Inside iQuant.fund: AI built for institutional trading

Where iQuant.fund started: traders automating their own strategies

iQuant.fund began with a group of traders who were tired of doing everything by hand. We ran our own strategies, and the work never stopped. Every signal had to be checked, every position watched and every decision made in real time.

I described it this way to Kanebridge News in 2024: "We often found ourselves juggling multiple screens, analyzing vast amounts of data, and spending countless hours away from our families."

What we wanted was straightforward. In the same interview I said we were after "a system that could automate our trading strategies". If a set of rules could be written down clearly, a machine could apply them faster and more consistently than a person watching several monitors at once.

That idea sits at the heart of systematic investing. A systematic approach replaces gut feel at the moment of decision with rules that are defined in advance, tested against history and applied the same way every time. It doesn't remove human judgement. It moves judgement to the design stage, where there's time to think, and away from the middle of a volatile trading session.

I co-founded iQuant Fund Ltd in February 2021 and served as its Chairman. The company built AI-assisted systematic equity research and AI trading technology for institutional clients.

From the ASX to global exchanges

The early focus was the Australian Securities Exchange. It was a natural place to start, a market close to home.

Over time the research expanded to NASDAQ, the New York Stock Exchange and Hong Kong. Eventually iQuant's coverage spanned more than 20 exchanges and over 5,000 equities.

Expanding across markets is harder than it sounds. Every exchange has its own trading hours, currency, holiday calendar and data formats. A signal that works cleanly on one market can behave differently on another because of differences in liquidity, sector mix or the kinds of investors who dominate trading there.

This is where machine learning earns its place. Watching 5,000 stocks across several time zones is beyond any human team. A well-built system can scan that universe continuously, flag what meets the criteria and leave people free to focus on the decisions that need them.

How AI supports institutional equity research

iQuant's research work covered momentum signals, market-regime analysis and portfolio diagnostics. Each one answers a different question for an institutional investor.

Momentum signals

Momentum is one of the most studied patterns in finance. In simple terms, stocks that have performed strongly over recent months have tended, on average, to keep outperforming for a while, and weak performers have tended to stay weak. Researchers have documented the effect across many markets and long periods of history, although it can reverse sharply after major market turning points.

Machine learning helps by testing many ways of measuring momentum at once, across different lookback windows, and by weighing momentum alongside other information rather than reading it in isolation.

Market-regime analysis

Markets don't behave the same way all the time. A calm, trending period looks very different from a high-volatility sell-off, and a strategy that suits one can struggle in the other. Regime analysis tries to identify which environment the market is in, so that other signals can be read in context.

Portfolio diagnostics

Diagnostics look at a portfolio as a whole. How concentrated is it? Which sectors, markets and risk factors is it exposed to? How do the holdings move relative to each other? For an institution managing a large book, these questions matter as much as picking individual stocks.

I've written more about the strengths and limits of these techniques in what machine learning can and can't do in equity markets.

Modern Portfolio Theory inside an AI trading system

Alongside machine learning, iQuant's approach drew on Modern Portfolio Theory. It's worth explaining, because it's one of the most important ideas in investing and one of the most misunderstood.

Harry Markowitz set out the framework in a 1952 paper titled "Portfolio Selection", and he later shared the 1990 Nobel Memorial Prize in Economic Sciences. His key insight was that an investment shouldn't be judged on its own. What matters is how it changes the risk and return of the whole portfolio.

Diversification

Two assets that are each volatile can form a steadier portfolio if they don't move together. When one falls, the other may hold its value or rise. The lower the correlation between holdings, the more diversification reduces overall volatility. That's why a portfolio's risk isn't simply the average risk of its parts.

The Sharpe ratio

William Sharpe, who shared that 1990 Nobel prize with Markowitz and Merton Miller, developed the measure now known as the Sharpe ratio. It takes a portfolio's return above the risk-free rate and divides it by the volatility of those returns.

The result shows how much return a portfolio has earned for each unit of risk taken. A higher figure means more reward per unit of volatility, which makes the ratio a useful way to compare strategies that take on very different amounts of risk.

Where the theory needs help

Modern Portfolio Theory has known limits. It relies on estimates of returns, volatility and correlation, and those estimates come from historical data that may not hold in future. Correlations between assets also tend to rise in a crisis, which is exactly when diversification is needed most. And volatility treats upside and downside moves the same way, even though investors only lose sleep over one of them.

In my view, this is where machine learning and classical theory work well together. Portfolio theory gives a disciplined framework for thinking about risk. Machine learning can help refresh the inputs as conditions change, and regime analysis can flag when historical relationships may be breaking down.

Why iQuant served institutional clients only

iQuant's products were built for institutions: family offices, allocators, funds and proprietary trading teams. We didn't offer them to retail investors.

That was a deliberate choice. Institutional clients usually have investment committees, risk teams, compliance functions and clear mandates. They can examine a model's assumptions, test its output against their own research and decide how much weight to give it. They also have the systems to act on research responsibly, including execution platforms and position limits.

A quantitative signal without that context can do real harm. An individual investor who sees a strong momentum reading may not know how often such signals fail, how trading costs eat into small positions or how quickly a market regime can change. Keeping the product in professional hands is, I believe, the responsible way to build this kind of technology.

When we announced our Series A, I put it this way: "Our AI trading product is a sophisticated tool that provides financial institutions with the ability to make highly informed trading decisions." The word "institutions" in that sentence matters.

iQuant was also a member of the AWS Activate and Nvidia Inception programs. Both support startups building with cloud and AI technology, which matters when you're processing data on thousands of securities across many markets.

The Kanebridge News interview and the Series A round

In June 2024, Kanebridge News published an interview with me under the headline "The Evolution of iQuant.fund: A Quantum AI Powerhouse in Financial Markets". It covered where the company came from and how it had grown.

When asked for advice, I said: "My advice would be to focus on innovation and never stop learning... embrace new ideas, and be willing to adapt your strategies as needed." I still think that holds, and nowhere more than in financial markets. Strategies that worked for years can stop working, and the teams that last are the ones willing to question their own models.

Two months later, in August 2024, iQuant announced the completion of a Series A funding round. The amount wasn't disclosed. For a young technology company, a Series A generally marks the shift from proving an idea to building it out properly, with more people, more infrastructure and more accountability to investors.

The announcement also set out where we wanted to go: "Our vision is to redefine institutional trading through the power of AI."

What I'd tell other founders in AI and finance

• Start with a problem you've lived. iQuant came from traders who wanted their time back, not from a technology looking for a use.
• Be clear about who the product is for. Deciding who you won't serve is as important as deciding who you will.
• Be honest about limits. Models fail, data has gaps and markets change. Building those realities into the product earns more trust than any marketing claim.
• Keep learning. The field moves quickly, and what counts as an edge today is often common knowledge a few years later.

About the author

Adrian Campbell is an Australian entrepreneur based in Indonesia and the founder and CEO of Kinnara, a global property marketplace. He co-founded iQuant.fund in 2021 and served as its Chairman, and has a strong interest in applying AI to financial markets. Read his biography.

This article is general information only and is not financial advice.