AutoML Capital Sets out to Harness AI Technology to Democratise Smart Portfolio Curation

Rebecca Lim is the Founding Partner of Hong Kong-based AutoML Capital, an SFC-licensed digital asset management advisory firm that aims to leverage its proprietary machine learning algorithm to optimise portfolio allocation for mass affluent and entry-level HNW investors. Lim ‘met’ with Hubbis recently to explain that their mission is to democratise access to AI-driven investment tools so that investors can outperform the market with a selection of tailored portfolios recommended by AutoML. She estimates that there is a vast market – indeed in Hong Kong alone she reckons the number of affluent individuals grew by 500,000 in just two years between 2017 and 2019, reaching 3.7 million. She explains that these mass affluent individuals as well as many HNWIs lack access to professional advice and the right digital tools to help curate their investments, and that AutoML’s AI Engine addresses those needs, leveraging more than three decades of market and economic data. The firm is young, and Lim has not given up her day job until very recently, but with financial support from the Hong Kong government and external investors, she anticipates a bright future ahead.

“The mass affluent and entry-level HNW segments are growing rapidly yet remain vastly under-served,” Lim reports. “Hong Kong’s affluent population accounts for more than 60% of its total population, yet the private banks do not really serve this huge market. And the same is true in Singapore, where the equivalent figure is roughly one-third of the population, or China, where more than 5% are in this classification, and of course that represents a vast number there. Brokers in Hong Kong do not offer rigorous investment insights that meet this segment’s needs and expectations, making it somewhat of a no-man’s land.”

She explains that the firm’s analysis of the market allows them to categorise these individuals with three core profiles. They might be passive ETF investors with little or no asset diversification and no potential to outperform the market. They might be mutual fund investors, with diversified assets but also with delayed market response. And they might be stock pickers, which she says is often a time-consuming game of roulette, especially for those with little formal knowledge or without a very rigorous approach.

Curating professional-grade portfolios

“We identified the market and the hurdles these individuals are facing, and devised our solution to democratise access to AI-driven investment tools so that investors can outperform the market while saving their time,” she elucidates. “AutoML Capital’s AI has been devised by machine learning experts and PhD level scientists and is based on our R&D in the automation of the data processing pipeline, feature selection, model search, as well as self-evolving machine learning algorithms. AutoML Capital spent the equivalent of over ten ‘man years’ and leveraged over 30 years of financial data – such as asset prices, news, futures, and a host of economic data - to build our proprietary AI engine. It provides access to all major asset classes with equities spread across developed and emerging markets, giving investors a remarkably broad range of exposure.”

Lim maintains that in the midst of rising global debt and inflationary pressures, the typical non-professional or relatively unsophisticated investor faces immense market uncertainties and an overload of information.

AI to laser guide investment decisions

“Frankly, AI is much better at understanding the vast treasure troves of data and discerning the implications than people, and we believe that AI will only grow exponentially,” she states. “Moreover, with even more data processed in the future, and rising all the time due to technology and accelerating speed, the gap between what computers and people can absorb and analyse will also grow at an exponential rate.”

She also debunks the view that there is a lot of competition already. “There are some advisory firms claimed to be AI-driven but most of their AI capabilities focus on risk assessment and their voice/chat bot to enhance customer experience,” she says. “Instead, put 95% of our R&D efforts into portfolio design and construction.”

Beating the odds

She adds that with the investor’s consent, the service automatically re-balances the investor’s portfolio based on their objectives and risk appetite. “Our algorithm has consistently provided a much lower drawdown but similar returns to market averages,” she states. “In fact, we are seeing an average annual investment return of 10%, outpacing market performance, with very carefully controlled maximum drawdowns and a focus in getting the right timing to reinvest back into the market after the downturns.”

AutoML’s work directly with asset management firms is expanding, with the firm providing its service as an add-on. “But we are careful to steer well clear of competing as an intermediary, so we never hold any money or assets ourselves. We do not cross lines in the sand, all investments therefore reside with the asset manager or bank the investor chooses. The range of EAMs and others we are working with is expanding constantly.”

A simple premise and a simple fee model

She explains that AutoML Capital operates on a simple and low flat-fee model with no other performance fees. “Our customers have a high degree of flexibility to deposit or withdraw their cash with a very low required minimum balance,” she says, “and unlike other AI black boxes, we provide an extra level of human touch to ensure the models and algorithms are explainable and understandable.”

The three founders first met as students at the University of Oxford some 15 years ago. They then ventured into different careers upon graduation – some into research, some into technology, and some into finance. “And after our careers developed, we all put our heads together and the idea came up in 2017, so we formed a team covering every expertise we needed, leveraging our different skills and experience. We have funded the business to this stage, with some support from the government and a few external parties. The next stage is for the firm to be self-sufficient.”

Boosted by external capital

Indeed, the firm’s progress has been encouraging thus far.

“As a home-grown FinTech, we were honoured to have in 2017-2021 received recognition and funding from the Hong Kong government, from Croucher Foundation, and also Cyberport, raising around HKD4 million that has enabled us to polish our research and development to a level that gave us the confidence to roll the solution out and also to embark on our regional expansion plans,” Lim reports.

Horses for these courses

She characterises the typical customers. One, she says, might be like her father, a businessman who has been working for 30-40 years, who needs some liquidity, who needs stability and who hopes for modest growth more than racy returns. The other is a younger professional more like Lim herself. “Neither of us want to be distracted by the markets, we don't want to spend much time to on that, yet we want a really diversified portfolio with a very flexible fee structure, a very flexible operating model that allows us to withdraw or deposit money at any time.”

She is quick to state that AutoML is not a robo advisory. “Our work centres entirely on optimising the AI to achieve the return for the clients, whereas a robo advisory is garnering AUM and we are not, all assets and money stay with the clients or the intermediaries we work alongside. Our mission is singular - AI is actually for predicting markets in order to generate return while reducing drawdowns.”Acting on the AI crystal ball

She elaborates on those comments, noting that the AI predicts the directions of the markets, then the output goes into optimising an all-weather portfolio, and the third element is a smart protection mechanism that will continuously predict the market direction. “If it looks like it's going to fall or possibly crash, we'll de-risk the portfolio, and then re-invest when we see the market about to or actually rebounding. We focus these actions through ETFs to represent all major asset classes and globally.”

She explains that liquidity is extremely important. “On a long term view, if you buy and hold, the portfolio will make sense, but on a short-term basis, the client should also have the flexibility to liquidate as and when required,” she reports. “We measure our performance against two benchmarks, one is the S&P500, and the second one is a portfolio with 60% bonds and 40% in equities. The result is that in any single month, our drawdown will be lower than the S&P500, and in most months our drawdown will be lower than a 60-40 portfolio.”

Choose your partners wisely

Lim also clarifies the firm’s business model is to source customers mainly through Interactive Brokers; those clients then set up an advisory account and AutoML offers a selection of four portfolios it manages on a discretionary basis for those clients.

“We are partnering, we are not ourselves seeking those end-customers,” she says. “In that way, we gain access by partnering with those who want to enrich their offering, and the customers are enticed by the additional product and service suite we bring. Custody stays with Interactive Brokers, and we manage the account for the clients from a selection of four portfolios we currently offer, with the clients selecting their preference based on their particular needs and expectations.”

She elaborates on these four portfolios, explaining that they come under the headings of growth, moderate, conservative, and fully invested conservative. “Each portfolio has its own optimisation target, and its target volatility,” she reports. “They comprise ETFs, stocks and derivatives such as futures. And as to fees, we only take a fixed management fee, which is levied by Interactive Brokers and then we divide the fee between us. We do not currently charge any other performance fee.”

She explains that this same model also applies to their partners in the EAM community, which she says is also a natural channel, as they have clients, but they do not have easy access to this type of algorithm-driven portfolio or managed account. “Their clients then have access to AutoML Capital as a managed account, but it is not a white label product, it has our name on it,” she explains. “The same type of fixed management fee arrangement and split works as with Interactive Brokers.”

Key Priorities

Lim reports that her first priority is to build more channel partners. “Realistically, we realise that we are good at building the engine, developing the AI, but actually not so good at going out and selling our name and products, so we believe the relationship with these partners would help us massively in terms of being able to focus to do what we are good at doing,” she explains.

Her second priority is to expand the research team so that they can continuously improve the performance and can layer in other markets or assets, such as cryptocurrencies, for example. “And brand awareness is a key mission for us, and the more people that come to us, the more referrals we will earn and the more momentum we can achieve,” she says.

Better by design

She closes the discussion by commenting that the channel partners the firm is garnering are working with them due to their appreciation of their technological expertise, their experience, and their trustworthiness.

“We have the capability to do this better than anyone in the market,” she says. “We offer these partners the incremental technological enhancement and optionality that they and their customers seek. We are open to discussions and imaginative and transparent in our dealings throughout. It is a great combination, we believe.”

Ready for lift-off

Her very final comment is that she is passionate about the technology they have built and optimistic that the business will take off and become self-sustaining. “We have invested our own money and we are lucky to have obtained some other financial support,” she says. “The next stages will prove that the technology and the model work. It is very challenging, but we are deeply committed and very excited about the future.”

Getting Personal with Rebecca Lim

Lim was born and educated through secondary school in Hong Kong, and later obtained her Bachelor and Master’s degrees in physics from the University of Oxford. She then embarked on a career in finance, centred on quantitative analysis. “We are building AutoML right now, but most partners are still working full-time,” she reports. “Once we have the critical momentum, we will probably dedicate our entire time to the business.”

As if she were not busy enough, Lim has three young children, twins aged two and a baby of just one year old. “As you might imagine, I do not have much time, but I do love running and train for marathons. I did my last marathon in Hong Kong some ten months after giving birth to our third child and managed a fairly respectable 3 hours and 45 minutes.”

She says she is also a keen advocate of encouraging young women to take on more technological roles. “I am part of an NGO that promotes this and helps underprivileged girls, including teaching them about coding and some basic AI concepts,” she says.

There is little doubt that Lim is a driven and insightful individual, nurturing three very young children as well as trying to build a new business concept, yet still finding time to help those less fortunate and also beat her own physical endurance records.