Feng Ji is the founder and CEO of Baiont, a top-performing quant fund in China that uses artificial intelligence to develop trading strategies.
He argues quant trading is fundamentally a computer science task and predicts that quant fund managers failing to embrace AI will not last another three years.
In this conversation with the Financial Times’ Asia Technology Correspondent Zijing Wu, Ji talks about how his team of young computer scientists with no finance background is disrupting the quant trading sector in China and has ambition to go global. He says quant trading is attracting the best AI talents and offers fertile ground for start-ups, such as DeepSeek.
Zijing Wu: Quant trading is still relatively new in China compared with the US and Europe. Can you describe the current landscape in China?
Feng Ji: The first wave of quant trading here started with some very talented Chinese traders coming back from Wall Street. Around 2013, regulations changed to allow quant trading, and more hedging tools were introduced in the Chinese market, which created a fertile ground for this first generation of Chinese quant traders. They did very well and remain leaders of the biggest funds today.
We are the second generation and very different. We come from “out of the circle” with zero finance background. We believe that quantitative trading is the same as many other tasks of data mining and analysis. There is nothing special about it. We regard it as a pure AI task, so our team is composed of only computer scientists and engineers.
ZW: How do you apply AI in quant trading, and what’s the difference between what you do versus the old-school quant trading?
FJ: AI technology has made significant progress in the past 10 years, especially in time series data modelling. Whether it’s language or multimedia AI models, fundamentally it’s all about modelling time series data. For example, the core task of ChatGPT is to predict the next word. It’s essentially the same with quant trading. Instead of predicting the next word we predict the rise and fall of prices in the next time interval.
A traditional quant fund would divide up its team in to several functions focusing on different stages of the pipeline, mainly factor finding, signal generation, modelling and strategising. These functions are independent and somewhat isolated from each other.
We see all these stages essentially as the same machine-learning task and approach it holistically with the same foundation model. This has a far-reaching impact on operations. It’s like before ChatGPT, language processing companies also had team divisions which focused on word separation, tagging, analysis etc. Now ChatGPT can do all of them at the same time with the same model.
AI Exchange
This spin-off from our popular Tech Exchange series of dialogues examines the benefits, risks and ethics of using artificial intelligence, by talking to those at the centre of its development.
See the rest of the interviews here
ZW: Why is your holistic approach better than the traditional division of labour?
FJ: First of all, you can predict and plan the upgrades of a system based on machine learning. Like when ChatGPT launched its first generation model, you basically have an idea what the second generation would be like and how long it would take to get there. The ability to continuously upgrade in a systematic way is key to a quant fund manager.
The second advantage is cost effectiveness. Instead of hiring 50 people to find factors, we use 100 GPUs and one person who writes the algorithm for factor finding. The result is even better and much faster. The same applies to all stages.
ZW: How big is your team and how much are your assets under management?
FJ: We currently manage close to Rmb7bn ($970mn) and our team has only about 30 people. Two-thirds are doing research while the rest focus on operational work. Our research is mainly about improving the algorithm and our own foundation model.
ZW: Does the industry see you as disruptive?
FJ: When we first started doing it, about four years ago, many people thought it was impossible. How can a bunch of computer scientists understand business and the markets? The truth is — we don’t and we don’t need to. Actually none of us did any trading before this. We see this as a pure machine learning task and one that’s totally doable.
Now very few people doubt us any more. Instead everyone is asking us frantically how they can better use AI.
So my prediction is that in three years, quant managers who do not complete their AI transformation will be eliminated by the market. Because the space is getting more and more competitive, and machine learning will become an essential tool. There is no reason why one shouldn’t adopt it.
ZW: Do you build your own model from scratch and can you give us an idea of how it works in trading?
FJ: Yes we built it all by ourselves. Because market data and behaviour is very different from, for example, language data. What we deal with is a lot more complex and we need to build specialised models for it.
Typically we focus on short term trading, from minutes to hours. This is what AI is best at. It’s like predicting the weather. If you have to predict the weather in a month it would not be so accurate, but if you predict in five minutes, the accuracy is very high, because you can catch many signals. Short-term signals are relatively predictable and we have analysed enough data to make quality predictions.
We will comprehensively evaluate the prediction of different signals from minutes to hours in real time. Then make a comprehensive score of these predictions, and based on such scores we build a dynamic combination of trades.
ZW: Does it mean you don’t care about the fundamentals at all?
FJ: Basically yes. Fundamental factors and alternative data factors change very little during the day. We mainly rely on trading data. The core of short-term price fluctuation is driven by trading data.
ZW: Why did you and your team, coming from a machine learning background, decide to get into quant trading instead of the more popular AI start-ups focused on large language models for example?
FJ: After graduating with my PhD in machine learning, I spent about a year looking at various directions in which machine learning and AI can have a truly disruptive impact, rather than a simple upgrade of the existing tools.
The second factor I considered at the time was whether it can bring a good cash flow. I realised at the time most of the AI unicorns don’t make money. They may be doing valuable things but it’s difficult to sustain. Also for many of them, their success depends significantly on capability of sales, not technology because there’s limited differentiation in their core tech. I felt like being a super nerd, I’m not interested in anything that’s heavily sales driven.
Then I found quant trading, which ticks all boxes. It’s an industry we can redo all over with AI. It’s not just a traditional linear model, but with the potential to create a neural network or a random forest. It’s a challenge I’m excited about. And it’s disruptive. It’s like designing a new electric vehicle factory, totally disrupting the old car manufacturing.
The other good thing about quant trading is it’s easily verifiable. You find out right away if you are on the right path or not by doing more than a thousand trades in one day.
It’s also almost purely technology driven. Most quant companies are led by people with a tech background. Because you can’t manage a team of nerds and geniuses if you don’t understand the technology yourself.
ZW: What kind of nerds and geniuses are we talking about here?
FJ: Our team, myself included, came from a computer science competition background. Out of our 30 people we have 13 gold medallists. Our team’s gold medal “density” is probably higher than any tech giant out there. Quant trading is an industry where you see the highest proportion of geniuses. It’s the same in the US. I believe the top machine learning talents are 80 per cent in Wall Street and 20 per cent in Silicon Valley.
ZW: Is this why DeepSeek came out of High-Flyer, one of China’s biggest quant funds?
FJ: Indeed. I was not surprised at all about that. [The] key contributions to LLM [that] DeepSeek made was to reduce engineering costs and improve communication efficiency between GPUs. This comes naturally to quant traders, because how we quantify time is in nanoseconds to microseconds, while traditional internet companies have a timescale of seconds, or milliseconds at best.
For example, a big tech platform with a billion users online at the same time wants to make sure there’s no lag and human reaction is between 50 to 150 milliseconds. It’s fine if you have a 10 millisecond delay. But in quants trading one millisecond is forever.
Quants trading is also where you have very healthy cash flows to attract the top talents. Ten years ago it attracted the smartest people from maths and physics, because they can transfer their data analytical skills to finance. But today it’s gradually taken over by computer scientists. Because we don’t even need to transfer skills — machine learning is essentially the same in designing the best tool to analyse data. It doesn’t matter whether it’s data from finance or any other area.
Making a lot of money also means the team has the luxury to branch out to do things they are more interested in pursuing. I call this a technology spillover.
When you have a large amount of geniuses and ample resources, they are able to spin off some unrelated technologies based on similar core skills.
It’s happened many times in history. For example, [hedge fund] DE Shaw’s founder created a large scientific research centre to use self-developed super computers for chemistry. It has nothing to do with quants trading but applying similar core skills.
ZW: Just like DeepSeek, your team is all from a Chinese education background. How do you compare the young talents in China and the US?
FJ: There’s very little gap these days. We are competing on basically the same level. And China has a larger pool of such talents thanks to our education system with a stronger focus on science and technology. We are particularly strong in engineering capability and algorithm innovation.
In the past decade, smart young people from everywhere in the world can freely communicate with, learn from and work together with each other on open source AI platforms. This has given our generation of Chinese coders a great opportunity to catch up with the world’s leading technology in this area.
The other thing about this young generation of talents in China is, unlike their parents, they grew up in mostly middle class families where they didn’t have to do things they didn’t like in order to make a living.
Most of our team are in their twenties. I’m 37 and the oldest by far. Their number one priority is to have fun. So instead of going to big tech companies where they will probably have to deal with politics one way or another, they would much rather come to a smaller research-oriented team like ours, where they work with similarly smart colleagues and a manager who speaks their language.
Having grown up in well-off environment also means this generation of Chinese young talents are more idealistic than their parents. You see more going into research instead of finance for quick money. We actually want to do something to change the world.
ZW: What’s a daily work schedule like for your team?
FJ: It’s basically like a research institute. No dress code — shorts and slippers are the most common. We arrive before markets open, start programming and discuss our work together, and review the performance before the market closes. Run a few more experiments, read and discuss some papers, and go home. The difference between us and a research institute is we have better resources. We build our own computing power. The more compute you have, the faster you get the results and the more efficient you are. It’s very important.
ZW: What’s the ultimate goal for you and your team?
FJ: In the midterm we want to build a world leading AI-native quants fund from China. We mainly trade in the Chinese markets now and we are looking to expand into the key overseas markets. When people talk about quants funds they all think about the Wall Street top firms, few knew about the Chinese funds. While the first generation of Chinese quants funds used the methodology learned from Wall Street, we can differentiate better by being AI-native early enough. We have a chance to compete with the global leaders.
In the long run we would like to build a computing company. There are many potential areas we are excited about, where we could spill over our technology. LLM isn’t necessarily the best use of AI.
This transcript has been edited for brevity and clarity.