can value the company better so it leaves less error for me. It
also allows for better market imbalance detection. That is if I can value a
company on secure cash flow, when the market overreacts or underreacts, I could
then analyze the true valuation of the stock.

So, when you when you look at it in that perspective, um, one has to be very,
very careful in not throwing the baby out with the bathwater is what’s happening
is if you look at what happened at SVB in the market and why I have a contrarian
response to actually supporting the Fed’s intervention is because you have the
asset side and the liability side. The deposits are now getting have to be paid
much higher at about what? Four, four and a half percent on the short term
deposit side.

[00:33:01] Pierre Daillie: Right.

[00:33:02] Sri Iyer: And in those days, you are paying, you’re getting paid 25
pips or 50 pips. So, that’s been a dem- demonstrative jump in the liability
asset mismatch right now. And so what happened here is to create net interest
margin and to be viable as a, as a profitable business, most of these banks went
on the asset side into very long duration assets, and when I say long duration
assets, you can go from 10-year treasuries all the way to some kind of private
equity-

[00:33:32] Pierre Daillie: Right.

[00:33:33] Sri Iyer: … agreement on a lending agreement. Now, what’s happening
here is in the last six months to nine months, with the short end of the yield
curve, and the rates starting to go from 25 pips to a target rate of around 5%.
A lot of these 10-year duration assets are now under the water, but my question
is, just because a 10-year bond is under the water doesn’t make it a bad bond.
It doesn’t-

[00:33:59] Pierre Daillie: Yeah.

[00:33:59] Sri Iyer: … because that yield to maturity, you’re gonna get your
money back.

[00:34:02] Pierre Daillie: Right.

[00:34:03] Sri Iyer: So, what’s happening here is the Fed intervention was not
to secure bad assets. Fed’s intervention was actually to secure good assets,
which are underwater. So, this is almost like a REPO in the sense that the Fed
is saying that we’re going to guarantee customer deposits so that the customers
don’t make a run on the bank because if we take the money out of the bank, the
banks will be forced to sell 10-year bonds when they don’t want to be selling
it.

[00:34:36] Pierre Daillie: Yeah. Yeah. So, I mean, that was… That, but that’s
good, right?

[00:34:39] Sri Iyer: That is very, very good.

[00:34:40] Pierre Daillie: It’s a great plan, because it, it… By, by protecting
depositors, you’re protecting everybody.

[00:34:45] Sri Iyer: You’re protecting the duration of an asset by protecting
short-term depositors from running, making around in the bank. So, this is not a
bailout.

[00:34:55] Pierre Daillie: Yeah, so a bond-

[00:34:56] Sri Iyer: This is a measure of intervention for protecting the
duration of an asset class. Now, if you went and gave your money to some private
equity, which is not going to do any, make any cash flow for the next 30 years,
that’s a risky asset that should never have been on the balance sheet beyond a
certain percentage was a deposit. That kind of risk should not be bailed out,
but the problem here was happening was that people were making a run on mid cap,
small cap banks, and a lot of the assets are sitting in long duration, fixed
income bonds, and that could create a liquidity problem in the market, and to
avoid that, the Fed guaranteeing deposits, in my opinion, is a very, very good
thing. It’s not a bailout. It’s, it’s a, it’s stopping irrational behavior.

[00:35:41] Pierre Daillie: I think, I think they’ve actually they’ve actually
avoided using the word bailout, haven’t they? I mean-

[00:35:45] Sri Iyer: Correct. Well-

[00:35:46] Pierre Daillie: It’s, it’s more of a stop gap or… Yeah.

[00:35:50] Sri Iyer: Well, banks that have massive liability mismatches will
fail.

[00:35:55] Pierre Daillie: Yeah.

[00:35:55] Sri Iyer: But not at the cost of clients making a run out of the
banks, but because eventually the mismatches will play out in the earnings and
the liability. So, let bad banks fail, but not let all good banks fail. That’s
the main thing and the separation of mega cap banks to mid cap, there’re a lot
of good banks in the mid cap, small cap space too. Unfortunately, this contagion
had moved further down the pipe. You can see a lot of small good banks also get
destroyed and that’s extremely unhealthy for a economy like the United States,
which has thousands of banks, not like Canada, which has-

[00:36:32] Pierre Daillie: Well, I said… That’s one of the… Uh. That’s really
one of the huge differences between, between us and them, right?

[00:36:40] Sri Iyer: Correct.

[00:36:41] Pierre Daillie: Is the, is the regional, you know, all of the
regional and smaller banks that they have.

[00:36:45] Sri Iyer: The lifeblood, the lifeblood of micro financing into small
businesses is done by regional banks, not by large banks.

[00:36:54] Pierre Daillie: Interesting. So, just to put things, just put things
back into perspective, um, what, just so that investors… I think you touched on
it at the beginning, at the beginning of our conversation about dividends with
regards to the coupon, and the coupon from dividend stocks, and correct, correct
me if I’m wrong, but what share of total return of the S&P 500 is attributable
to dividends over the last 20 years?

[00:37:30] Sri Iyer: You have said 30 years since 1970s.

[00:37:33] Pierre Daillie: Yeah.

[00:37:33] Sri Iyer: Between dividends, buybacks, which is money being taken
back, and debt reduction, or what we call dividend yield plus buyback yield plus
debt reduction, which is called shareholder yield. That is the shareholder gets
his yield and his cash back. This represents about two-thirds or slightly more
than two-thirds of the compounding of equity returns are since the 1970s.
Two-thirds.

[00:37:58] Pierre Daillie: Yeah.

[00:37:58] Sri Iyer: So, when you’re capturing dividends, which is a very large
part, um, almost 70 to 80% of that two-thirds with some very strong buybacks and
very little debt on balance sheets, you’re able to capture a very significant
part of the very reason why you should be owning equities in the first place.

[00:38:20] Pierre Daillie: Yeah, and that seems to be really an under,
underappreciated aspect of, of investing in stocks is, is the dividends, but I
think, I think… I want to, I don’t know, I think I wanted to make a quick point,
because we went straight to dividends and I know that’s not all you do. But, but
it really, it’s kind of like synonymous with our conversation about how the
treasury has promised depositors that their, their deposits are safe. Um, that
looks after everything that comes after it. Everything that’s related to those
deposits, such as the duration of the bonds in the held to maturity portfolio or
the available for sale portfolios in the banks.

[00:39:11] Sri Iyer: Correct.

[00:39:12] Pierre Daillie: But I’m not quite exactly that way. Maybe it’s I’m
making a terrible connection there, but, but the, the connection I see with,
with dividends is that if you, if you, if you focus on dividend investing as a
core equity strategy, um, by focusing on the quality of dividends and dividend
growers, companies with, with a long dividend history and a history of growing
their dividend, aren’t you in effect taking care of some of the more important
aspects of building a quality portfolio of equities?

[00:39:51] Sri Iyer: Absolutely. So-

[00:39:52] Pierre Daillie: Yeah.

[00:39:52] Sri Iyer: … if we can go to finance 101-

[00:39:55] Pierre Daillie: As a core principle.

[00:39:56] Sri Iyer: Absolutely. If you look at CAPM, capital asset pricing
model, look at what the Fed did here in guaranteeing short term deposits. When
you start with capital asset pricing model, you start with money market or
money, a three month. A three-month T bill or money market is the most sacred
instrument you can own. If you start making a run on banks and the mid cap bank,
you cannot even be comfortable owning a three-month T bill at a local bank, then
anything beyond that is suspect for you.

So, what’s happened as interest rates started to rise is, in general, people
have dramatically reduced their duration, gotten out of fixed income, gotten out
of Bitcoin and in general, even equities, and are sitting with copious amounts
of cash in the asset allocation right now.

[00:40:50] Pierre Daillie: Right.

[00:40:50] Sri Iyer: And on that, you’re giving headlines that people are making
a run on banks. Can you imagine the spike in implied risk premium, in short-term
bonds if you start making run on bank, bank deposits? That is extremely risky,
because the short end of the CAPM line has to have an implied safety mechanism
always built into it. Otherwise, it has a dramatic cascading response to risk
further up the CAPM curve. So, this approach by the Fed actually kind of stopped
a tsunami, a credit risk mounting up the, the, the CAPM curve in a very
aggressive way.

Um, the correction in, in fixed income last year would look like a picnic if we
start seeing long duration, duration assets start to default. You’re seeing it
in private equity right now. There’s no bid for exits. So, there’s a lot of, um,
uh, underwater private equity exits now sitting on big firms and big
conglomerates right now that cannot get out and so… and they don’t want to
market these products.

[00:41:57] Pierre Daillie: Right. Right.

[00:41:58] Sri Iyer: It gets marked to market only when you go to market to
exit. So, these unrealized losses sitting on balance sheets right now, uh, could
have a devastating effect on capitalism in general and it all starts with the
short end of the CAPM line and so what the Fed did here is to make sure that the
cash is protected. The problem with cash at 4% is when inflation is at 6%, you
still have a negative rate of return, and I’m trying my best and our job as
investment managers is to make sure that people should not get afraid of
duration, but embrace duration the right way and so jumping from cash to
dividends, which is still equities, not a bond, gives you a very good conduit to
capture some duration visibility on McDonald’s, or Johnson Johnson or an
AstraZeneca, um, gives you that long visibility of secular cash flow with a
three to three and a half percent coupon, and that’s as good as it gets in this
market when you’re afraid that you’re not getting your deposit back tomorrow
from a bank local around your corner.

So, there is a true value added to make sure that the short end of the CAPM is
protected at any cost and give people a viable response to move away from cash
because at 6% persistent high inflation, um, it’s actually giving you a negative
rate of return. So, you do need some duration debts to get appreciation along
with a payout and dividends play a very, very important role in that transition
between safe deposits and risky duration. Dividends play the mid space extremely
well in this market. I cannot explain it any further than this. It’s as-

[00:43:49] Pierre Daillie: You did a great job. You, you did a terrific job
there. Um, so, uh, does, does systemic risk does or does the potential for
systemic risk in the financial system make dividend payers and growers more
attractive?

[00:44:05] Sri Iyer: I, I will only answer it in one word. Yes.

[00:44:11] Pierre Daillie: Okay, great. So, um, how does inflation, assuming
that it remains sticky for the foreseeable future, how does it impact fixed
income and equity portfolios? And what is the difference in how they’re
affected?

[00:44:29] Sri Iyer: I think, again, these are my opinions in my analysis of
data. I think consistent inflation is significantly worse than inflation and
people do not differentiate between inflation and persistent inflation. If I
have to ask an economist a question, I’ll ask him, "Please define inflation to
me as being persistent or not. If inflation is not persistent, I actually like
it," because it allows us to… It acts as a transfer mechanism from the company
to the user and you could still make money if inflation is not persistent. The
problem with persistent inflation is it eats into real income.

[00:45:12] Pierre Daillie: Right.

[00:45:13] Sri Iyer: And the compounding negative aspects of high inflation is
never really perceived by the market or by the individual. So, when you’re not
perceiving the compounding of persistent inflation, um, you start seeing risk.
Show me inflation, and I’ll show you rising rates. Show me rising rates and I’ll
show you credit risk. So, that’s how I see inflation, if it stayed sticky,
impacting the market. So, just credit risk, perception that’s it volleying in
the market right now and ended with the Fed intervening to a degree is all a
logical culmination or a cumulative response to persistent inflation.

[00:46:03] Pierre Daillie: Yeah, that’s the key. I think, I think the what I got
from, from your response is that it’s persistent inflation that the Fed is
hell-bent on tackling.

[00:46:16] Sri Iyer: Correct.

[00:46:17] Pierre Daillie: It’s not inflationary spikes or-

[00:46:20] Sri Iyer: No.

[00:46:21] Pierre Daillie: Events. Interesting, very interesting. I, I… You
know, it’s the sticky components that they’re, that they’re really focused on,
right?

[00:46:32] Sri Iyer: There were… At the peak of COVID, in March, March 2020,
there were accumulative 1,000 rate cuts around the world since subprime and we
expect all that to go away in a matter of two years.

[00:46:49] Pierre Daillie: [laughs].

[00:46:49] Sri Iyer: There’s not enough liquidity in the market that is driving
prices up, does enough. Lack of labor force participation that is keeping wages
high and unemployment low. These are all structural issues that just don’t turn
on a dime after giving 1,000 rate cuts for 10 years. It takes time to bleed
excess out of the system and what the Fed is trying to do is to make sure it’s
done on an orderly basis.

[00:47:19] Pierre Daillie: I think, I think they have established that. I think,
I think, you know, because investors at large are not very, not, not pleased,
obviously, with, with the outcome of Fed intervention, you know, that, that
that’s why you have this. You know, on one side, you have folks who are
accepting of what the Fed is doing and understand it like you, uh, and on the
other side is, is the other, uh, other cohorts of, of opinion on the, on Fed
policy that that are in disagreement with it.

[00:47:58] Sri Iyer: Well, if I may phrase it, I rather be wrong and save my
money than right and lose all my money.

[00:48:07] Pierre Daillie: Did you find it interesting before… I find it
interesting that we went from there is no alternative to what almost feels like
the opposite and not that there… I mean, now, there’s lots of alternatives. Of
course, I don’t think that… I don’t think that that we have to stop investing in
equities and start strictly investing in short duration, but um, in a, in a, in
a strange way, like this, you know, in… You know, people for example pulling,
and let’s, let’s let’s, let’s, let’s say this X, this SVB thing that just
happened. But when people know that they could get, you know, 4.5 to 5% on a
treasury bill versus a bank account, um, then, then did the Fed not… has the Fed
or the bond market not put investors at odds with deposit-taking institutions
even though-

[00:49:25] Sri Iyer: I agree with you to a degree, yes, but if I may see the
positive in this, perhaps you’re going to see the bank’s work much harder in
raising their interest on their deposits to entice clients to stay around and
perhaps-

[00:49:43] Pierre Daillie: That’s the reason, right?

[00:49:44] Sri Iyer: Perhaps not buy the T bill. So, this is gonna-

[00:49:45] Pierre Daillie: Yeah.

[00:49:46] Sri Iyer: … make banks in my opinion more honest going forward to the
extent that if you’re going to sit with a bank and get a, a, a deposit rate
higher than the T bills, and by the way, T bills is a sophisticated context. The
average John Doe person doesn’t know what a T bill is. So, while we-

[00:50:07] Pierre Daillie: You’re absolutely right.

[00:50:08] Sri Iyer: While we’re gonna have a nice conversation about T bills,
the average John Doe person doesn’t even know what that is. They’re looking at a
bank deposit and income on a bank account including myself and my bank.

[00:50:18] Pierre Daillie: Let alone a, let alone a money market fund, right?
So-

[00:50:20] Sri Iyer: Exactly. I’m not buying… With my free cash, I’m not buying
cables. I’m still owning my bank’s savings account. I’m hoping they bought the T
bill and not something crazy, but they-

[00:50:31] Pierre Daillie: Well, there’s a whole degree, there’s a whole degree
of separation involved in that. I mean-

[00:50:35] Sri Iyer: Exactly.

[00:50:35] Pierre Daillie: … accessibility being one of them.

[00:50:37] Sri Iyer: Exactly. So, banks will compete in raising their deposit
rates to attract high quality assets. It’s as simple as that.

[00:50:46] Pierre Daillie: Yeah.

[00:50:46] Sri Iyer: And the larger banks will be able to do it better than
smaller banks. That’s also true.

[00:50:52] Pierre Daillie: That reduces their net interest margin.

[00:50:54] Sri Iyer: Correct. It does, which means earnings for banks could get
a little bit volatile and our AI systems have been showing for last six months
that bank earnings don’t look good.

[00:51:07] Pierre Daillie: Well, we just saw, we’re starting to see some
reporting from banks now. Right? Of, of decline in earnings.

[00:51:16] Sri Iyer: Canadian banks in general look significantly more stable
because we have a large repository of global banks around the world. Uh,
relatively speaking, Canadian banks are much, much, much more stable, but when
you see some of the predictions from AI, with respect to a one-year fraud
earnings estimates or even dividend risk, uh, probabilities, you could clearly
see a spread between, say, mid cap, small cap banks and large cap banks around
the world clearly.

[00:51:45] Pierre Daillie: Interesting. So, sure, yeah. How does, how does the
recession affect the demand for near term predictable and attractively priced
cash flows and why does this make dividend payers valuable?

[00:51:59] Sri Iyer: Well, I would say recession amid persistent high inflation
is deemed as stagflation. The mistake the market is making is I think it feels
the Fed can engineer a soft landing. However, the principal components of
inflation are what? Supply chain driven problems, labor costs, persistent high
labor costs, because of lack of participation in the labor force. Remember, I
can’t get a 25-year-old to come to work and I’m trying my hardest to keep a
60-year-old from leaving work and retiring. That’s my job today.

[00:52:35] Pierre Daillie: [laughs]. Yeah.

[00:52:36] Sri Iyer: So, I got both details leaving or not being there. So, I’m
the sandwich generation at 53 years old trying to keep the high talent alive.
That means more labor costs and persistent labor costs. So, when you look in at
the global geopolitical problems with the Ukrainian war, and everything else,
and now, the China aspect looming like a big, big Albatross right now, which
leads to supply chain disruptions, labor cost driven. These are all very
structural in nature. A mere rate increase here and there will not be enough to
suffice a change. So, it’s going to take some time for structural issues to sort
themselves out.

In the meantime, you’re going to see volatility in the market, and you’re going
to see turbulent periods and so you have to be very careful right now, where
beta and alpha should be separated, where it’s simply buying the market means
you’re buying volatility. When you buy active management, or you’re buying
responses to the duration and credit through active management, you probably are
going to be far better off than just going passive right now. So, these are
changes that the average person might feel, but does not comprehend yet and
that’s how I see the recession if it does come around, which I think is in our
opinion, Europe is already in a recession from an earnings standpoint and when
it comes to United States, the earnings cycle is starting to slow down, are
starting to decelerate at a reasonable pace. So, when you put all that together,
I think my thesis might play out.

[00:54:13] Pierre Daillie: Interesting, I think, you know, some, some in the,
some in markets, some people in markets are of the sort of simpler opinion that
a recession would lead to lower interest rates and therefore, you know, that
would benefit higher duration investments-

[00:54:32] Sri Iyer: To a degree.

[00:54:34] Pierre Daillie: Yeah.

[00:54:34] Sri Iyer: Mathematically, yes. Structurally, no.

[00:54:39] Pierre Daillie: Well, that’s, that’s the short sightedness.

[00:54:41] Sri Iyer: Right.

[00:54:41] Pierre Daillie: Is that… Yeah. So, we’re on the subject of events
and, you know, systemic events or systemic… Sorry, not systemic events, but
turbulence. A little bit of chaos this week. When these kinds of disruptions
happen, how do you, how do you and like, how do you instruct or how do you teach
your AI to recognize these types of situations? Um, and I guess the, maybe the
relevant question is, do you want to?

[00:55:21] Sri Iyer: Well-

[00:55:21] Pierre Daillie: Or to what degree, to what degree do you want to-

[00:55:24] Sri Iyer: Good question.

[00:55:25] Pierre Daillie: … do you want to teach your AI to recognize something
like SVB or, or a rise in, in systemic problems? Is it possible?

[00:55:36] Sri Iyer: It’s possible. So, if I give you an anecdotal response, um,
our, our dividend strategy has added a track record since inception of having
zero dividend cuts, and we’ve been running it for 15 years. So, when… Uh, so we
had no dividend cuts during subprime because doing subprime, most dividend
managers had about 40 to 50% in banks. Our systems allowed us to exit bank so
we’re a non-bank dividend strategy doing subprime, if you could wrap your head
around that. During COVID-

[00:56:08] Pierre Daillie: Impressive.

[00:56:09] Sri Iyer: During COVID, we had zero dividend cuts. So, people asked
the same question here. Is your model learning from COVID? The answer is no,
because COVID was not predicted. You have to train the model on COVID data. So,
today, has the model learned from COVID data? The answer is yes. We’ve trained
the model on the disintermediated responses of the market to the COVID crisis,
but during COVID it learned from the subprime response. So, the source of the
problem cannot be predicted, but their reaction to the problem is much more
predictable and consistent across multiple regimes.

So, regime detection, in my opinion as a data scientist, is very hard to do in a
very explicit way, but implied shifts are much more manageable. So, what we do
is we feed macro data into the model in the form of features. So, all the lead
economic indicators, we feed into the market. What happens is the stocks tend to
react and interact with these features to develop patterns of behavior. So, a
stock that are highly sensitive to rising rates, some are very less sensitive to
rising rate, some are very high sensitive to rising oil prices, some are less.
Some are very highly sensitive to manufacturing water PMI cycles, some are not.
Some are high, highly sensitive to the steepening of the yield curve while some
are not. These kinds of things have to be thought or trained into the model.

And once they are trained in the developed patterns and behavior recognition,
then when you feed it today’s data, whether it’s SVB, or whatever, the model can
use that training data to do what? Infer.

[00:58:02] Pierre Daillie: Infer. [laughs]. Thank you.

[00:58:05] Sri Iyer: So, that’s what we are building, um, in our model context
is to allow the model to infer and it can infer from the past to predict the
future with a degree of certainty and we use probability estimates. In some
cases, we use random forests techniques where we’re using the average of 2,000
or 20 founder predictions, because we do not rely on one prediction. We’re doing
2,000 predictions and taking the average. So, there are different techniques, a
supervised machine learning for certain features and certain parameters for
decision making, and there are certain very advanced deep learning type
responses we use for a market neutral hedge fund where we actually labeling
stocks as when to short and when to go long.

So, there are a lot of complex variables, where we start with machine learning
all the way to deep learning and now you got to regenerative AI, which I have
not really ventured into just yet, but perhaps someday my all-star data science
team might come back to me and saying, "Sri, you’re an old hag at this. Let me
get to take a crack at it," and I’ll be like, "Go for it."

[00:59:09] Pierre Daillie: So, you’re not yet at stage, at the stage where you
would consider adding regenerative AI to your processes yet?

[00:59:17] Sri Iyer: No, we’re not.

[00:59:18] Pierre Daillie: Yeah, it’s, it’s, it’s, it’s, it’s a whole other
ballgame. Right?

[00:59:22] Sri Iyer: You, you… It’s hard. If I, if I use ChatGPT as a
subscription, enterprise subscription, and use it to make decisions it, to a
degree, it’s a black box, and in the world of investment management, where we
have fiduciary responsibilities, I rather build a glass box than a black box and
so us being data scientists, our feature engineering, feature selection, we have
an innate understanding of how our artificial intelligence frameworks work. So,
as we do our academic research, and our readings and theory as to how
regenerative AI works, uh, we could probably start to learn from it.

One of the things that we do very unique, which is kind of similar to regenerate
AI in the sense regenerative AI uses transformers, um, in its deep learning
framework that it uses ridiculously multiple layers of neural nets, um, and any
kind of data coming out of the neural net is very abstract in nature and
regenerate AI can create recognitions of vectors and clusters of words and all
that stuff. So, it’s very abstract in nature. We are now actually feeding the
output of one AI model into a another AI model. So, one AI model of learning
from the predictions of another model.

So, things have progressed quite significantly in the world of artificial
intelligence. Um, it’s somewhat proprietary, but we’re having a chat,
conversation. Hopefully people can get some knowledge out of these things, but
that’s what, um, transformers and abstract learning is all about. It’s how the
human brain reacts to certain relationships between abstract datasets, and we’re
doing that right now, but as far as ChatGPT goes, the size of data… You know how
many GPUs we use? We use about three to four GPUs. You know how many ChatGPT
uses?

[01:01:11] Pierre Daillie: How many?

[01:01:12] Sri Iyer: Over 100,000 per query.

[01:01:14] Pierre Daillie: [laughs]. That’s mind blowing.

[01:01:18] Sri Iyer: It is.

[01:01:20] Pierre Daillie: I, I can see where where you don’t want to introduce,
uh, what’s potentially, a, a… I mean, maybe I’m not using the right term, but
it’s a wildcard. It’s too many wildcards into your process by, by if you were to
try to do something like introducing regenerative AI, but thank God you’re not
because it’s, it’s, it’s a… It, it… You know, your, your, your situ… your AI is
by design.

[01:01:45] Sri Iyer: Yeah and that’s a large language model. So, a new sentiment
data specialist might use that to create better perspective or projected
sentiment for a stock. So, not only are they telling you what the sentiment is
today based on news media. They might be able to tell you that we could
regenerate the sentiment that is going to happen tomorrow or the next minute
based on it. That could happen in the future and that could be a very positive
feature set for us, because we currently use new sentiment data as a momentum
feature set because we don’t read anything. The machine reads for all of us. So,
can sentiment become more intelligent with ChatGPT and regenerative AI? My
opinion is yes, but that’s not our domain. We actually subscribe to that data.
So, hopefully the data scientists are doing that for us who work on that.

[01:02:31] Pierre Daillie: [laughs]. Speaking of sentiment, um, what are your
expectations for market volatility or equity markets for 2023?

[01:02:40] Sri Iyer: Um.

[01:02:40] Pierre Daillie: How do you, how do you see things unfolding for the
balance of the year and, and, uh, and even ahead?

[01:02:46] Sri Iyer: I have mentioned in the recent past that the only macro
risk I worry about is liquidity and credit risk. If inflation stays sticky,
impact on technology, in general, I think goes into warp speed. Technology’s
gonna go into warp speed if inflation says sticky, because productivity and
labor force participation will create opportunities and tech that me and you
haven’t really comprehended yet. Human and robot interaction becomes a norm much
earlier than one expects. It already has become.

[01:03:24] Pierre Daillie: Right.

[01:03:24] Sri Iyer: Market wall stays elevated, but not in any panic state, in
my opinion. The four-letter word hope, which are H stands for housing, O stands
for orders, P stands for profitability, and E stands for employment are all
shifting and an economic bottom at least in our opinion is not in sight yet. So,
there’s still more fog on the road. So, don’t speed up, slow down. That’s the
overall expectation I have for market volatility in 2023. So, there’re a lot of
nuggets here I put in that I will not explain-

[01:04:03] Pierre Daillie: I love it. I love it. No, that’s perfect. That’s I
think, I think it’s prudent to be cautious, and uh-

[01:04:14] Sri Iyer: But also, I put in some components of where the optimism is
also there. So-

[01:04:19] Pierre Daillie: Yes.

[01:04:20] Sri Iyer: … don’t invest in beta, invest in alpha. Invest in areas
where the market will continue to evolve and grow.

[01:04:29] Pierre Daillie: Sri, thank you so much. I, I, uh, you know, love
talking to you. We love you. It’s always a mind-blowing conversation with you.
You’ve always got a world of, of intelligence to, to share with us, and um, you
have, you’re doing so much. I think, I think if anybody actually got to, to see
what you’re actually doing, I know that’s proprietary, but I, I… It’s
mind-blowing to. To me, it’s mind-blowing. I think what you’re doing with, with
investing is, is, uh, on another scale that that… and it’s not something you
just… You didn’t just start it this year. This is something you’ve been doing
for your whole career, um, working your way up to this, and, and then since,
since 2018, which doesn’t seem like that long ago, but that’s five full years of
full implementation-

[01:05:25] Sri Iyer: Correct.

[01:05:25] Pierre Daillie: … that you worked up to in the previous decade of
your career or decades. And, and um, I, I… It’s astonishing is all I can say. I
think that, that, um, you know, you’re, you’re, you’re a pioneer in, in AI. I
think that’s why I was excited, particularly excited to talk to you about AI
today because it’s been such a big year for, for that and uh, I… I imagine it’s
going to be… It’s making… You know, it’s making the work that you do even more,
uh, profoundly satisfying, gratifying.

[01:06:05] Sri Iyer: Thank you very much. These are exciting times, indeed.

[01:06:08] Pierre Daillie: Thank you, Sri. Thank you so much for your incredibly
valuable time and, and your thought.

[01:06:12] Sri Iyer: Thank you very much.

LISTEN ON THE MOVE

Insight is Capital™ Podcast
ChatGPT, AI, Systemic Risk & The Case for Dividends
1:06:27
ChatGPT, AI, Systemic Risk & The Case for Dividends
1:06:27

30
30
1x

Listen and subscribe on your favorite app.

Apple
Google
Spotify

In this episode, Sri Iyer, Head of I3 Investments™, and Portfolio Manager at
Guardian Capital LP, joined us for a provocative conversation about the banking
crisis, the Fed, monetary policy, ChatGPT and AI, and the case for dividends as
a core and resilient equity allocation for all portfolios.

We discussed the current state of regional bank stocks in the US, which took a
hit due to the uncertainty surrounding the Silicon Valley Bank
meltdown/collapse. However, Federal Reserve Chair Janet Yellen's statement
guaranteed all deposits, implying that the US government will guarantee 100% of
deposits, providing some relief.

TURBULENT TIMES

In a turbulent period, investors are seeking companies with wider moats and
stable or growing dividends, as well as companies whose moats have shrunk.
Depositors may pull their money out of lower-tier banks and shift them into tier
one banks, while investors default back to high-quality, dividend-paying stocks.
So, it's essential to focus on consistent companies with a proven track record
of growing dividends during uncertain times.

STAGFLATION?

We also discussed the concept of 'stagflation', which is a recession amid high
inflation. The market believes that the Fed can engineer a soft landing, but
this is a mistake. The main components of inflation are supply chain problems
and persistent high labor costs due to labor force participation issues.

MORE MARKET VOLATILITY AHEAD

Geopolitical issues, such as the Ukrainian war and China's impact on supply
chains, are structural issues that cannot be solved by mere rate increases.
These structural issues will lead to market volatility and turbulence, making it
essential to separate beta and alpha. As a result, many experts believe that
active management may be better than passive management right now.

DEMOCRATIZATION OF AI

We discussed the democratization of AI through open source tools such as
Pytorch, which is making AI accessible to a wider audience. Chat GPT is one such
AI tool that can be used for making decisions. It has the potential to
revolutionize several industries such as software development, big data storage,
cybersecurity, search engines, media content generation, music, legal sector,
healthcare, pharmaceuticals, predictive analysis, and aerospace engineering.

DIVIDEND INVESTING

Lastly, we talked about how dividends can provide a good conduit to capture
duration visibility and have a mid space between safe deposits and risky
duration, playing a vital role in this market. Also, we discussed how the Fed's
response to the market cycle is measured, and it's more concerned about
protecting the average investor than bailing out failing institutions.

Thank you for listening to our podcast. Stay tuned for more exciting episodes!

TIMESTAMPED HIGHLIGHTS:

[00:01:46] The Fed is responding to the market in a measured way, balancing
inflation and protecting the average investor; not bailing out failing
institutions.

[00:06:06] ChatGPT revolutionized the average person's interaction with AI,
leading to new levels of "humans and bots merging" and the emergence of new
forms of AI like regenerative AI.

[00:09:00] AI processing data and training has been drastically changed with the
introduction of ChatGPT, leading to democratization of AI.

[00:15:30] ChatGPT is a revolutionary AI that can change software development,
data centers, cybersecurity, search engines, communication, media content,
healthcare, pharmaceuticals, banks, and aerospace engineering.

[00:26:32] Dividend paying stocks help investors in turbulent markets with low
volatility, downside capture, cash flow visibility and increased yield at cost.

[00:31:35] Secular cash flow/dividend growth gives consistency to company/cash
flow, allowing for cleaner valuation and better market mismatch detection.

[00:39:51] Fed guarantees bank deposits to protect against cascading credit risk
and inflation. Dividends provide mid-space between deposits and risky duration.

[00:51:59] Recession and high inflation due to labor costs and geopolitical
issues, leading to market volatility and the need for active management.

[00:55:35] Dividend strategy has no cuts since inception; trained model on COVID
data to recognize behavior and infer future trends.

[00:59:24] AI using transformers to create abstract data and learn from
predictions of other models.

Guardian Capital LP is a sub-advisor on numerous funds for BMO Global Asset
Management, BMO Exchange Traded Funds, and Horizons Exchange Traded Funds, in
addition to managing its own suite of investment funds, and assets for large
institutional clients.

Total
0
Shares
Share 0
Tweet 0
Share 0
Share 0
Author
Insight is Capital™ Podcast

advisoranalyst.com Open in urlscan Pro 192.124.249.3 Public Scan

Form analysis 2 forms found in the DOM

GET https://advisoranalyst.com/

POST

Text Content

advisoranalyst.com Open in urlscan Pro
192.124.249.3 Public Scan

Form analysis
2 forms found in the DOM