Trading Tomorrow - Navigating Trends in Capital Markets

Unlocking AI's Power in Finance with Fawaz Chaudhry

• Numerix • Season 3 • Episode 31

Share episode

In this episode of Trading Tomorrow, we explore the groundbreaking ways artificial intelligence is reshaping finance with insights from Fawaz Chaudhry, the Head of Equities for Fulcrum Asset Management. Fawaz provides a rare look at how AI tools are harnessed to interpret complex data, streamline coding, and improve reporting in finance. We delve into the future of AI for pattern recognition in images and video, and Fawaz shares the impact of hardware advances on AI's capabilities in finance. Tune in to discover how AI affects productivity, market efficiency, and the future of portfolio construction.

Speaker 1:

Welcome to Trading Tomorrow Navigating Trends in Capital Markets the podcast where we deep dive into technologies reshaping the world of capital markets. I'm your host, jim Jockle, a veteran of the finance industry with a passion for the complexities of financial technologies and market trends. In each episode, we'll explore the cutting-edge trends, tools and strategies driving today's financial landscapes and paving the way for the future. With the finance industry at a pivotal point, influenced by groundbreaking innovations, it's more crucial than ever to understand how these technological advancements interact with market dynamics. Today, we're discussing artificial intelligence and its evolving role in finance, particularly within equity markets. Ai is reshaping everything from now casting to portfolio management and beyond, but how much of this revolution is transformative and how much is overhyped? To explore this, we're thrilled to have Fawaz Chowdhury, the head of equities at Fulcrum Asset Management, with us today.

Speaker 1:

With over $8 billion in assets under management, fulcrum has been at the forefront of integrating advanced AI tools into financial analysis, helping to refine investment strategies and even collaborating with central banks. Fawaz Chowdhury joined Fulcrum in 2017 as head of equities and is responsible for building the firm's equity capability. A key member of the broader investment team, fawaz is also the lead portfolio manager for the thematic and climate solution strategies. Prior to joining Fulcrum, fawaz spent over 15 years developing his long-term thematic investment approach, including at Hadron Capital and More Capital. Welcome to the podcast, forwaz. Thank you so much for having me, so you know. Let's just start with your basic pulse on AI tools within the industry. What have you noticed in terms of usage?

Speaker 2:

Well, I mean, the AI tools have been used to create algorithms for trading black boxes, but in a very hesitant way, because you don't actually know what the black box is spitting out. There's dangers of hallucination, as they call it, and the results have been mixed. So on that sense, I would say progress is still limited and halting in general from industry. What I have observed more in terms of gathering and interpreting data, in terms of gathering and interpreting data in terms of where the AI is best suited is a language. So, in terms of gathering investor reports, gathering knowledge from transcripts, gathering generally lots of reading, all of the transcripts that are written in language that can be summarized put together, I think that's where we've seen a lot of progress now and it's continuing at this stage and at Fulcrum, you're using AI to write a little bit of code.

Speaker 1:

Perhaps you can explain that a little bit further.

Speaker 2:

Yeah, I mean code goes asset management. In every industry you write code now and obviously so do in asset management. We write code for tools for monitoring, for risk, for Implementing trade there, obviously systematic trading and even in discretionary trading. You also use to risk management tools. So everyone is well aware of the adoption of large language models for increasing productivity, for producing very high degree of code.

Speaker 2:

We see that the data as an investor in ai companies and, because I'm a equity investor, we see one of the best and early adopters of productivity has been in code writing and you see it in the github usage and so, yes, so do we. In my team I have analysts who are writing code and tools and scripts and they are more productive than they used to be. So I think this is a general usage across all producing software and that's a trend that will continue.

Speaker 1:

So you know in terms of you know your opinion, your experience, whether it's large language models, writing for code, you know. If you had to create two very clear buckets at this moment in time, because everything is changing regularly, you know what would you say AI is most useful for and what is it not useful for.

Speaker 2:

I mean, the most useful part is obvious, which is language, and it's just that producing written reports is, while it is productive, like we also I mentioned uh gathering, reading transcripts, etc. But a low level work would could be, for example, writing our monthly reports and etc. So what we used to spend time on uh perfecting the language gets fitting it in a certain fact format in the fact sheet. That is still time saved if the while we our first rough draft can then be perfected in minutes, not hours or even a day. So even we are using it. But language in general, the high level work in language is coding. You are paying graduates from stanford two hundred thousand dollars a year, right?

Speaker 2:

now in silicon valley to write code, or or so you are actually saying, hey, this language work, which is what coding is. It's a software language is where you see some of the highest productivity. But language in general because this is why they call large language models, and AI is good at identifying patterns and trends and then transferring output to match that. And then because we have so much written language in terms of the entire web of written language, ai is able to do that. In the future, ai will be doing other pattern recognitions. They're doing that now for image creation and video. It's not as good yet as language, but it will keep getting there identifying patterns for video and image, but it will keep going. Anything that you have to identify patterns and reproduce AI is going to be very good at.

Speaker 1:

We've all seen AI generating images of dogs with four legs or five legs.

Speaker 2:

So it's getting there, yeah it's not as good as language, but you will not see that, so let me explain.

Speaker 2:

You do not currently notice how much for example, all the journalists are using AI to write an article, for example, from the first draft to the final product, how much? How they could write an article now, maybe in two days, not a week, because they don't have to perfect it the same way, because the AI can do it for them. So you don't see that, because the end product is that good that you don't see that. So the image and video are not there yet, but it will get there. Where you don't see, you won't be like seeing how much AI has been used because the end product is that good.

Speaker 1:

Well, you know, previously you've mentioned that AI and macro trading is very overblown. You know. Perhaps you can elaborate on why you believe its impact has been limited so far in this way.

Speaker 2:

Well, macro trading I mean. The more it's about data, then again, the point is, data can break. Data and macro trading, by its inherent nature, is more short term, so you are reacting quickly to market events. And then, whether it's inflation or economic data or etc. Or the NFP report, employment data, so you're reacting quickly to it. So, long-term trends and recognizing the pattern in long-term trends and repeating them becomes a bit harder when everything is more short-term and more unique. And then once you get a spurious data point like hallucinate, the AI model can hallucinate, you could put the trade in the wrong direction, and et cetera. So it's a bit harder.

Speaker 2:

In one area that I've seen, macro, in which AI is now starting to be used, is again coming to language reading the language of, for example, all FOMC members or all ECB members or all within ECB. There's all these different central banks at Spain, italy, et cetera. All of them are putting out some transcripts. No one has time to read every Spanish central bank's interview launch transcript, but the AI can read all of it and see whether it's are they leaning more hawkish or leaning more dovish. Again, it comes back to language and how the language is used. So language is where we can still see, and obviously Fed speeches matter. So that's where language is important, that's where it can help. Macro Data lessons.

Speaker 1:

You know you make a really interesting point as it relates to, like, fed speeches. But you know, I would also argue sometimes it's what's either not said or the way it's said. In that nuance I've seen move markets as well. I mean, how are you? You know it's wonderful to have the interpretation of, you know, all the central banks, but at the same time, how are you interpreting the nuance on the front lines?

Speaker 2:

Well, ai is actually decent at it. So if you ask AI, is it more dovish or more hawkish than the same members or the previous 10 speeches? It can pick up on what he is not saying and hence, is leaning more dovish or leaning more hawkish. As long as it pertains to language, ai is actually decent at it. That's what I'm trying to say, and you could see all of the FOMC members are out there making speeches. Every week, you get 12 of these, and so you can actually now even and all the banks are now doing it Bloomberg is now doing it Everyone is now doing natural language processing using AI and giving you a sense of each of the members' position, turning dovish or hawkish, based on their recent speeches compared to their own previous speeches. So I think, in terms of language, ai is quite advanced. I would not say that you can pick up some nuance and the AI is missing that nuance.

Speaker 1:

Wow. So you know. I guess another kind of follow-up question in that regard is you know what kind of skills you know are the teams developing now on the front lines in terms of prompt engineering to be able to extract that nuance from these models?

Speaker 2:

Well, what the example I just gave, and in some sense the most important example, which is Fed, and whether the FOMC is leaning, dovish or hawkish, because that moves markets, it's the prompt engineering required lean, dovish or lean hawkish and to prompt it. What those words mean, ai can even understand what those words mean. So actually it doesn't take too much. Feed them the speeches, ai will be able to understand what you mean by dovish and hawkish Because, again, that's part of language.

Speaker 2:

So it's actually just about getting the fiddling with it and getting it right. I would say work is being done and they are expanding it to be able to get the nuance correctly and put it on some scale and et cetera, how, how, how, and so you can actually get some useful information out of it. So that's where the prompt engineering put it on a, assign a number to it and put it, chart it over time, so the dovishness and hawkiness can be put a quantity on that and then chart it over time in a graph and stuff like that. That's the kind of prompt engineering you're referring to. That's what I've seen produced by these various banks and Bloomberg and others.

Speaker 1:

And has AI reshaped your views on optimal portfolio construction? And you know? If yes, why?

Speaker 2:

I would say optimal portfolio construction theory is not being challenged by AI. I think AI can get news into the market quicker, get it priced quicker. It's a continuation of the pattern that we have seen over the last two decades. It's not that machine learning wasn't happening before. What was machine learning we call AI now, I would say in terms of academic rigor and now the portfolio construction, academic rigor has not been challenged.

Speaker 2:

And the balanced portfolio of optimal frontier, etc. All of that is not being challenged necessarily by ai. It's just that ai is just another tool which recognizes patents and gets that information. Even, let's say, fomc I'm giving the example of now if someone starts running a strategy that, as the Fed, is leaning more dovish, sell dollar. If it's leaning hawkish, buy dollar. Or I mean if people are running AI, live AI, as the the FOMC member is making a live speech, ai is assigning a quantity on the dovishness and hawkishness live over time it is translating into the market quicker. So all I'm saying is what it does is it takes more of the information and gets that priced in into security prices quicker, makes markets more efficient. So, to the extent that the academic rigor required, an efficient market hypothesis, yeah, is just actually making market moves more toward the academic rigor of efficient market hypothesis, not the markets being semi-efficient or inefficient. Ai is making markets more efficient.

Speaker 1:

And you've said that Fulcrum avoids black box AI systems due to fiduciary responsibilities. Obviously, you know EU regulation that's coming out is, you know, very concerned around the black box concepts with AI models. Perhaps you can explain a bit further how your team has come to this decision and is implementing that.

Speaker 2:

Well, I mean in general, about any strategy that you put forward, systematic you should have a very high degree of understanding of what those models are, how those models are constructed, how they're implemented, how they react to different market conditions. The more black box it is, the less of a risk you can assign to it, less of the allocation you can do to it. And neural networks, by their design, are constructed to be black box. In effect, it's very difficult to understand how, because of the number of nodes and the number of hidden layers and the way it propagates. And just as a bit of background, I did my bachelor's and master's at MIT in computer science, with a master's thesis in neural networks at MIT AI Lab.

Speaker 2:

So this is the same thing we were doing even two decades ago, and I understand that you cannot actually determine what the driver behind each of the decision making was. It's a black box, it spits it out. Behind each of the decision making was. It's a black box. It spits it out and there is some something it saw in the pattern that and it could have been a spurious correlation that it assigned a lot of weight to and then ultimately gave you an answer, which is what we call hallucination and etc. It. It is some pattern that it saw that it didn't really have a causality behind it and ultimately we basically say it's hallucinating, but ultimately you cannot then run a large amount of risk and large amount of your assets dedicated to such a strategy which can hallucinate, but you cannot explain why the decision it did, why you cannot. So even other asset managers are in the same kind of conundrum and then hence they don't.

Speaker 1:

What is your view on the future of AI and coding and automating complex financial processes?

Speaker 2:

I am extremely bullish and the largest risk in my portfolio is AI-related businesses and has been for a year and a half and continues to be, despite the July-August pullback in these stocks. I think adoption rate is going to surprise everyone to the upside and will continue to be extremely robust, and we see at the moment the market being supply constrained with not enough hardware available. I gave you a very simple example that I mean, do we really need to pay people so much for writing code when it's just language? Yes, writing code has a high amount of value add, but it's still cheaper to buy a graphics card. So it's going to be the case where the code will be written by the hardware, and I think we are at a seminal moment and these kind of pivots happen in the technology hardware. And I think we are at a seminal moment and these kind of pivots happen in the technology landscape.

Speaker 1:

And we're about to put the silicon back in Silicon Valley, basically. So you know, as someone who's spent time at MIT made a decision as it related to technology 20 years ago. Your words, not mine. I'm not trying to age you here, but I guess you know. What would your advice to young people be as they look at their education? Yes, we might be paying 200K per programmer now, but if these skills are being automated, perhaps how should younger people be thinking about the evolution of the industry?

Speaker 2:

Yeah, one thing, for example, in pattern recognition, something like radiology. Even 10 years ago everyone was telling me it's going to die. And we're now at that point that AI can do radiology and detect stuff that doctors can't, because it sees patterns across them. Having been trained on millions of x-rays, it can do better. We are going to anything that that ai will threaten. Those jobs will be at risk.

Speaker 2:

And I am saying software, writing, coding is something that will become commoditized or much cheaper in the future, which means software itself will become much cheaper in the future and more of the value in the tech stack will go towards hardware, not software. So software has already eaten everyone's lunch. Those predictions from 15 years ago have come true for Mark Henderson, and now it's time to give some back. Software ate too much. We're getting indigestion from software. Software ain't too much. We're getting indigestion from software. So it's I would say. I mentioned MIT. Mit actually has one department for which they call Department of Electrical Engineering and Computer Science. So even my bachelor's and my master's degrees were called. In Elect engineering and computer science they say course six. So they think of electrical engineering and computer science as the same department, which across a spectrum. So either you're more in hardware and transistors and the silicon and doing hardware level coding, or you're more in the software which is an upper layers, like the base layer versus a operating system layer versus this app layer and etc.

Speaker 2:

So you keep going upper or you keep going lower base layer. So we are now in the world where we're going back to the chip. Now you have to produce the better silicon so then we can get more productivity out of it. The large language models are, in my opinion, getting commoditized pretty quick. Meta's model Lama is open source, and others as well, and the way they make them better is adding more layers, adding more nodes, basically more hardware. All of which requires more hardware to train them more hardware to run the inference on them.

Speaker 2:

So everyone is desperate to get their hands on more hardware. So the software better AI model is basically just get your hands on better hardware. So, yeah, get into hardware. That's my advice to the young kids.

Speaker 1:

You make me think you know going back. Maybe you know let's call it eight years. You know back when NVIDIA came out with CUDA and you started having all these debates around. You know C++ and you know getting and moving over to GPU versus CPU. It seems now, if AI was where it is today, back then maybe we would have seen even more rapid movement and sophistication around GPUs.

Speaker 2:

Yeah, I mean, ultimately market did not invest enough in it. Okay, we had Google who decided to develop its own accelerator called Tensor Processing Unit, tpu. So Google did a lot of investment into it, which kept NVIDIA honest. But AMD fell behind with their competitor ATI. They bought ATI. It was NVIDIA and ATI before and AMD fell behind.

Speaker 2:

It was more interested in the server market, competing with Intel, which they've been winning on, and let this accelerator market go. And there were not enough people trying to keep NVIDIA honest. Nvidia kept on producing on a certain speed, but if people appreciated that, hey, you should be investing more into an accelerator technology. If Microsoft or Amazon were doing more, it would have spurred more innovation in it and we could have had better and cheaper accelerators now already, and we would already be at a pace where we're not paying software programmers so much because we would be getting code produced more cheaper already. So anyway, it's, it's all happened. It took a breakthrough technology like chat, gpt to come along, to get wake people up to the potential, and once you get the breakthrough technology, we're're on that path now no stopping.

Speaker 1:

Well, as a CMO of a software company, I'm a little nervous by this conversation. So you know. Lastly, you know there's a lot of excitement around AI infrastructure companies, as you know potential market winners, you know. What role do you think AI infrastructure will play in the financial sector and how should investors position themselves in this space?

Speaker 2:

I mean investors should be not fighting the trend, in my opinion at all. I think when I look at every software company right now Salesforcecom, they're coming up with their agent AI agent OK, sounds like they're buying some NVIDIA cards. Or ServiceNow, they've got any. Sounds like they're buying some NVIDIA. Every software company is now boasting how they have an AI version of their software. They are all becoming more CapEx intensive. That's what they're saying. Read between the lines they're all buying hardware. There's they. They are all going to. They are going to be companies that used to be very capex light. They used to hire programmers to create a new feature on their software and charge you more for it. Now they're all they're doing is buying more cards. They're all buying hardware. When they're offering you AI versions of their software, all they're saying is they're out there trying to get their hands on some hardware. So ultimately, all of these guys are going to become more CapEx intensive. More CapEx intensive businesses have lower returns, not negative. It is still above their cost of capital. They are right to do this. If they don't, someone else will and they will fall behind. It just means where the value is going is towards the hardware guys. The value is moving towards the hardware guys, the CapEx end. So guys like NVIDIA, guys like Arista Networks, broadcom, these are the guys like Micron, high Bandwidth Memory. These are the guys who are going to now be taking their piece of and obviously, international markets, asml, tsmc. These are the champions. They're going out there. Sk Hynix in Korea is a high bandwidth memory producer. These guys are now creating.

Speaker 2:

We have went through a 30-year hardware down cycle where basically the whole hard because of Moore's law, there was so much productivity gain in hardware that hardware got commoditized. You got this. You got better and better hardware for cheaper and cheaper and you got used to it. And we went from 20 memory makers to three and we went from 10 leading edge fabs to one leading edge fab, tsmc. We went from 20 graphics cards maker to one. So the point is the hardware industry consolidated because it had to, and now we need more hardware. They have pricing power and they're coming back. And yeah, and you can now produce a software competitor like that by getting some accelerator and telling it to write a code. Write me the forcecom platform. It will produce billions of lines of code and reproduce it for you. Wow.

Speaker 1:

You know it's funny. I think of some of the larger companies that have taken hits from the market, but I never thought about it in the context of that. They're becoming too CapEx intensive at this point, but we've seen that play out with the Facebooks and others over the past couple of months. That's a really interesting way of looking at it.

Speaker 2:

But they are CapEx. They will become more CapEx intensive and CapEx intensive businesses will have lower returns. Doesn't mean negative. Doesn't mean it's less than the cost of capital. They are still going to create value for the business by doing it, but in essence it's a free cash flow of all the software businesses being transferred to the hardware.

Speaker 1:

Wow, so sadly we've made it to the final question of this podcast and we call it the trend drop. It's like a desert island question, and if you could only watch or track one trend in AI and finance, what would it be?

Speaker 2:

The CapEx of the mega cap tech because they are the ones reselling. Microsoft is a reseller of AI infrastructure. It is building data centers, buying nuclear power capacity, buying NVIDIA graphics card and hopper system and reselling those flops. Aws is reselling those flops GCP, google Cloud. So the capex is increasing. Clearly someone is buying them, obviously and they are. And Microsoft. They came out and said their capacity constraint, supply constraint, people, there's infinite, infinite demand. Everyone wants more of it. So the capex of these mega cap tech, as long as that's the trend I'm watching, that is increasing, will continue to increase. That's the revenues of all my ai companies that I own. That tells me that there's demand, hence they're doing the capex. So that's the one to watch. And if that rolls over, then something is wrong because they don't see the demand on the other side and hence they're not doing it. And that would mean the AI trade, all the AI picks and shovels guys, their revenues are going to roll over the whole equity market could roll over from it.

Speaker 1:

So it's the one thing to watch, very insightful, and I want to thank you so much for your time today. What a great conversation. Thank you for having me. I appreciate it. Conversation Thank you for having me, I appreciate it. Thanks so much for listening to today's episode and if you're enjoying Trading Tomorrow, navigating trends and capital markets, be sure to like, subscribe and share, and we'll see you next time.