Why Generative AI Still Can’t Trade | David Wright on How Quant Alpha Actually Is Done With Machine Learning, Decision Trees, and Gradient Boosting

Very pleased today to be joined by David Wright, co head of Pictet Asset Management's quantitative investments franchise, which manages over $30,000,000,000 in their quant group. This includes most recently two ETFs they launched, PQNT, their Pictet AI enhanced international equity ETF, and PQUS, their Pictet AI enhanced US Equity ETF. This is a sponsored fireside chat. David, so glad to have you here. Welcome.

Thanks, Jack, for having me.

Can you start out by defining the quantitative tools that you use? There's AI. There's machine learning. There's this new thing of AI, generative AI, that's come on the scene past three, four years. There's neural nets. What exactly are the tools that you use? What are not the tools that you use? I think it's important for us to be precise.

Yeah. I often when I'm having these discussions, what we're not doing is almost as important as as what we are doing. So, again, a few people are gonna argue over these definitions, but but let's kind of take it from the top. Artificial intelligence, I think generally what we're suggesting here is a computer doing something that historically would have been a human task. Now once we get into the realm of machine learning, that is where that task, rather than being programmed by a programmer, is, an algorithm that is trained on data to do that task in a repeatable way and has been found from da

So the techniques and the large amounts of competitive power that are powering GenAI, that is not what you're doing. I'm curious why. Is that because generative AI is basically making things up and people are it's now famous for its hallucinations, which are getting better. And the last thing you want as a quant investor is for the computer to be making things up.

The last thing you want from any investor is is is information that is incorrect or or the or or that is being made up. So I would just take a little bit of a step back. In many industries and and even within parts of finance outside kind of quant investing, the the arrival of generative AI has been very much a revolution. There's kind of a before and after in terms of people investigating and using artificial intelligence. In the quant industry, we have used technology and data from our birth in the seventies and eighties and have evolved the usage of that technology as the years have passed.

That makes sense. David, just zooming out, how much more computing power and advancements in quantitative finance has the quant finance, you know, world, not just you and your firm, advanced over the past three years. I think in the world of generative AI, it's gone from extremely, extremely you know, very little achievements to rapid, rapid achievements. I understand you're saying, okay. We in the quantitative finance, we've been using neural nets. We've been using a lot of these techniques for a long time, so it's not new to us. But but just tell us about how quickly the field has been advan

It's two dimensions. So you're completely right. I in in the quant world, it's not like a it's it's not like really a before and after the arrival of of generative AI. So, again, machine learning, AI, it's been looked at in some capacity five, ten, fifteen years by by quite a few different quants, particularly in in in the hedge fund end of the market. But certainly, the last ten years, many of the the things that have aided the rise of generative AI, more data, cheaper compute power, faster graphics processing chips, more access to those types of things. Rise of cloud computing, they've provi

That's really interesting. Other leaders from the