51% Want an AI to Pick Their MPF Funds. The Market Has One Real Robo-Adviser.

Track 1 Data Mirror | 27 October 2026

The core claim: demand has arrived, supply is still a catalogue

The most revealing number in GUM’s MPF Confidence Index, published on 21 October 2026, is not the headline confidence score, up to a 15-year high of 5.1. It is the demand for help. The survey covered 808 MPF members aged 18 to 64, interviewed on 13 to 15 August 2026. 51% of respondents were interested in AI-assisted investment analysis and fund selection tools, a further 51% rated regular market updates as the most valuable support, and 35% wanted tools to assess their risk tolerance.

In other words, one in two workers no longer wants to read fund fact sheets alone. They want a tool that answers a question: which fund should I buy?

Take the tools on the market apart, and the answer is awkward. Most products sold as “smart fund selection” will not select a fund for you. They will filter funds for you — and between filtering and advising lies a gap measured in fees and risk. Demand has arrived. Supply is still at the catalogue stage.

Piercing the data: a four-rung ladder, and only rung three advises

Rung one: information and screening — the MPFA Fund Platform. The MPFA’s MPF Fund Platform offers a fund information table with filtering and sorting, five- and ten-year performance charts, and “My Selection”, which compares up to 10 funds side by side. Data as at 30 September 2026 covers 382 constituent funds, with a median risk indicator of 9.86%. It is the best cross-trustee price-comparison tool in the market. But it will never tell you which fund to buy. It assumes you already know your risk tolerance — and GUM’s survey shows that 35% of members want a tool for precisely that step.

Rung two: risk questionnaires and fund matrices — the status quo at most trustees. Many trustees offer a fund selection matrix or a risk assessment that maps you to a category such as aggressive, balanced or conservative. That is useful, but it is still not personalised advice: it gives you a category, not a portfolio. It does not compare your current portfolio with the recommended one, and it does not state the total fee of the recommended portfolio.

Rung three: a genuine personalised robo-adviser — AIA MPF Smart Advisor. This is the only tool found in this audit that is publicly promoted to members and outputs a personalised portfolio recommendation. It runs inside the AIA+ mobile app: five investment risk assessment questions, a recommendation drawn from up to 50 portfolio options, a comparison of the current portfolio with the recommended portfolio, the ability to customise the allocation and instantly view simulated historical returns, plus a retirement gap calculator. The small print matters as much as the features: the service is independently operated and provided by Magnum Research (AQUMON), and the investment recommendations are solely provided by that company; AIA and its affiliated intermediaries do not provide investment advice. Magnum Research holds MPFA licences and is authorised by the Securities and Futures Commission for Type 1, Type 4 and Type 9 regulated activities. In short, this tool earns the title “adviser” not because it uses the word AI, but because a licensed entity stands behind the recommendation.

Rung four: AI headlines that are not fund selection — do not mistake them. Manulife joined the GenA.I. Sandbox led by the Hong Kong Monetary Authority, the Securities and Futures Commission, the Insurance Authority and the MPFA in August 2026, with two admitted use cases: BetterClaims AI, which helps advisers answer medical and critical illness enquiries, and a Red-Flag Tool for intermediary fraud risk review. Neither helps a member pick a fund. HSBC’s Portfolio+ is a retail bank-led digital discretionary investing service with automatic rebalancing — but it is not an MPF product. Reading these headlines as proof that “AI fund selection has arrived for MPF” is the single biggest source of the demand-supply confusion.

One honest audit caveat: this audit covers tools publicly promoted to members as at 27 October 2026. A trustee may run a member-only tool that is not publicly promoted; if so, the trustee should say so. But on what an ordinary member can find in the open market, “one real robo-adviser” is the current state of play.

The compounding cost: a filter dressed as an adviser is more expensive than it looks

Why does “filter dressed as advice” deserve Track 1 treatment? Because the cost of a wrong steer is not a one-off handling fee. It is a fee gap charged every year.

The market-average fund expense ratio is about 1.36%, against about 0.77% for the Default Investment Strategy (DIS) in practice — a gap of 0.59 percentage points. On an illustrative basis (monthly contributions of HK$5,000, gross return of 7%, compounded monthly, contributions at month-end), that gap is worth about HK$26,000 over 10 years, HK$160,000 over 20 years and HK$559,000 over 30 years. Against the average member balance of HK$338,950, it costs about HK$2,000 a year. A tool that steers you from a low-fee DIS portfolio into a high-fee active fund, without proving it can persistently outperform, hands you a bill that renews annually — and calls it advice.

The danger of chasing past performance has numbers behind it too. MPFA provisional data as at end-September 2026 show equity funds averaging 10.3% over the previous 12 months — but the best fund gained more than 98.4% while the worst lost 15.7%, a spread of over 114 percentage points within the same broad category. Any tool that merely ranks funds by last year’s winner is inviting you to chase the tail of that dispersion. The MPFA’s own warning stands: MPF is a long-term investment spanning more than 40 years, and members should avoid short-term approaches or market timing that risk buying high and selling low.

The benchmark for any fund-selection tool should be the DIS Core Accumulation Fund: an average annualised net return of 7.1% since its 2017 launch (MPFA provisional data as at end-September 2026), with a fee cap of 0.85%. If a tool’s recommendation loses to that benchmark on both fees and diversification, it is decoration, not advice.

The way through: five tests before trusting any tool

MPF has two hard execution limits that no tool can remove: forward pricing (switch prices are unknown until the next dealing day, T+1/T+2) and a transfer window of about one to two weeks out of the market. A tool that encourages frequent switching is not helping you capture markets; it is multiplying your unknown-price risk. Five tests are the minimum before trusting any selection tool:

Test one: does it ask about capacity, or only about feelings? A real risk assessment must cover horizon (years to age 65), job stability and other savings — not just “could you sleep if your balance fell 20%?” If changing your age or horizon leaves the recommendation unchanged, that is labelling, not personalisation.

Test two: does it show the total fee? What is the fund expense ratio of the recommended portfolio, and how does it compare with the DIS actual average of about 0.77% and the 0.85% cap? A recommendation that hides its fee should be treated as advertising.

Test three: does it compare with your current portfolio? AIA’s tool does; other trustees should copy that feature. Without a comparison, there is no way to tell whether the recommendation is an improvement or merely a switch.

Test four: who is responsible for the recommendation? Look for the name of the licensed entity, not the brand logo. The AIA example shows that the brand and the advice provider can be two different companies; the clearer the attribution of responsibility, the more credible the tool.

Test five: does it warn about switching costs? That includes unknown pricing, the out-of-market window, and the fact that bonus units themselves attract applicable fees. A tool that offers “one-click switching” without mentioning the window is hiding the one risk it cannot eliminate.

If you are not an AIA member and have no personalised tool available, the most practical stack today is: use the MPFA Fund Platform’s “My Selection” to compare up to 10 funds of the same category side by side (like for like — never an active Hong Kong equity fund against a Hang Seng Index tracker), use the DIS as your fee and return benchmark, and check risk class and fund expense ratio in the fund fact sheet. It is not “smart”. But it will not pretend to know your answer.

The 51% figure is a signal: members are ready to accept help. The market’s response should not be more filters renamed as AI. It should be an adviser that shows its fee, names who is responsible, compares with your current portfolio, and admits the limits of unknown pricing. Until then, every tool that falls short of those four tests is a catalogue — and a catalogue should not make your decisions for you.