My next guest was coding machine learning models back when most of Wall Street still thought AI meant science fiction. Two decades later, AI is everywhere, and he has watched the hype catch up to what he has actually been doing the whole time. Evan Szu is a 30-year commodity trader and the Founder of Gamma Prime, a marketplace opening up hedge funds and private deals to investors who normally never get in the room. Evan, good to have you.
Hi, Johnny. Good to be with you.
You have been coding machine learning models since the 90s, long before this was trendy. What is the biggest misconception people have about AI in trading right now?
First, just to give fair credit — this is not because I had some sort of insight. My father is the founder of the International Neural Network Society, and he did that in the 80s. So I grew up with this over the dinner table. A lot of people think AI is so smart and so fast that it is going to ruin trading. The reality is that is just not how it works. It is faster. It can run 24/7. But it is entirely data dependent. What people do not realise is that it is not the models — it is always the data that matters in machine learning. I will give you an extreme example: how much data is there about how to trade commodities in the middle of a war where the Straits of Hormuz are closed and we have a mercurial president that changes his mind every four hours? Zero data. And so a machine learning model is going to do very poorly with this because there is no precedent.
You have been using machine learning in your own trading since around 2005. What has been the biggest change you have seen?
From a trading perspective, the technology has been around quite some time. Most hedge funds have already employed it to the extent they can. The biggest change is that we have now seen large language models come to the fore — which means people can now access AI without actually being technical, without understanding how to code the mathematics behind it. That has caused AI to explode into the public consciousness. But the actual math, the actual underlying guts that run these things, they have been around for decades. It is more of a public awareness that has caused AI to be such a hot commodity now.
If AI gives every trader the same tools and the same data, does the edge disappear?
The way machine learning is effectively used in trading is this: if you already have a particular style and approach for trading that works, you are not going to be able to just ask ChatGPT to go make a bunch of money — it is not going to work that way. However, if you have a particular style that you have already honed, that generates some degree of alpha, a machine learning model can definitely make that more efficient, more effective, and run it 24/7. That is really where the edge comes in — but it still takes human judgement.
Walk us through what Gamma Prime does and who it is actually for.
Just because you have a few million or even tens of millions does not mean you have access to all the private markets you need. In order to get access to private markets, you need a certain amount of mass. Strange to say, but there is actually an underserved class in private markets. If you have five to ten million in net worth, private bankers do not pay you much attention because you are just not worth their time. So there is this underserved gap where you actually need to diversify like a large family office or institution, but you do not have the resources and tools. That is where Gamma Prime comes in. We hit that underserved medium to high net worth individual — anybody who is at least accredited. Our sweet spot is in the tens of millions, maybe hundreds of millions of net worth.
What AI claim from a fund manager should make an investor walk away?
If they come back and say we use AI and we know how to do this better than anybody else — that is not an edge. If you are using machine learning, you need a fundamental approach that already makes money. You need alpha and an edge that already vests with someone with experience. Then AI can supercharge that. But if someone says we are just a smart quant fund that uses AI but we never had an edge before AI — I would be quite suspicious.
Evan, thank you for joining us this morning.
A pleasure. Hope to see you again.
