Quants worth following: Ernie Chan
I asked my partners, "IBM is one of the leading language modeling groups in the world. Why would Bob Mercer and Peter Brown leave our group?"
"Well, they went to a hedge fund called Renaissance Technologies…"
I asked, "What is a hedge fund? What does that mean?"
"Quants worth following" is our latest interview series highlighting thought leaders in the quantitative trading industry who actively share their knowledge and resources with the community. This week, we spoke with Dr. Ernie Chan, a two-time founder and respected author in the field.
Ernie founded QTS Capital Management, LLC., a quantitative commodity pool operator and trading advisor, in 2011. He managed a hedge fund and SMAs at the firm for many years and still serves as its non-executive chairman. In 2020, he went on to build PredictNow.ai, a fintech startup that he continues to lead as chief scientific officer. PredictNow.ai leverages machine learning to optimize trading strategies, offering investors and traders predictive insights to assess the probability of profit for their next investment.
Ernie has also written three influential books on quantitative trading: Quantitative Trading, Algorithmic Trading: Winning Strategies and Their Rationale, and Machine Trading: Deploying Computer Algorithms to Conquer the Markets.
In this interview, Ernie shares insights into his unconventional career trajectory, the contrasts between managing a hedge fund and a startup, upcoming plans for each company, and strategies for sourcing top talent.
With a PhD in theoretical physics from Cornell University, Ernie's early work as a machine learning researcher at IBM's T.J. Watson Research Center's Human Language Technologies division facilitated his transition into the quant industry.
"I definitely did not grow up thinking that I would be in finance. I grew up thinking that I would be a physicist. But after six years in graduate school in a very snowy, out-of-the-place campus in upstate New York, I discovered that I'm very bad at physics. It wasn't a lack of foresight, but when compared to people there who were leaders in the Manhattan Project, you can perhaps excuse me for thinking that I'm really bad at physics.”
Despite these challenges, Ernie completed his PhD with the support of his late supervisor. In 1994, the prospect of securing an academic appointment was improbable, so he pivoted into research—ultimately leading to his first exposure to hedge funds.
"I was in a world-renowned language modeling group at IBM. Coincidentally, the guys who started the language modeling group were Bob Mercer and Peter Fitzhugh Brown, and they had just left the group before I joined.
I asked my colleagues, 'IBM is one of the leading language modeling groups in the world. Why would these guys leave our group?'
And they told me, 'Well, they went to a hedge fund called Renaissance Technologies…'
I asked, 'What is a hedge fund? What does that mean?'"
Following Bob and Peter's departure, more researchers, including David Magerman and the Della Pietra brothers, left IBM for the hedge fund scene. Three years later, Ernie joined Morgan Stanley's AI group in Manhattan to kickstart his own career in finance.
There are significant differences between running a hedge fund and a fintech startup. Here are three lessons Ernie has learned running PredictNow.ai and QTS Capital Management:
Marketing plays a vital role in startups' success, especially in acquiring funding or customers. Still, excessive self-promotion in the hedge fund industry can result in regulatory issues or closure.
"Hedge funds are a heavily regulated industry. We can't easily tell the world about our hedge fund's performance because marketing is not allowed by the definition of being a hedge fund.
It's a strange feeling psychologically: knowing you had a great year but can't talk about it. You have to be very careful when soliciting investors, such as going through an intermediary. You have to go to industry conferences and talk about yourself without telling people how well your fund is doing. You have a great product, but you can't really tell people why it's great.
Thus, everything is done via indirect references and rumors, so to speak, such as the Wall Street Journal publishing stories about you."
"When running a quantitative hedge fund, there are two metrics for success: one is AUM, and the other is the Sharpe ratio. Both are highly numerical, so there's no need to exaggerate your performance. People can simply look at the numbers to know how good or bad you are.
Whereas with running a startup, at least as an early-stage startup, you can't boast about your revenue because it's minimal. And even if there is revenue, your expenses so outweigh the revenue that it's embarrassing to look at.
Are you profitable? Of course not. That's usually the answer. There is a lot of talking about how great you will be instead of how great you have been. In my experience, it's a really different culture."
"It's very possible to start a hedge fund as a solo founder. You develop a strategy; you might hire a programmer to build an automated system, maybe hire a data scientist to get the data, and so forth.
I've found that I have much more of a researcher’s personality than a salesperson’s. It's easy for me to talk about algorithms and technology, but it's very difficult for me to convince non-technical people about the business prospects of the company. Investors don't necessarily care about how great your algorithm is. Most of them couldn't understand it anyway. So you need someone who has the personality to always sell, sell, sell.
And so it's a little bit harder if you're a solo founder, but it's not impossible. There are many successful hedge funds with solo founders compared to tech startups because the latter requires even more sales, marketing, and story-telling acumen from the very beginning, whereas the former is more of a numbers game."
Ernie discusses the evolution of QTS Capital Management as a multi-strategy investment firm. Instead of solely relying on their proprietary strategy, they've successfully onboarded unconventional emerging traders and portfolio managers.
"Just one example: one of the guys who had been doing very well under our umbrella traded out of a town no one had heard of in East Africa. If he knocked on the door of Millennium Partners, he probably wouldn't even get a return phone call.
We worked with him over the years to ascertain that he really has alpha. Lo and behold, he has made seven figures in profit for our firm. And he's just one person sitting in a room somewhere in East Africa."
The firm plans to continue this onboarding strategy, which has been crucial for building a high Sharpe ratio vehicle beyond pure research. However, finding success in onboarding didn't come quickly.
"In the early days of building out the multi-strategy approach, there were instances where our sub-advisors didn't work out, but we eventually learned how to do effective due diligence and testing.
Now that most of the onboarded people perform well out of sample, not just having a great track record, they have started to generate profit for us after being hired.”
As for PredictNow.ai, Ernie reveals the company's plans to explore a new direction.
"In addition to serving the asset management industry, we are going to start looking to other verticals, such as oil and gas, to apply what we learned in applying optimization based on machine learning to finance.
And people say, 'Well, you don't know anything about oil and gas. How would you be able to do that?'
What we found is the wisdom in financial AI: Finance is maybe the hardest domain for AI because of the low signal-to-noise ratio, arbitrage activities, and regime changes. Thus, if something works somewhat well in finance, it probably will work very well outside of finance. Based on that assumption, we wanted to explore other verticals going forward."
Ernie finds hidden talent by giving back and mentoring the next generation of researchers from as early as their high school days.
"Everybody that I have worked with since I became an independent trader is inbound. I have never gone out and found any of them. These traders reached out to me because of the books I published. So essentially, I use the books as a way to tell people about myself and our work.
Beyond business, there are some really amazing high school students I have met over time. For example, there was a high school student recently from Bronx High School of Science. He reached out to me, and he talked about AI in finance. I asked, 'So, what have you read?' He said, 'I already finished Marcos López de Prado's book.'
Recently, my firm hired an intern who had just finished college at Columbia in finance. He said, 'Do you remember me?'
I didn't.
He said, 'Well, you know, you mentored me when I was in high school. Now I'm graduating.' So, I seldom refuse anyone reaching out to me, even if they're in high school, as time flies a lot faster than you expect."
Ernie is also a frequent lecturer and speaker:
"I've been teaching part-time for the last few decades. I've been an adjunct faculty member at Northwestern University for five or six years teaching financial data science and machine learning, but I stopped doing that because I got busy with my startup.
I've always taught at universities, independently, or at QuantInsti, a for-profit organization for emerging traders. Through teaching financial data science and machine learning, people tend to come to me out of nowhere for advice and mentorship, and later on, perhaps for internships, supervisory roles, and so forth.”
Check out the full interview on our YouTube here.