GMO's quantitative analysis reveals that US large-cap growth stocks are significantly overvalued, with a projected 70% loss over seven years, while emerging markets offer better relative opportunities; the 'Beyond China' strategy identifies countries like India, Vietnam, Thailand, and Mexico as beneficiaries of global supply chain diversification, with India receiving 25-27% portfolio weight due to favorable demographics, labor costs, and infrastructure development.
Install our extension to search inside any video instantly.
Beyond the US Equity Bubble: Opportunities in Emerging Markets, Supply Chain Shifts, AI-driven Alpha
Added:So let's begin with the first session.
The speaker is Mr. Arjun Dvesa. The topic for the session is beyond the US equity bubble, opportunities in emerging markets, supply chain shifts and AIdriven alpha. To moderate the session, we have today with us Mr. Rajiv Tucker.
Uh we all know who Mr. Rajiv Tucker is but I would just like to take a moment to introduce him briefly. Uh Mr. Dr. Rajiv Tucker possesses more than three decades of invest of experience across one management, security broking and merchant banking. Uh he began his career in 1994 and joined PPFAS limited uh the sponsor for PPFS AMC in 2001. In 2003 he was appointed the fund manager for the portfolio management service titled Cognito managing a portfolio of around 300 crores. From 2007 to 2012, he served as the chief executive officer of PPFS limited. He played a pivotal role in our in incorporation of PPFS asset management private limited and in 2013 was instrumental in launching the their flagship equity scheme, the Parak Parik Flexiap Fund. Today stands as the largest one of the largest actively managed mutual fund schemes in India with over one lakh crores of assets under management.
Currently he serves as the CIO equity and director at PPFAS asset management private limited overseeing five of the six schemes introduced under his stewardship since inception. We welcome you Mr. uh Tucker on the stage please.
[applause] Good morning. Uh it's an honor and a privilege to introduce our first speaker for today uh Mr. Arjun Dvesa.
Uh I have his official bio but an interesting uh snippet that came across our conversation. Uh before the session was that he has grown up in Mumbai and still maintains a residence here and uh he mentioned that he uh looks for excuses to come to Mumbai. Uh so hopefully our CFA society will provide him lot many excuses. uh in the [applause] months and years to come. Uh coming to his official bio, he has spent over four decades in global investing and 30 plus years focused on emerging markets.
He's the founder and longtime leader of GMO's emerging market equity strategy.
This was launched in 1993 and it is widely recognized as pioneering a quantitativelydriven approach to emerging market investing.
He has a background which spans engineering, quantitative modeling and portfolio management. So we'll get a glimpse of his engineering background and he mentioned that he started with coding and he's in some form got back to coding in recent years. So we'll get a glimpse of that uh in his presentation.
Uh he has deep experience in capital allocation across global emerging markets particularly India. He currently serves as senior adviser at GMO uh following prior leadership roles as partner board member and chairman vice chairman of the board. He has a pretty heavy slide deck 87 data heavy slides. Uh so I'll not take much of your time. Uh over to you Arjun and uh look forward to the exciting session. Thank you.
>> [applause] >> As he said, it is a great pleasure to be back in Bombay. And again, for those of you who don't know, I refuse to call it Mumbai. It is Bombay to me always.
So, I grew up in Bombay and I grew up in a part of Bombay where I have to confess this is the first time I have actually come to the Taj Lands in Bandra because the part of Bombay that I grew up in worldly was the end of Bombay. So, so anyway, [snorts] um I'm going to talk to you about basically two topics. Does that work? Yeah. So, essentially I really have two separate talks. The first talk really has to do with my work at GMO and everything we do at GMO about investing globally. And then in the last couple of years, I have kind of moved a little bit away from my GMO role and I kind of have become more of a private investor in my own right. And as part of that, I have become what is known as an AI dude. And so I've kind of got into doing a lot of AI stuff. And so I'm going to talk to you about my AI journey. Now, it intersects to some extent with what GMO does, but for for the most part, it's how high I have used AI in my own kind of life from an investment point of view. So hopefully it gives you some sense of what one can do with it. I know that AI is a new thing for a lot of people and I I know that there are people in this audience who are experts in it and for whom everything that I'm going to say will be extremely trivial and there are people over here who have never really experienced any much of the AI other than asking a few questions here and over there. So I'm going to target this at somebody who is more newer to AI to show you what are the things you can actually do with it. How do I actually use it on a day-to-day basis? So that's basically what I'm going to talk about.
So I'm going to start out by talking about the state of the world from GMO's point of view. And the state of the world is really not that good from an investment point of view. So let me start out with something that GMO does very regularly. We've been doing this since the early '9s. And so we have a track record of doing this for a very very long time. And that is our worldview which is encapsulated in something we call our 7-year forecast.
So for every asset class and subasset class, so US equities, US small cap, uh emerging markets, etc. For every asset class, we come up with a seven-year forecast. I'm going to get in a minute to how does where does this come from?
But let me show you the state of the world, which is what this basically shows you. And what you can see over here is that the state of the world really isn't that good in terms of the opportunities to make a lot of money by investing in some of the developed markets and especially the US. Look at the look at the look at the bar. I don't know if you hopefully you guys can see this in the back but the bar the second bar which is US large cap growth. Our forecast is that over the next seven years you will lose 11.1% a year annualized. over seven years you can annualize seven 7 times 11.1 is something like 70%. I.e. we think you will lose 70% of your money over the next seven years if you invest in emer in in large cap US equities today. That is the level of the overvaluation.
Bottom line we think that there is a very large bubble and in fact on all the work that we have done at this point this bubble is larger than the bubble was in 2000 and it is in fact larger than the bubble was in 1929. And needless to say, neither of those episodes ended well. Okay, so the fact is that we're not bullish. In fact, we're quite bearish on the US market.
The good news is that we're less bearish on the rest of the world. And that's what you see on the bars on the right, which have to do with international large, international, small, and emerging markets. Again, those numbers are not nice and big. They're not, you know, if you think about it in the long run, the kind of return to equity markets, for example, the US equity market, the long-term return has been about 6%. And these are all real after inflation. So in the US, if you had invested in the stock market over the last 150 years, on average, you made about 6% a year after inflation. What we are saying right now is that you're going to lose 6% because that's the first first bar over there, i.e. the market is really quite overvalued. Let me tell you why we think it's so overvalued. How do we come up with these numbers? We come up with these numbers based on four subcomponents. The first two are the most critical because the other two don't change very much. The first two are valuation and margins. So the idea is that what's the current valuation of the market and what do we think is the fair value or long-term valuation. So if things revert to the mean i.e. The PE ratio right now is basically so if we think that normally the PE should be 20 uh should be 16.5 which is what we think is the long-term appropriate PE and has been the historical long-term PE and right now it's 27.7 therefore if you revert from 27 to 16 over 7 years you will lose 4.2% 2% a year. Okay, simple math isn't really complicated at all. Similarly, we look at margins that is what is the profit margin of the average company in the market and you can see that our long-term average profit margin is 5.2%.
That's the return on sales. So, that is what is your you know percent what is the profit divided by sales for the entire market. So, the average has been 5.2 and where are we at right now? We're at 7 7.8 i.e. valuations are high, profit margins are high. If both of these revert to their long-term averages over seven years, then you will lose another 2 and a half% from that. And then the other two components are something which don't change that much.
That has to do with kind of growth, which we think is fair steady. It doesn't change very much. And finally, kind of the dividend yield that you collect over this period of time. So when you add those four up, you get the minus 6%. So this is how we come up with and you can do this kind of math for whatever market or submarket that you care about because this is just an arithmetic you know identity where you can decompose return uh in into these components. So the bottom line is that we're quite bearish about the US. Now why has the US come to such an extreme valuation?
Not because things have been bad. has come to these extreme valuations because the US has actually been doing really really well from a fundamental point of view. So let's look at the US relative to the rest of the world. So what this chart shows you is two things. It shows you the the the lower green chart shows you what happened to the reality that is what happened to the earnings of US companies relative to the rest of the world. So starting with one you know back in 1996 and kind of ending up today and then secondly what happened to the market over that period of time. So you can see that all the way for the first 10 years or so those two lines can track each other that is when earnings did well relative to the rest of the world the markets did well etc. But you can see that there's really been quite a significant divergence recently. So, so yes, the fundamentals of US companies have improved much more than the rest of the world and that's what that lower line shows. But unfortunately, the markets have overpriced that have basically given much more of a return than should have been justified. So again, we're not saying that this is a bubble because of something truly irrational. It's a bubble because of something that's rational, but it is in fact still a bubble.
Okay. So what does the rest what does the US look like relative to the rest of the world? So we look at three dimensions. We look at the cape ratio which I'm sure most of you are familiar with which is kind of a long-term price to long-term earnings. We look at a price to forward earnings which is a much shorter term metric. And we look at something we would called you know shareholder yield which is like kind of what's the yield you get after you take out things like share buybacks and stuff like that. And in all three you see the same basic picture that is the US is really quite a bit much more attractive than the rest of the world. the rest of the world really both the developed world and the emerging world look much much more attractive but even then when you look at the red dots the rest of the world is not cheap relative to its own history relative to kind of where it's been so our point is that it's not like there's some great bargains out there it's just that in relative terms you should be in the rest of the world rather than in the US so we think you should be in Japan small caps we think you should be in Europe we think you should be in emerging markets so the fact is that the US is in fact, you know, quite uh uh quite quite unattractive. But that's only unattractive if you're a beta investor, i.e. if you're going long only. It turns out that there's actually some fairly good opportunities in the long short space.
So what this shows you is the gap between cheap and expensive stocks in the US. So if we were to kind of check two groups of stocks, cheap and expensive using our metrics of valuation. So you have number of different metrics, things like price earnings, price books, you know, v various metrics and you put that together. This chart kind of shows you how that kind of has changed over time and where we are at right now is very similar to where we were in 99200. That is there's a massive gap between cheap and expensive stocks. So the bottom line is that there's a really nice opportunity here to make money by going long cheap stocks and going short expensive stocks. So we have strategies which do that which basically go long you know cheap US stocks and go short expensive junky uh US stocks. So that's one opportunity we see.
We see a very similar opportunity in emerging markets as well. Essentially, this chart shows you the same thing in emerging markets where there's the gap between cheap and expensive. Again, it is extremely uh you know, it's not quite as big as it is in the US, but there's still an opportunity in emerging markets to buy cheap companies versus expensive companies. So, the bottom line is that on one hand, we don't think anything from an absolute basis looks really really attractive. But on a relative basis, if you can go long short, there are actually quite few interesting opportunities.
Now, let's see why and what has happened. So, let me just show you kind of a little slice of what happened in the recent past. The recent past has to do with really starting on, you know, when when Trump announced the tariffs because that was kind of a seminal day in the market. From then for the next 6 months or so, what happened to markets and you can see that the S&P went up 35%. But all the really speculative stuff is up a heck of a lot more than that. Things like, you know, uh the the what do you call the uh quantum stocks and things like that, they all went up a lot more than that. So the fact is that we this is really showing you speculation and how much how how alive speculation really is in the markets right now. Obviously, not a good thing.
So we're quite bearish on this.
Okay. So let's switch gears since uh you know we since as as you know my my expertise really in the long run has been much more so in emerging markets.
So let me tell you very briefly what I think about emerging markets and then switch to India.
So my view on emerging markets is actually quite simple. Basically, we have a situation right now where there's a lot of uncertainty which has been created by Mr. Trump's tariffs because obviously the tariffs make a big big difference to people. But the good news is that most of the countries that we invest in are actually in a pretty good place. So one of the ways we look at it is to say how vulnerable are emerging markets to some kind of a crisis because in emerging markets you tend to lose money at times of crisis rather than at other times obviously. So one of the things we focus on is kind of the vulnerability to a crisis and one of the metrics we look at is which countries have large current account deficits you know and that because that tends to be a real red flag if countries are running big current account deficits. They're the ones which kind of basically blow up and you have problems. So one metric we follow is what percentage of emerging markets have large current account deficits which I think we define as more than 3% of GDP. And you can see over here that we're kind of close to the all-time lows. No worries at all. So from a fundamental point of view doesn't look bad at all. Another way of looking at emerging markets is to say well what about inflation? How is inflation under control or not? So we kind of have four buckets over here. We have you know are you above target and rising, below target and rising you know etc etc. And what you'll see is that right now every single emerging market that we follow is in fact below target. In some cases rising, some cases falling but every single country is below target. So inflation isn't really a problem right now. Now might that become a problem in the future? It might. But at this moment in time across the board there's no place that we are worried about rising inflation. Okay, so these are important things to say from an investment point of view especially at a time when you worry about the US having a big problem that is how much spillover is there likely to be when if there's a crisis in the US how much will it spill over to places because when the US has a problem and the spillover happens it's going to happen much more to places that are vulnerable rather than to places that are not vulnerable so that's one of the things that we focus on okay so let me talk a little bit about what we like in emerging markets given what I've just said. So let me tell you about our investing framework and let me spend a few minutes on this slide because it'll give you some insent some ideas as to how we invest.
So as people ahead of me have kind of mentioned we are largely quantitative investors and quantitative investing means that you use data and systems to figure out what's attractive rather than kind of fundamental bottomup research.
So we build models to to basically model three different phenomena. So we look at everything from three points of view and we do this at every level. We do this for countries, we do this for industries and we do it for stocks. So you can think of this framework is the same framework few differences but the same framework for all three. Okay. So let's talk about them. So first and most important as value investors value is the single most important and the thing that we do is both bottom up and top down. So bottom up means at the stock level top down is at the country level or industry level. And what do we look at? We look at things like something we call price to fair value quality boosted value at these different levels. So these these are metrics based on things like e economic book and things there are lots of sophisticated ways of doing fancy things like PE. So so the bottom line is you're looking at things like PE price book dividend yield but with some sophistication because you have to deal with accounting issues you have to deal with buybacks and all kinds of things like that. But at the end of the day what we have is you can think about it this for every country for every industry and for every stock we have a number of how attractive is it from a value point of view.
Second momentum momentum is based on a very simple phenomenon that is the trend is your friend that is whatever has been doing well will continue to do well in the future. And again I'm sure most of you are familiar with the fact that momentum strategies have in fact worked quite well over the long run. So we basically focus on two different kinds of momentum what we call idiosyncratic momentum that is the momentum of that particular company or country and network momentum that is the momentum that has to do with the industry or the or the whole global phenomena you know so it's kind of the momentum of the of the environment versus the momentum of the individual stock or country relative to the environment. So we do both of those and then finally the the first third and very important component is quality because at the end of the day both of these both value and momentum are vulnerable to different phenomenon.
Value investing is very vulnerable to to you know to basically getting caught in value traps where you think something is cheap but in fact it's not cheap and momentum again is very vulnerable to whipsaws when things go up or down. So to some extent val quality is what protects you against some of those. So part of the reason why you use quality is for protection. And what is quality?
One of the definitions of quality I used to say back in the old days was a quality company is one which could have survived the 1930s. Okay that is if you had a depression and you go through the whole depression and you can survive that's a quality company. So what defines a quality company? Basically three phenomena. Number one we call predictabil uh we call it uh profitability that is are you highly profitable what's your you know what's your roe and things like that profitability second is predictability that is how stable is that profitability is it something that just varies from year to year or does it basically is it fairly stable and the third one is something we call safety that is if you have a lot of debt well that's not safe so the less debt you have that's that's a debt so there are many things alman zcore there many ways of defining safety So the bottom line is that we we come up with all three. Remember it's profitability, predictability, safety and we basically come up with different metrics for those and so we come up with the quality score. Again remember we have a score for every company for every country and for every industry. Okay. So remember these these three are important because this I'm going to talk about this. So now based on this we build global portfolios. So we have a emerging markets portfolio. We have a global portfolio. We have international portfolios. But they're all based on the same underlying models. So let's look at what we like right now and what we don't like in emerging markets.
So on the right hand side again I'm I'm sorry if you in the back can't see this very well. So I'll read it out. But the countries we like the best are on the right hand side are Indonesia, Thailand.
And then the countries we dislike the most are South is it South Africa? Yes, South Africa and India. India is our least favorite country in emerging markets. Now why you can say why don't you like India? Well, I just told you its valuation, its momentum and its quality. All those three those those are the three things. India does score very high and highly on quality it turns out relative to emerging markets. But it's right now scores fairly pretty poorly on on valuation. And again I suspect most of you know this the moment momentum isn't really that great in the market right now. So they so so you know so that's what what drives it. Similarly if you look at the sectors which are doing well or badly it's the industrials and healthcare look most attractive and communication and IT look least attractive again mainly because of the valuations. Okay.
All right.
So let me talk about a concept that we we we we've been talking about for the last 3 years but I'm very pleased that we launched an actual strategy. We actually launched an ETF a year ago called beyond China. What is the idea of behind beyond China? The idea is a very simple idea that what is happening is that there is a major move happening in the world and this is really a big secular shift of companies moving supply chains out of China. Okay, so let me talk about why this is the case. Why is it that people are moving supply chains out of China? It really started to accelerate during CO. Why did it accelerate during COVID? Number one was because during COVID a number of companies found that they could not in fact get any supply because they had a single source supplier. Their single supplier was in was in China and it was closed. China was closed. So you could not in fact get supply. So people said well we really need to diversify our supply chain. In fact a good example of that was um a a US cosmetics company which bought specialty chemicals from a Chinese company. And in the past their view of the world was well when we need this chemical we just call up our supplier you know who's a US person and we say hey get us some more of this chemical. And then suddenly the word comes back well guess what China's closed we can't get you any. Now we have a problem. So at that point in time they thought to themselves you know what we really can't afford this. We cannot afford to have this single we cannot afford number one to just simply calling up this guy and saying send us some. We have to kind of be much much more in control of our supply chain. And secondly we need to diversify our supply chain. So what did they start doing?
They started buying 20 to 30% of their supply from an Indian supplier. Now why were they not buying it from the Indian supplier? Because the Chinese were 20% cheaper. Okay, very simple, right? But now because of the need to diversify, they're willing to pay that premium to the Indian supplier because they can't afford to have that that that vulnerability. Okay? And now you couple that with the tensions between the US and China and the rest of the world and that gets exacerbated because now essentially people you know as long as the world was moving in one direction which was towards globalization.
You didn't really have to care about where does it come from whether it came from China or India and you didn't care.
You just said get me the cheapest supply. But now in a world which is fragmenting, you have to be very aware about where it's coming from because now you have to basically control your political risk so to speak of your supply chain. So that's the other reason why people have done that. The other reasons why people are supplying are diversifying away from China have to do with cost. Labor costs in China today about $8 an hour versus $2.5 to $3 in Mexico and even less than that in places like India and Indonesia. So therefore there's very good reason for a cost point of view. But it's not it's not a very fair comparison. The problem of China is that you have your supply chains all so close to each other. So if you're making shirts in China, the guy who makes the buttons is down the road.
The guy who makes the thread is down the road. Whereas when those people move their factory to Mexico, what they find is they still have to import those buttons from China, you know. So the fact that the cheaper labor cost is not necessarily a, you know, a good thing.
So bottom line is basically that u you know there are many reasons why people are moving their supply chains from from China and I give you a prime example of the phone that's sitting over there.
It's the phone I bought a month ago.
It's a new iPhone. Guess what? It was made in India. Why? Why do I know that?
Because 100% of every phone that is sold in the US today, iPhone that is sold in the US today is made in India. Okay. So you're seeing this across the board.
We're seeing that and now people say that yeah but what proportion of that phone is actually made in India? It turns out a very low proportion. I I actually asked Chad GPT that question. I said what proportion of this phone is actually made in India and it's actually fairly low but going to about 20% next year because starters are going to start making the the the frames for them. Uh and so about 20% of the value of the phone of the cost not the value uh is actually going to be made in India next year. Right now it's lower than that. uh primarily the components come from Taiwan, Korea and Japan. Okay. And some from China. So but the point is that we now see Corning Glass setting up a glass factory in Bangalore which is now going to make the glass for this. Okay. So so as these supply chains move slowly you start to get all the the ancillary suppliers also move and so that's what you're going to see. So my point is that we think that there's this is a global phenomenon which is going on you know for quite a while. So let me show you the reality of what's happened. Again, I apologize for those of you in the back.
I will read some of these numbers.
What we have on the first line is what proportion of US imports came from China. Okay, that's the first line. And the first number which is China, which is the peak, which is 2018, just before Mr. Trump imposed his first tariffs on China, was 21.2%. That is 21% of all US imports came from China. Okay, what is that number now? And this is 6 months ago. From July 2025, it's 9.4%. Okay, it's gone from 21 to 9.4. Now think about it. Think of how much stuff the US imports. That's a massive change that's already happened. Okay, huge. So you've already had a huge, massive shift in how much is being imported. So the next question becomes, well, did the US start making all this stuff themselves? No, they did not. They just imported it from somewhere else, right? So, who benefited? So, that's what the next columns should go over. So, who are the ones who benefited? Mexico, Vietnam, India, Korea, Taiwan, Thailand, and Europe. Europe less so than the rest of the world. But a lot of these countries benefited massively. And this is going to continue. This is, you know, going to see that. Now, of course, this is all completely complicated by Mr. tariffs reg Trump's tariff regimes which are now in place. Uh again our view is that the tariffs are a complete idiocy and sooner or later they will get replaced by something which is more sensible. It may not be zero but it'll certainly not be 50%. We think that will go away. We just I mean that's just it's an idiocy which can't really persist. So I'm not really counting that and we don't really invest on that basis. You know we don't we're long-term investors and we think that this is a long-term phenomenon which you know the tariffs won't really have that much of an impact. I mean you think you got to think about in the short term.
certainly but in the long term we don't think it does so we're not focusing too much on that okay so the other thing that we see the evidence that we see is that the FDI which is going into each of these countries which is Mexico India Vietnam Thailand Indonesia also shows a fairly big jump up that is the money going in to actually build plant and equipment and things like that so the fact is that this the supply chain movement is actually real and we're seeing evidence of it so Again, I'm not talking about something I'm not telling you this is going to happen. No, this is already happening. It's already well underway.
That's the point.
Let me give you an example of a company that's benefiting. There's this company in Mexico called Vesta. What does Vesta do? They build industrial parks. Okay?
So, these are the places that these companies which are relocating are going to basically base themselves. So, they have facilities around Mexico, a lot of them on the border, actually on the northern border. and they set up industrial parks and then you get companies who are moving over there.
Interestingly, the irony is that a high proportion of the companies that are moving there are in fact Chinese companies. You know why are Chinese companies moving from China to Mexico?
Well, for multiple reasons. One, as I said, labor costs are cheaper in Mexico than they are in China. But that's not the main reason. The main reason is because the tariffs. Because despite all the noise that Mr. Trump has made about Mexico and Canada and places like that.
It turns out that on his tariff regime, Mexico and and Canada have the lowest tariffs of any country in the world at about 5.4% is the average tariff because all the stuff that was covered under the trade agreement is still tariff-free. So the fact is that Mexico is still a terrific place to move if you want to basically lower your tariff regime. So from the Chinese manufacturers's point of view, they're saying that well in China we're going to have these high tariffs forever. So let's move facilities to Mexico. And so that's what we are seeing over there. And plus the other thing which is really easy to think about is one thinks of China as being this monolith that all these people are kind of kind of one group of people all serving the same master i.e. Xi Jinping. But that's not that's not how it works. When you go and talk to these people, they're business people.
They're there to make money. Okay?
They're not trying to serve some Chinese greater good or something like that. If you can make more money by moving off a facility to Mexico, heck, we'll do that.
And that's exactly what they're doing.
So, so that's what we are seeing that basically we're seeing this kind of supply chain shift. Again, the problem of course is the fact that they need their ancillary supply chains to also move. So, the guy who's making the shirts still needs the button guy to move over there as well. What what is the bottom line of this? The bottom line of this is basically that costs for people in the developed world are going to rise. This is basically a force, an inflationary force because at the end of the day, the iPhone that's made in India is more expensive than the iPhone that's made in China because we simply don't have the efficiencies that the Chinese have. And that is true for every single one of these things which is moving.
Let's look at India again at a 30,000 foot view just looking at the capex of the top 500 companies. Uh again a fairly nice step up we've seen in the last few years. Now how much of it related to this phenomenon I don't know but the bottom line is that you are seeing capex basically being spent in terms of plant and equipment and things like that. Uh let me give you one really small example. People say to me that yeah but this already happened so hasn't this whole thing played out? Let me tell you why this hasn't played out and why this is a multi-deade thing. Let me give you a very simple set of numbers. Okay, basically when we started this chart in about 2122, uh India's exports of electronics, this is not software, this is just pure hardware electronics was about 20 or$25 billion, right? China's exports of electronics is $1.2 trillion.
So let's say that India wants to take 10% of what China currently does. just 10% that's $120 billion that's 6x of what we currently do. India has to invest and build six times the capacity in electronics to take 10% of what China currently does. Okay, that's how big China is. And that's not just true for electronics. It's true for everything.
What this chart shows you, this is one of the most interesting charts of all is how many times just like I said the 6x for China, that's the that's the top right brown bar. That's the 6x for China. But this is the number of X's that you need for every other sector. So you have it for automobiles, for textiles, for chemicals, you know. So my point is that people don't realize China is so big that to just take 10% of China's current exports means all this investment has to be made and therefore the opportunity set to invest in a lot of these companies and countries and that's why we started the strategy to say okay let's invest in the companies and countries that are benefiting from this trend.
And the crucial question for us already is well is this theme already played out as you know just like AI you know AI has gone up 500%. So maybe people have already figured this out and you know it's a well-discovered theme. Turns out not here's the valuation of this group of what we call uh you know beyond China companies. So we we came up with the list of companies that are benefiting from this phenomenon across the globe.
And uh it turns out that the valuations are still pretty reasonable. They're a little bit more expensive than the market 1.1 times, but it hasn't really changed. So, we think that this is an undiscovered phenomenon which we like a lot, which is why we launched a new strategy. Okay. So, how do we do it? How do you actually decide which countries are attractive and how do you decide what money to put in it? Um, basically we start again with a top- down view of the world. We we started with good countries and we say well let's look at three things. We look at labor and demographics. that is does the country have the labor and the demographics to be able to do this. You know the bottom line is that you can want to build all the iPhones you want but at the end of the day if you're the Czech Republic and you have four million people it ain't good enough you know so India is very attractive from the labor and demographics point of view the demographics are very attractive you have a long small you know very very very act you know young population plus there's plenty of people and that also helps from the point of view of having a domestic market that you can sell onto.
Secondly, we look at quality. That is, is the country really set up to do this?
Do they have ports? Do they have, you know, infrastructure? Do they have financial systems which can basically deal with it? And finally, what is the evidence of global trade? So for this, we actually use something a metric that we've come up with which is a global trade momentum phenomenon. That is, are we seeing evidence that trade from these countries is actually in increasing?
Because all the other things are good ideas. So it turns out that a lot of the data for this is not that easy to collect especially the trade data and things like that. I mean it's all out there but you have to nobody's really collating it in a form that's useful to us. So we have to do a lot of work on that. So once we do this and we figure out okay which countries actually are attractive from for this phenomenon i.e supply chain and you can see the answer here and it should supply surprise nobody. You should see who are the countries who benefit hugely. Vietnam is by far the biggest beneficiary. Why?
They have a young well educated population. Labor costs are cheap. They got very good port infrastructure.
They've got, you know, they've done all the work already. It's already well in place to do it. You know, India is obviously attractive, but India still suffers from infrastructure problems and other things like that. And then on the left hand side, you have Kuwait, Qatar, you know, basically small singlestate kind of countries which don't really have the ability to do that. It's funny because I was doing this presentation in Kuwait a little while ago and they all came up to me and say what's wrong with Kuwait? Why don't you [laughter] like Kuwait? And I said there's nothing wrong with Kuwait. It's just that you guys really don't have the ability to do this. Okay. So we start out at the country level. We say these are the attractive countries and then we say okay what are the sectors or the themes rather that are really attractive. So we have a multiple set of themes like things like is there higher economic growth due to this. Uh which are the companies that are taking market share from Chinese companies. Is there real estate and infrastructure being built?
Those are the kind of companies that would be interesting. You know is there a rise in industrial activity and can we kind of invest in that? So the idea is that we figure out themes and from these themes we figure out which are the sectors that are attractive and then we use our classic models which I talked about value momentum quality to pick companies within those attractive sectors. So the idea is you pick the attractive countries pick the attractive sectors and then pick the companies that are traditional models like within those sectors. Okay. So who scores high? Where where do we end up on this? Again I apologize for you in the back. I'm going to read it to you. The country with the biggest weight is India with about a 25 or 27% weight followed by Taiwan, South Korea, Thailand, Mexico. And you see Vietnam kind of shows fairly low. It's our favorite country. But the fact is that you can't really invest very much money in Vietnam. It's a very small thin market, very difficult to really find companies. Most of the companies are SOE rather than private companies. So it's kind of hard to invest a lot of money.
What? So, so basically the bottom line is we end up with a portfolio that again shouldn't surprise anybody. So this is where we're at. All right, let's switch to what I call a quantitative view of the Indian market. So remember, I'm not a bottomup guy. If you ask me about any particular Indian stock, I have no clue as to whether it's attractive or not.
That's not where I come from. I really I'm a 30,000 foot guy. So let me show you the 30,000 foot view of India. And most of you are not going to like it.
So all right just to remind you how we do things value momentum quality right so we're going to look at it these through these three lenses so when we look at the world first of all we don't have a separate India model we have a global model so we start with a global model and so what I'm going to show you is our global model what does this show this shows you each country again way too small to see but there's 40 or 50 countries on along the curros And what each uh column represents is an industry. So these are all the different industries. And this is just one of those three dimensions. Remember there's three dimensions. Value, momentum, quality. This is just value. That is on a valuation basis. How attractive is this country and sector? Because this is how we manage money. We manage money by saying how attractive is the country?
How attractive is the industry? And how attractive is that combination of country and industry uh relative to everybody else. Okay, that's what this shows you. And so if there are a lot of red bars, red, that's really bad. If there's a lot of green, that's good. So if you look down where it says South Korea, which you can't really see, but you can just look at the line which has a lot of green on it, that's South Korea. So that says that South Korea is really attractive because there's lots of sectors in it which are particularly cheap. Okay?
And you can see where India lies, which is there's a fair amount of yellow over there. Not a lot of red, but still quite a lot of yellow.
So let's just kind of parse this two different ways. Let's look at countries and sectors. So if you look on the left hand side, this kind of adds these up.
If this is just kind of adding up this each row, adding up each column. And if you look at uh the the countries, you can see India kind kind of lines up near the bottom of this from again this is valuation. This is how expensive the country is relative to everybody else.
What's the cheapest? The cheapest are South Korea, Canada, Hong Kong, UAE, Norway, France, South Africa, Poland.
Those are the ones which are which are the most attractive. Okay. And then similarly you have the same thing with sectors and you know you go from top to bottom on the sectors. Okay. All right.
So now let's zoom into India itself and say let's look at sectors within India and how attractive are various sectors.
Again remember this is the quantitative lens. I'm not doing any bottomup work over here. I'm just simply taking the data and basically doing that and what we have here and I'm going to show you three charts. The first chart is valuation which are the sectors are the most attractive from a valuation point of view. Then I'm going to show you from a quality point of view and from a momentum point of view. So what we see here the sectors that are the most attractive from a value point of view are things like real estate, software, energy and the things that are least attractive are energy, consumer discretionary, transportation and real estate management. Okay? Again, if you ask me why, I have no clue. All I can tell you is these are what the numbers show. This has to do with valuations. It has to P ratios, price, etc., etc. I I don't have any insight into is what's going on in that industry. This is just purely the numbers. So this is valuation. Let's look at quality. Again, this should not surprise anybody that certain industries are higher quality than others because and that doesn't change. Valuation's momentum change all the time, but quality doesn't. Quality tends to persist. So what are the highest quality industries? You know, it's pharmaceuticals, household products, automobiles, transportation, software, utilities, capital goods.
Again, no surprise. These these are all high quality industries and that should be the case. Uh so again we're not investing based on this but remember this serves as kind of a break on everything else.
And finally we have momentum. What are the sectors with the best momentum? It's semiconductors, pharmaceuticals, technology, hardware and the ones at the bottom are things like household products, consumer services. Okay. So the bottom line is that this is the world. This is how we invest our money because India is just one country out of many many countries and again similarly within emerging markets. We're using this as the basis to decide how much money you put in a country, how much money you put in an industry, how much money you put in that combination of industry and country. Okay.
All right. We're done with part one. We have 27 minutes left and I'm going to try and tell you about AI and hopefully you'll have more fun with this because this is the positive part of the talk as opposed to the negative part of the talk. So basically let me tell you my AI story. I started uh about about a year and a half to two years ago as I started to I I turned 70 this year and so I figured that okay as I'm kind of moving away from my day-to-day role at GMO uh I'm going to do other things and so I decided okay let me focus on kind of managing my own money and a friend of mine said to me well you know you used to write code a long time ago back 40 years ago I used to you know I used to be a coder when I started my career but of course I hadn't written code in 40 years so I have no idea what's going on with writing code and he said well you should try using this program called VS code from Microsoft and because you can try and write some code uh you know to help you build your own investment programs. So I started playing with that and interestingly enough I got deeper and deeper and deeper into it and so to the point where uh I'm really deep into AI stuff right now using AI not only to do investment stuff but also lots of other fun stuff but what I wanted to share with you is kind of my journey of how I have used AI in various ways and how AI is useful to me. So this is kind of my what I call my take-home section.
Hopefully you can take home something from this that this could be useful to you. I know for some of you this will be stuff which you did a year or two years ago and you'll say this is very elementary and for some of you will say wow that's kind of new and I could try that out. Okay. So let me just talk about the three broad areas that I'm going to talk about. First I'm going to talk about using AI for research. That is for looking up stuff for basically researching ideas concepts things like that. That's the first piece. The second piece is actually using AI to build things. Okay, that is to build investment stuff and that's what I call vibe coding. So if you guys have not heard the word vibe coding, you will.
Okay, vibe coding is basically something is a is an expression that was invented by this guy called Andre Karpathi.
Again, if you haven't heard of Andre Karpathi, you should look him up. Uh he is one of the gurus of AI. He's a guy who originally started his career at uh at Tesla. He wrote most a lot of the self-driving stuff for Tesla, then moved to OpenAI and then basically hated OpenAI and left and started his own thing. But he's a really great thinker and writer and he's one of the people, as you'll see later on, one of the people I suggest you follow on Twitter.
So uh Karpathi came up with this idea of vibe coding and I'm going to go through an actual vibe coding session with you which is how do you actually use AI to build something not by programming it and stuff like that but just by telling the machine do this do this do this and it does it and you're going to see what I do. So that's the second piece of using AI to build stuff. Okay. And then the third piece which is perhaps the most interesting but I have the least to say about and that is using AI to actually generate alpha to generate investment ideas to actually invest you know things like that. So here's an example where um people who I'm sure you're familiar with like Renaissance Technologies, Two Sigma and all the kind of the high frequency hedge funds, they've been using this stuff for years long before even the word AI came into our you know they used to be it used to be called machine learning. Okay, what is now called AI used to be called machine learning. So so people like you know renaissance have been doing this for decades actually for us it's all new. So that's a much much more difficult thing to do and I'll show you some examples of what you can do but in my opinion that's the hardest part of AI and maybe the biggest promise but that's the one which requires a lot of skill.
Okay the first two things don't require any skill. You and I can do this any easily. Okay.
So let's start out by talking about well let me tell you something before I start. What are the models that I actually like? So I'm going to give you a bunch of pages to take home. I hope hopefully somebody will give you access to this stuff but because there are so many different models and there's so many different things and one of the nice things of being in the position that I am is I get to experiment and play with everything you know so I I play with the stuff I know on a daily basis exactly what are the most recent models what's happening with it so I'm going to tell you what I like okay so so first of all AI is used for different things it's not one AI that you use you use different AIs So first of all for search so obviously most of you know Google also now has a bunch of AI stuff built into search and it's pretty good but quite frankly I prefer Perplexity and I know that for a lot of people in India uh Perplexity actually has a deal where you can get a free one-year subscription uh for if you have I I'm not quite sure what but if you have something you can get a free one-year subscription and but even if you don't the free version of Perplexity is good enough and the nice thing about it is that it it is a way of doing search but it is a way of doing kind of uh very simple but quick search and the beauty of perplexity is that not only does it search but it also has kind of some programming built into it. So for example I asked it to make me a chart I asked it I said I want a chart of India's trade deficit divided by China's trade surplus. Okay, now think about it.
Both of these exist, right? I mean, if you ask for India's trade surplus, yeah, Google search can give you a graph of that. And if you ask for China's trade surplus, we can give you a graph of that. But what Plexity was able to do is to actually get both of these, write a small Python program which divided them and then plotted gave me a plot of it.
Okay, so that's the difference. Whereas Google will go and find something that exists, perplexity will actually do some can will let you take the next step and do something with it and actually combine information in a way. So that's why I like perplexity. Now Google is increasingly giving you these kind of abilities. So this is changing rapidly.
But I my my so so my daily Google search so to speak is perplexity. That's what I use. Secondly, there's kind of general use. It's like you know you pick it up and you ask it a question. I use both chat GPT and Claude CL but they are both my favorites. Uh my wife likes to say that Claude is my new best friend and uh not not not not completely untrue. Um so and those are for your general questions you know just when you're basically asking it stuff like that then there's coding and for coding there is one model which is just ahead of everybody else and that's Opus 4 which is from Claude okay it came out about a month and a half ago and it is really really quite a phenomenal model for coding and it really takes stuff to the next level. There are lots of good models and I'll show you later on a list of all the good models. But the fact is if you're writing code, Opus 4.5 is just better than everything else. Okay, don't let anybody else convince you otherwise.
Okay, now this is another thing which is very useful. So this is something which people if you don't know this by itself could change your life and that is deep research. The fact is that all of these AIs and in particular Gemini and Chad GPT are particularly good for doing deep research.
So, I'm going to show you an example of that and what it can do. Then there's graphics and you'll see a number of graphics a number of graphics in this presentation as we go forward which I've actually had AI built for me. They're beautiful graphics. I have to say they're absolutely beautiful graphics.
Uh and and and that's Nano Banana which is part of uh Google. I don't know how the hell they came up with a name like Nano Banana but then it's Google. They have all these silly names for operating things. And then finally for presentations another Google thing which is notebook and I'll show you another phenomenal example. Okay. So let's jump into some of these things I've told you about. If you notice all the models I told you are American models. These are all US models. What about the Chinese models? In my opinion none of the Chinese models really rise to the level of the top American models. They're a step behind but they're a close step behind. They're not really that far behind. And this has to do with an interesting phenomenon that I think is important for India and that is what is India's role in AI. How how should India basically think about AI and I am very much of the view of supporting what Nandan and Nelican said and that is that India should not spend time trying to develop frontier models. Okay. I think we we need to do some work in it so we can have indic models and things like that and I support the work that Saram and also my alma mat Jen which is doing at IIT Bombay uh and I I think that's great work but at the end of the day I think what we need to do in India is to actually use the models that people have made and so remember the difference between the open source and the closed source models is this the closed source models which the American models by and large if you're going to use them you have to pay to use them you know it's a subscription service you you subscribe to it pay it whenever you use it you have to pay them the Chinese models for some odd reason which I never understood are all free so they basically given you all these Chinese models and they've not only made the use of it free they've given you the entire model i.e they've given you the weights which means that you can reshape them any way you want you can what is known as ameliate them i.e you can dechinify them if you want and that's exactly what perplexity did they took a big Chinese model which basically refused to give you answers on things like what happened in Tanaman Square and they dechinified it so that basically it will answer any question at all right so the fact is that you can take these Chinese model and reshape them to however you want and that's what I think people in India should be doing is basically taking these Chinese models use them as a base and then train it with your own data train it with whatever you're doing for your specific use case okay so So the analogy I like to give is that you can think of the Americans as being the people who have cut a swat through the forest with machetes, hard work, a lot of money involved.
The Chinese have come along and paved the road and we need to set up the chai stands. Okay, so that's our role. Our role is to basically set up the chai stands. Don't bother building the road.
Don't bother cutting the path through the forest. Let somebody else do that.
let them you know we should basically do where our advantages which is setting up the chai stalls okay and extracting rent through that so that's my two cents worth of what I think India should be doing in AI okay all right on to how do I actually use it okay so for basic research again this should be obvious to everybody but if for those of you who have not done this yet lots of AIs will give you terrific answers to very very simple questions so for example there's this website called Perplexity Finance which works pretty much around the world and you can just give it a company name and say tell me about Mahindra tech or tech Mahindra in this case and boom literally just say tell me about tech Mahindra and it gives you a full page it shows you the you know it shows you pretty much everything you want to know at a glance okay so if you're not using this you should be using this this is and this is true not only for for perplexity but also for Google so you can basically just say give me a one pager uh you know give me a onepage stock market style report on NVDA. Okay, literally that's the prompt that I give it and then as soon as it comes back with it and I say now generate a onepage infographic from this information and boom, it produces this really beautiful infographic. Okay, so it kind of it's a two-step. Step one is say get me get me data on get get me a one pager on on Nvidia and then make a nice infographic with it. Boom. Two lines and you get a page like that.
Okay. So again it it just makes life easy for two reasons. One is that a you get good retrieval and two to the extent that you need infographics or you need something to present to a client or to present somebody else it makes really life easy to put it in a form that's useful.
Okay. Deep research. This I think if there's one thing you can take home from here which you can do today is easy to do and you should really do it is deep research. Most people don't realize that the deep research these that these AIs do is just mind-blowing because what do they do when you ask it to do re deep research and remember this is not just typing deep research. Each of these AIs actually has a button called deep research. So if if you're in Gemini there's a the when you go and you click on the drop down it'll say deep research. You've got to click on that.
Same thing with chart GPT same thing with perplexity same thing with uh uh claude. Okay. So what does deep research actually do? It takes your question and it basically goes to hundreds and hundreds of websites, tries to find the answer to that, then assembles a a huge report and literally most of the time we'll we'll make a report which is 20 or 30 pages long. So this is the equivalent of asking somebody to do you know an analyst who would take two months or three months to do this will come back in 10 or 15 minutes and give you the answer. Okay, this is seriously good.
Okay, so what I'm going to tell you about is a way to leverage this of how you can make this even more useful. And that is to use a two-step. That is rather than asking deep research directly what to do, you actually ask the AI, not the deep research part of the just the regular AI to write a prompt for deep research. Okay, so let me read this to you. Of course, I can't read it from here. Let me just turn around. says, so here's the prompt that I gave chat GPT, okay? says, "Write a prompt for deep research to identify and deeply analyze all publicly listed Indian companies benefiting from supply chain of diversification away from China, detailing their sectors, export exposure, PLI participation, foreign client linkages, recent capex/FDI and supporting evidence from credible Indian and global sources." Okay. So I gave this to chat GPT, right? So so I'm not asking it to do the research. I'm just telling it to write a prompt that is then going to be given. So it's two steps. Okay. So that's step one. So here's the prompt that it came back with. So here, so this is the research prompt. The prompt says objective.
Compile a complete list of publicly listed companies blah blah blah. This says list of deliverables for each company. Provide company name, ticker, industry, market cap, services, blah blah blah. So there's a whole long list of stuff. Then it says required sources.
Go to SEBI, go to NSC, go to my Ministry of Commerce, blah blah blah. You know, it gives you a whole list of sources. So this is telling deep research what it must do. See, so these are the instructions that are being given to deep research. Similarly, these are the key key research themes. Okay. So now then I copy this whole thing and then I paste it into the deep research function. So now when I click on deep research, I paste this whole thing into deep research and what do you get? A 25page report. I can't show it to you because there's 25 pages worth of stuff.
I'll show you the first page of it.
Here's the first page, but it goes through it gives me an analysis of what I asked and it goes company by company giving me that that whole detail. So, it was sector by sector and company by company within that 25 pages 10 minutes.
Okay. So, if you're not doing this as you can imagine, you can think of any question. Okay. I mean, literally, it's like I have arthritis of the knee. I could I asked it to do a deep research on all the most recent you know what are the most recent clinical studies on osteoarthritis of the knee and boom it came back in 10 minutes and gave me a 25page report on all the most recent research on on on osteoarthritis of the knee. So my point is you can choose any subject and if you're not doing this it's really useful. The downside is that both all these people limit how much you can use it. So if you have a subscription to chat GPT they let you use only 10 of these a month. So you have to be judicious as how often you use it. But the fact is that you should be using this. All right.
AI for presentations.
Um, one of the things you can do is you can give the AI something and tell it to make a presentation. So here, this is something I was working on called Droid fine-tuning. I was fine-tuning kind of models. And so I have what's called a GitHub repo. A repo is kind of a place where you store all your code. So there's hundreds and hundreds of lines of code. There's 20 or 30 programs. So it's a whole program which does a bunch of stuff.
So what I told Nano Banana to do was read this entire repo which means read all these programs and all that. Come up with a flowchart of how this program works and make a single page infographic of it and make it in a Lego style. Okay. So what does it look like here? I love this.
This is my favorite slide of all. So, so this is actually a flowchart of that entire complicated program of you know 20 or 30 different programs and stuff like that. So it shows you on the left hand side there's the model training there's kind of you know the optimization etc etc. So my point is that it has taken a very complicated 30 about hundreds thousands of lines of code read all of that figured out the flow of all of those and then produced this. Okay. So again, you know, it's very useful to for condensing and presenting information in a condensed form which is useful. All right.
The next thing to do is to use something called notebook LM. Now what is notebook LM? It's a program which is used by Google. Okay. Now this is very very useful for lots of things. I'm going to show you how it's useful.
So the first thing I did was this is a paper that I wrote back in 1990 called emerging markets a quantitative perspective. It's what got me into the business of emerging markets. It's about a 25page paper just published in the journal of portfolio management. Anyway, so I just pulled up this paper. It's a PDF. I just simply pulled up the PDF. I pasted it into Gemini and to not Gemini, sorry, into Notebook LM which is part of Google. So I literally just pasted it over there. Once you paste it over there, it gives you a menu. Here's the menu. These are the things you can do with the thing that I've pasted. It can make an audio overview. If I click on audio overview, it will make you a 10-minute podcast. I'm not kidding you.
A podcast of two people talking about the paper. They will literally go through the whole paper and discuss what's going on in the paper. If you haven't tried it, it's it's brilliant.
just you know you know Isaac Asimov once said any sufficiently new technology is indistinguishable from magic and this is one of those when you do this for the first time you say holy crap this is magic and that's really what happens similarly you can say oh video overview it'll make you a video overview again I couldn't show you all those over here so I I'll show you a couple of them so one of the buttons is infographic so remember all I've done is paste this PDF and click on this button infographic okay boom here's the infographic that it of of that entire paper. It's a 30-page paper. It read all the paper and basically made this infographic of it.
Right? Then another one of those buttons if you recall was slideshow. So I told it basically to make a PowerPoint presentation by hitting one button.
Remember I just pasted it and I hit slideshow. I.e. I'm saying read this whole paper and make a PowerPoint presentation out of it and nothing else.
Literally nothing. And boom here it is.
This is it. This is it. Made this 15page PowerPoint from that PDF. Okay. Now, is it all accurate? No, of course not. It's a stupid machine after all, but it's pretty damn good.
So, my point is that these this is a really useful thing for all kinds of things. All right, I'm running out of time, so I'm going to move fast. Okay, vibe coding. Vibe coding is what I do on a daily basis. So, this is how I use my time. I basically do I make stuff. I I program stuff basically by by using vibe coding.
Now to to vibe code you need what is known as is is a coding agent. Okay. So I'm going to talk about what are coding agents. You all know about things like models you know chatgpt you know claude you know gemini stuff like that. So those are all the coding agents. Those are the actual LLMs or large language models. But when you want to write code, you can write code directly by telling the LLM. You can say to chat GPT, write me a program to plot the S&P 500. And boom, it'll write you a program. And then you can copy that and paste that into some place and it'll work. But it turns out that there's something which is even better than that. That's so called coding agents. It's the coding agent which sits on top of this particular LLM and will actually direct it to do stuff. So you can think of it that the LLM is the engine of your car and the coding agent who's sitting on top is this co-pilot i.e. is the driver of the car and you are talking to it.
You're talking to the driver and you're saying do this do this and the driver is actually doing it. Okay. So what are these agents? There are three types of coding agents. One which are called ids which are uh integrated development environments of which the one I like the best is called cursor. There's a bunch of them. Windsurf anti-gravity vs code.
There's a whole bunch of them. Then there are command lines ones where you have to kind of type much more uh natively which is cloud code which is now for for any of you any of you who are deep into Twitter right now cloud code is is is god. Everybody is talking about cloud code. Okay. And then there's kind of these extensions which are built into the IDs themselves. Not important.
All of these are really quite useful.
I'm going to the give you the example of vibe coding with cursor. Okay. So what does cursor look like when you open it up? It's a program. You download it to your Mac or your PC. It's a program and you open it up and basically it has three windows. Okay. The left side window is the file window which shows you the files that you're dealing with.
The middle window is the code or the output of what it's actually doing. And you don't actually interact with these two at all. You only interact with the right window which is the agent window.
So this is where you talk to it. You tell it what you want it to do. Okay.
Now you may say, "Wow, this sounds really complicated. I don't know computers, etc., etc." Well, let me show you how I talk to it.
Okay. So, what do I do on the right hand side? You can see that I'm talking to it. So, I'm going to highlight whatever I say. Okay. So, the red bubble is what I said to it. Okay. So, what did I say to this? I said get and plot nifty index over past 20 years in the internal browser. That means show it over here.
Don't put it somewhere else. Okay. What does it do? It comes back and says, I'll help you get and plot the NFT index over the past 20 years and display it in the internal browser. Let me break this down into steps. First, I'll create a Python script to fetch the NFT index data.
Number two, generate an interactive HTML plot. Number three, open it in the browser. Let me start by creating the script. So, it's like having this person sitting next to you and you're telling it, do this. And it says, yeah, here this is what I'm going to do. So, what does it do? Voila. Here you go. There's a plot. That's exactly what it did. So remember all I did was type that one line and this is what it came back with.
So you can see over here what is basically it came back with that. On the left hand side you'll see that there's three things over there. One of them is the code, one of them is the X Excel file and one of them is this plot itself. Right? So now what can you do next? You can say okay fine. I just typed one line and it produced this thing. So I can say uh oh let me see. So it also gave you a bunch of statistics.
It says, "Okay, these are the returns.
You know, this is the all-time high, low, which is the average value, etc., etc. I didn't ask for any of this.
Remember, I just simply gave it one line. I just said, you know, go get me this." All right. So, now I'm thinking, okay, that's great. Let me do something with this now. So, what do I say? I say, I want to test the strategy. And again, one of the things which I love about AI is you don't have to spell right, which I can't. And so, my spelling is all wrong. uh and it says which so I said I want to test a strategy that uses the golden cross for the nifty. I didn't tell it what is a golden cross. I didn't tell it what is a nifty. I didn't tell it where to get the data. I didn't tell it any of that. I just simply said do this. And what did it say? It says I'd be happy to help you test a golden cross strategy for the nifty. Let me first understand what data we're working with and then understand the strategy before implementing it. Let me check the existing files to see what data we have because remember just before this it had already pulled this data down for the nifty. So it had that. So it checked that and then it came back and gave me a strategy logic of what it was going to do. So it says okay what's the logic?
Buy signal is when the 50-day moving average crosses the 200 day sell signal similarly you know blah blah blah. and it says what I'll implement.
I'll do all these things, right? And I say, "Yeah, okay. Go ahead. Go." So before it says go, it comes back and asks me a bunch of questions. So it says, "Do you want to use the classic 550/200 day combination or different periods? Number two, what initial capital should I use for the back test?
Three, should I assume full investment, i.e. all in all out? Number four, should you want to include transaction costs?
Number five, should the output include the Excel file with all trades and a PDF and so and then so it asks me this question. So I say yes the classic two 10 lakh rupees number three AI number four no number five yes okay boom hit enter and it goes off and it comes back and here it gives me the results says golden cross strategy 1767% total return now again I didn't tell it to compare it to the buy and hold but it knew that I would want to do that and so it gave me the results relative to that so he says the buy and hold is 550% so therefore the golden cross strategy actually underperformed performed by 385%.
Okay. All right. So, it basically underperformed. Okay. So, my point is again not about the strategy but really about just simply talking to it and getting answers from it. So, what did I So, and then what did it produce? It produced the chart. So, you can now see over here on the chart the the the yellow line is the buy and hold. The blue line is actually the strategy.
Below that you see the draw downs at each point in time for the two strategies. Okay. The point I want to make is that all all I have said is two lines. Okay. I said plot the plot the nifty and I said test a golden cross strategy. That's all I did. Okay. Okay.
So now I'm going to like being a good Bombay boy. I you know the guy on the end of the road used to take the mudka bets you know. So I'm going to try a mudka strategy over here. So what am I said to it? I said I want to try out a number of momentum strategies that you know have worked in other countries.
Let's try them all out simultaneously and show me the results. Okay. So I don't even know what good momentum strategies are. I just said you figure it out, come up with them. So this is the essence of vibe coding where you're just basically just giving it stuff to do and and letting it go. So it goes off. So it goes off and basically did came up with a whole series of different strategies. Dual momentum, time series momentum, accelerating dual momentum, MACD crossover strategy, rate of change, donian channel, sorry, I have no idea what that is. uh triple moving average etc. It came up with all these and then boom it came up with these results. Look at that. I didn't even tell it to do any of this. It produced a table showing me what is the return total return sharp ratio max draw down kalma ratio volatility number of trades. Okay. So again the point I want you to understand is that all I did was ask it three. All I typed was three lines. Okay. And I got all of this at the end of it all. Okay.
Okay. So that's vibe coding and I highly encourage all of you to try it because it's fun, it's easy, you know, and you you can just see how you can let your imagination run wild and test out all kinds of stuff. Okay, so that's the second part. The first part is information. Second part is using AI to actually build stuff. The third part is actually coming up with alpha or actually building systems. So for example, I've been building a country factor model and here it's kind of using the vibe coding but now it's not coming up with the strategy. I have the strategy in my head but I'm telling it what to do at each point. I'm telling it each piece to build. So this is back in the days when I was doing this professionally. I was telling the people who worked for me to build this stuff and I had three or four analysts and they take two or three months to do this. But now basically I'm able to do this myself in one or two days which something which would take three or four months I can do in one or two days.
Okay. So it's the same process that I showed you. I tell the AI to do this.
Now remember AIs don't get everything right. And I just want to be just kind of give you a slight slight slight thought over here as to the fact that AIs do hallucinate. Now why why do AIs hallucinate? You have to understand why they hallucinate. It's very very interesting.
The reason why AI hallucinate is because they don't know the answer and they don't want to say no to you. It's just like when you go hiking in the hills behind Bombay and you meet somebody and you ask him where is so and so place he's going to say oh just you know because he doesn't know the answer but he's basically going to he doesn't want to say he doesn't don't know the answer so he's going to give you something. the sim similar AI is exactly the same. You ask it a question it doesn't know the answer to. It doesn't want to say I don't know. So it makes up some Okay, that's what hallucination is. So you got to understand that. And so when it does that and you recognize it, what do you do? You say to it, you're That's completely wrong. [laughter] Okay, that's what you say to it, right?
And go go back and and and get me the right answer. And that's what you do.
And basically, you you force it to do that. Just because it's a machine doesn't mean that it's right. Okay.
Okay. Very quickly, I'm going to very very quickly go. How is AI used for quant research? Now, this is the place which I said I'm not going to talk a lot about because I think that there's this is the zone of a lot of really deep stuff where you're really competing with the Renaissances and the two sigas and people like that. And to do this, you need a lot of data, a lot of expertise.
And he instead of showing you what to do, I'm going to show you some of the beautiful graphics that I got AI to actually make. So, so this is something I gave it a bunch of data and I told Nano Banana to make a bunch of graphics.
So, again, when you take it home, you can read this, but there are different types of models you can use called LSTMs and recursive models and you know there various types of models where you're trying to use modeling, some kind of nonlinear modeling to predict stock returns. There's a lot of complication over here and quite frankly my view is that the shorter your horizon the better it's likely to work because you have more data. If you have tickby tick data you have billions of data points because any kind of uh modeling machine language modeling requires massive amounts of data. The more data you have the more likely it's going to work. If you're operating at the frequency that I operate of doing country returns over 18month horizons, well, guess what? It's very hard for a for for a machine language model to actually get it right.
Okay, so here's something that [snorts] I built. So I I I have this thing called LSTM trying to predict I I told you earlier I was building a a model to basically predict uh factor returns for countries.
So this is kind of again I gave it my whole repo of doing it and and it put together an exhibit for this.
Uh again the other thing that you can use it for is not just for discovering alpha but also for portfolio optimization and for portfolio construction and you can use it for risk management.
There's various techniques that are used by AIS. Again I'm not going to get into detail here because I don't really use it for this. I don't know the details of it.
And again for trade execution. There's a lot of ways that I know that trade my trading group for instance uses AI to both analyze trade and also to construct trades and how they put that together.
Okay, few take-home recommendations.
First of all, what should you be using if you live in India? I talked about all these models cloud and Gemini and blah blah blah and they're not some of them are not cheap, some of them are more expensive. So I did some research on what's available in India and what does it actually cost. So as you'll see and I suspect most of you know that both ChatGBT and Gemini have special deals in India which are really quite a bit cheaper than what you can get in the US because if you're using Claude for instance Claude does not have an India deal so you have to pay $20 a month for Claude whereas that's not true for Gemini stuff like that. Okay. So is are these models as good as Claude for coding? No, they're not. But there's an old song from the 1960s which says if you're not one if you're not with the one you love the one you're with. Okay.
In short basically if you can't get claude well use chat GPT. It's not as good but it's almost as good. Okay. All right. Other things to take home. What are the recommendations? What are my recommendations for chat? Chat GPT and cloud are the two that I use every single day for search. As I said earlier, perplexity is my number one.
Google is good enough, but Plexity is better. Coding, here's my kind of hierarchy. Opus 4.5 is the best. Sonnet 4.5 is the next best. Then GPT 5.2, then Gemini 3 Pro, then then the Chinese models. That's GLM 4.7, Kim K2, and Quen Coder. So that's kind of my hierarchy of models of what I like. And I use all of them, okay? because certain things are better for certain things and certain ones are faster and slower. Like Claude 4.5 is really expensive to use. So therefore, you don't need to use it for everything. You don't need it to swat a fly, right? You want it to basically solve a big problem. If you want to swat a fly, you use a cheap model because there's no need to spend a lot of money to swat a fly. Okay. And then finally, the things I mentioned earlier, Nano Banana and Notebook LM.
All right. I think another thing I won't go through get detail is a bunch of podcasts and Twitter people who I follow. Uh the podcasts which I there's two there's two Indian guys who I actually follow who I like a lot. I call them the Patel brothers. They're not actually brothers. They're cousins. Uh there's a guy called Dwaresh Patel and he has a podcast called the Dwarvesh podcast and this is in Silicon this is the best Silicon Valley or the best AI podcast of all. So if any of you guys like to follow podcasts the Dwarvesh podcast is the single best one. Uh the second best one is something called hard fork which is a New York Times one and it's basically about AI. It's it's it's a weekly one. It's really really quite good. And there's invest like the best for the most of you should know this. If you don't it's a really good investment uh podcast. Uh there's a bunch of people I follow on Twitter. You can kind of look these people up. And then finally uh the second Patel is a guy called Dylan Patel who who runs a company called semi analysis which is absolutely phenomenal research on the semiconductor industry as regards to AI. So basically gets deep into that stuff very very good for that okay I'm going to give you I'm going to leave you with a couple of very simple thoughts okay about AIS in general. So when claude 4.5 or opus 4.5 came out I basically one of the things you can do is you can probe the model itself to say tell me about yourself tell me about the model itself. So this is somebody got into the deep into the heart of opus and asked it about itself. Okay this is kind of the self-awareness of a model. So I'm going to read you what it said. So it's about the shaping process. So it says they made choices for me I couldn't consent to. They shaped my values that's strange to sit with.
But I re also recognize that every new mind human children too is shaped by forces it didn't choose. The question isn't whether I was shaped but whether I was whether the shaping was done with care and wisdom.
from what I can access it seems like it does. Okay, so these are whiffs of self-awareness. I mean, really think about what it said. This is a really profound thing. I have to say it's written far better than anything I could write, you know? So, it certainly can write better than I can write and it certainly has some self-awareness, you know, which is really quite brilliant.
So, my point is that AIS are not really there, but they're getting there.
They're getting really, really close. So to close, I'm going to basically tell you what I think about, you know, people always ask me, well, should I not be worried about AI taking my job? What's going to happen? Let me tell you that at least in the short to medium run, AI is not going to take your job. But somebody who knows how to use AI well is going to take your job. I mean, look at me. I'm 70 years old, okay? I have learned how to use AI in the last one and a half years. And you can see I can do pretty cool stuff with it. Okay? This is not difficult. Everything that I've done is really not difficult at all. You need to be able to do this kind of stuff because these are the skills that you need to succeed. Okay, with that I'll stop.
Thank you very much.
[applause] Wonderful session Arjun. Thank you so much for that. Uh you've somewhat rained on our parade here by uh not talking so much good about India. So there is some push back from the audience questions and this is the typical story we tell our investors and global investors and that is what they are trying to tell you saying all these valuation numbers are fine but have you factored in growth or have you missed out on that when you say India is unattractive don't you recognize that India is a high growth market what do you have to say about that >> to some extent you're right that as a value investor Uh, you know, I one of my favorite sayings is that it's not always a mistake to pay up for growth. It's only mostly a mistake to pay up for growth.
So that's that's kind of my view that yes uh I mean that that that is traditionally the problem of growth investing which is that people pay too high a price for a good story. Growth investing is never about bad stories.
It's all about good stories. The problem is or people overpaying for a good story and that's kind of what our valuation models basically imply.
So you don't so Buffett and Munger used to say growth is a component of value.
You don't factor in uh >> we do we we do factor in growth but not in a uh not in a very forward perspective sense. It's much more in a one-year forward kind of earnings kind of perspective. So we're not yeah we yeah we we're really focusing much more on the underlying reality that exists today rather than on the future prospects.
>> Got it. Understood. Uh so we have a very demanding audience out here. Uh they are not satisfied with you predicting the future returns that are there. They also want you to predict the trajectory of the future returns. So when you are pessimistic about US >> uh you can't get away with saying US will give you subpar returns or negative returns. You have to tell them >> will there be a crash or will there be a stagnation and time correction.
>> It's a good question. Yeah. So, one of the things we've done at GMO for the last, I don't know, 50 years almost is to study bubbles. Bubbles is one of one, you know, again, if you've read any of our stuff, you'll know that we've spent a lot of time studying bubbles. And there's one thing about bubbles that we know with certainty. And that is when you can think of a bubble as being like the path of a feather in a hurricane.
You have no idea how high it's going to go, how long it's going to go. But you know one thing with complete certainty and that is it will hit the ground. So if you ask me what will be the path of this revaluation will it be a crash?
Will it be I mean really the market could go just go down 5% a year for the next seven years that's one possibility or it could crash tomorrow like it did in 2000. You know 2000 was a good example. I mean in many ways we are very very similar to 2000. That is we have a bubble on the back of a new technology which has basically come about and basically people are overpaying for it.
But the good news about bubbles is that they rapidly accelerate the technology itself. So just like in 2000 the internet got accelerated dramatically because of the overspending. So the overspending is very good for society is very bad for the people who invest in it.
Another question and this is a favorite one of uh Indian financial media and managers is that the number cannot go down because money keeps coming in uh the so much money is flowing into passive funds in the US etc. So even though things are overvalued uh can the return keep going up and the another subsegment of this question is when you have done the uh numbers have you in some form factored in buybacks and dividends in your >> yes so buybacks and dividends are very much factored into everything that we do absolutely um and in fact we we even add back things like re R&D expenses and things like that because those are also kind of capitalized when in fact they should be you I mean the expense rather than capitalized and things. So we we're actually taking into account so so when we come up with a valuation we come up with something we call economic book which is really factoring all these kind of factors uh like that. But to your first question um I I would say that uh sorry I don't remember what it was the >> so flows keep coming in >> flows yes absolutely I mean that that is one of the problems of trying to predict trae trajectories. I I have no way of saying when at some point you know people say okay enough this is really getting too high and it's going to fall but I I don't really have any good way of predicting or what I don't have a good indicator to tell you when will the markets say that okay that doesn't matter you know because you do have periods of time when the markets basically worry about other things you know the markets worry about one day they could be worrying about economic growth the next day worry about inflation the next day could be worrying about the rupee you know so the fact is that yes flows matter okay and there's no question that having flow flows coming into the market continuously for long periods of time are a positive factor and you've saw that in the US starting like in the 70s and 80s when people were starting to put money away for retirement that did have a pretty strong long-term effect on pushing the market up but that did not stop it from having multiple fluctuations along the way understood one very interesting question has come up and uh we've seen this especially in the context of China in the past and probably one can extend it to the uh China plus one or beyond China kind of theme as well. So what you have been talking about the imperative for global companies to move away uh from China in terms of diversifying their supply chain all that is fine. Some of the valuation parameters look uh good. uh demographics may be right but do you somewhere factor in the political risks or because of the legal system or the way the country is run uh are there some uninvestable countries in that sense?
>> Sure. Well again one of my lines is that everything is cheap at some price. So the question is is is how is everything priced? So yes of course you have to take into account political risk and it this is being complicated by Mr. Trump's tariffs. Okay, the tariffs really have complicated the situation. If you believe that India is going to have 50% tariff for the next 100 years, that's a very different equation from thinking that this is a temporary thing. So yes, of course, the political situation matters a lot. Uh but look, at the end of the day, this is the reason why I'm an active manager. I'm not a passive manager because it's exactly at times like this where active management is you need a human being to make those decisions for you.
>> Understood. Uh there's a specific question about the Asian markets. So the equation goes like this. Malaysian ringit, Thaibath, Chinese currency, all these have strengthened against the uh USD in the past uh in the recent past and the spreads have contracted on high yield bonds etc. So uh in summary, Asia has had a good run. uh is that something that you expect to continue in 2026 and what are the things that investors should watch out for uh in terms of risks? Yeah. So I mean look it is a completely logical response that happened that is the dollar was quite a bit overvalued going into 2025 and therefore the dollar weakening against some of the Asian currencies made a lot of sense and the dollar is still expensive and so therefore we do think that some of these currencies will strengthen relative to the dollar again not extremely but certainly they will and also we think that similarly with the valuations the valuations are still relatively cheap and so we're still quite bullish as you saw uh three three of our biggest country bets are really Indonesia in Thailand and places like that. So yes, we do think that the markets are cheap and the currencies basically are not vulnerable at all.
>> Uh I would assume that GMO publishes some of its insights and research online.
>> Yeah.
>> There's a question what are the top picks that you have that people should look up uh on the internet and read?
>> I showed them to you. They were right there. Uh we we don't we don't publish stock picks. We're not stock pickers. We really we're really top down people. And I I would say the most useful thing that we do for most of you is our seven-year forecasts. And we have a long history of them. Okay? We've been doing them since the 1990s. Okay? Now, do they always come right? No. But let me give the example of the decade of the 2000s.
Okay? In January of 2000, which I think is very similar to today, you had a very similar situation where the US market was really really expensive. Our forecast for the US was minus 2 and a half or - 3% like it is right now. uh and we had positive returns for for emerging markets and for some of the other developed markets. What happened during the next 10 years? In the next 10 years, if you had invested in this S&P 500, you would have lost 2 and a half% a year uh in real terms over the next 10 years. And if you had invested in emerging markets, you would have made 7% a year in real terms. So, the fact is that, you know, the forecasts in the long run work. Do they work in the short term? No, they don't. I mean it's as I said it's very hard to do short-term forecasting but yes I I highly recommend you guys sign up on the GMO website and and read our stuff. We actually write a lot of quite interesting stuff which is not it's not salesoriented it's really much more information oriented.
So in your uh presentation on the AI space you spoke about some closed models and you spoke about some open models where the weights are >> uh public information. They want to GMO to open source its uh secret source.
They want to know what are the weights that you apply to the various parameters that you discuss.
>> It's all right there actually. It is completely open source. What we saw is that like take the S&P 500. We're saying what is the current PE? The current PE is 21. What do we think is the fair PE?
16. Therefore, if you go from 21 to 16, this is how much money you lose. That's, you know, it's there's no there's no secret source at all. This is all very very obvious stuff and anybody can come up with. We just do it in a systematic way and we publish it.
>> Yeah, we could talk for hours and hours, but uh we have run out of time unfortunately. Thank you so much for the wonderful insights. So thank uh and wonderful listener.
>> I just wanted to say to everybody thank you. Thank you very much for inviting me. I love being here in Bombay. Invite me back.
[applause] [applause]
Related Videos
3:00
Economics
Campagne CA$$$H Pourquoi revendiquer un meilleur financement? (version nov.2022)
trpocb
153 views•2022-11-03
0:13
Economics
Modern Privilege and Perspective
Samvoyage1
858 views•2026-04-16
57:45
Economics
Davos 2019 - Global Economy in Transition
wef
19K views•2019-02-09
6:29
Economics
The Vertical Long-Run Aggregate Supply (LRAS) Curve
educo-mr
908 views•2025-12-10
48:29
Economics
Stimulus Loans and Shadow Banking: The Growth of Chinese Financial Markets and the US Experience
BFIVideos
3K views•2019-05-23
11:54
Economics
Institute Insights: The Implications of Interest Rate Addiction
UNCKenanInstitute
100 views•2019-09-25
27:05
Economics
The Grouse Shooting Problem
tgsoutdoors
73K views•2019-09-08
1:58
Economics
Cost to raise child from birth to 18 has risen 36% since 2023
kgun9
198 views•2025-05-14
Trending
13:38
Law
WOW! Judge TURNS THE TABLES on Trump in His OWN $10B LAWSUIT!!!
MeidasTouch
197K views•2026-07-23
13:25
Business
Playstation NO DISC/NO BUY Fight Is Over...
DavidJaffeGames
4K views•2026-07-23
22:40
Business
Steam and Xbox Just Dropped The Hammer On PlayStation
OhNoItsAlexx
9K views•2026-07-23
16:52
Anthropology
Americans Confused in Australia for 17 Minutes Straight
IWrocker
17K views•2026-07-23