A saver tells an app they are investing for a home purchase in three years. The system asks five questions, assigns a risk score and recommends a portfolio holding mainly shares. The chart looks professional and the annual management fee appears small. But does the proposed portfolio actually fit the date the saver will need the money? Who is responsible for testing the recommendation and correcting a poorly interpreted answer?
Robo-advisory services use algorithms and digital interfaces to recommend or manage investment portfolios, subject to the actual firm’s service and authorisation. The gold standard is not flawless predictions or beautiful rebalancing animations. It is an accountable chain from customer objectives to suitable asset allocation, properly executed investments, transparent costs, accurate reports and a meaningful route when the system does not understand a person’s circumstances.
MoneySense’s September 2026 robo-adviser guide describes digital and traditional advice in Singapore. This article separates automated customer advice from investment decision-making in Asset Management and fund distribution in Investment Fund Distribution, all within the Gold Standard hub.
An automated portfolio still has five human responsibilities
| Stage | What must be controlled | Failure to avoid |
|---|---|---|
| Customer profile | Correct goal, time horizon and risk capacity | Wrong investor category |
| Portfolio selection | Documented asset allocation and product fit | Opaque algorithmic choice |
| Investment execution | Valid orders, custody and pricing | App balance diverges from holdings |
| Ongoing management | Approved rebalancing and goal changes | Unintended portfolio drift |
| Customer protection | Full cost, withdrawals and complaint access | No person accountable when model errs |
Read the sixteen chapters
1. Verify the platform’s actual regulatory authorisation
An investment app is not a legitimate robo-adviser simply because it calls its model intelligent or displays a professional dashboard. A financial firm recommending or managing investments must operate under the regulatory permissions appropriate to its real activities. In Singapore, MoneySense directs users to check robo-advisers’ MAS regulatory status.
Confirm the contracting legal entity and services against the MAS Financial Institutions Directory. Understand whether the firm supplies personal investment advice, discretionary management, execution or information only. Regulation is not a promise of profit; it establishes a framework for accountable provision of permitted services.
2. The questionnaire must reflect the client’s actual objective
A client saving for retirement thirty years away can consider investment risks differently from someone saving for a tuition payment next year. A robo-adviser’s questionnaire should ask about goals, time horizon, financial situation and ability or willingness to bear loss, within the relevant regulatory and product rules.
The platform must not treat an unanswered or contradictory question as a confident sign of high risk tolerance. Review whether the collected data is sufficient for the recommendation and whether customers understand what they are being asked. If the system cannot handle the situation, it should identify a suitable alternate route rather than invent a precise score.
3. Distinguish willingness to take risk from capacity to absorb it
A confident investor may be comfortable seeing a large market decline but still need the money for a near-term commitment. Conversely, another investor may be financially able to withstand volatility but strongly prefer stability. Treating either issue as the entire risk profile can produce unsuitable portfolios.
A credible algorithm considers relevant objectives, horizon, available resources, commitments and product characteristics. Where the required facts are missing, it should not automatically recommend the most commercially valuable portfolio. The MoneySense advisory process explains why needs analysis matters for both human and digital advice.
4. Translate the score into an explainable asset allocation
A platform might describe an investor as moderate risk and recommend 60% shares and 40% bonds. That percentage alone does not explain the currency exposure, markets, bonds’ duration or the consequences of a recession. Different portfolios using the same broad label can behave differently.
Explain which underlying funds or exchange-traded funds are selected, how their risks relate to the questionnaire and what the portfolio is intended to achieve. A model should have a documented mapping between client categories and permitted assets. A customer should be able to distinguish diversification from a guarantee that losses cannot occur.
5. Control algorithm changes as investment decisions
A software update might change risk scores, target allocations or how questions are interpreted. That is not merely a user-interface improvement; it can alter customers’ holdings and investment risks. A provider should not push a new algorithm into production without considering the financial consequence for existing users.
Apply design approval, independent review where appropriate, versioning, scenario tests and rollback controls. Record which model version produced each material recommendation. Check the impact on different legitimate customer profiles before broad release. A firm remains accountable for what its algorithm does even when nobody manually chose each portfolio.
6. Know where the invested assets are actually held
The robo-adviser may recommend a portfolio and execute investment transactions while a broker, custodian or other authorised institution holds the assets through its appropriate arrangements. The customer’s app balance is not the legal record of every security entitlement.
Explain the custody structure, account type, safeguarding and the responsibilities of underlying parties. A robo-adviser’s corporate failure and a decline in the market value of ETFs are different risks. Singapore deposit insurance does not generally insure market losses on investment funds. The Securities Custody Gold Standard develops the recordkeeping implications.
7. Rebalancing is an investment transaction, not automatic improvement
A portfolio can drift from its target weights as market prices change. A robo-adviser may sell one fund and purchase another to restore a chosen allocation according to predefined rules. This can maintain the intended risk exposure, but it can also create trading costs, timing risks and other consequences.
Document when rebalancing is triggered, which constraints apply and how the firm handles markets that are closed or unusually illiquid. Explain the rationale to customers without suggesting that automatically buying falling assets makes a positive return certain. MoneySense describes rebalancing as a typical robo-adviser feature, not an assurance against losses.
8. Show the true combination of platform and underlying fund fees
An advertised platform management charge is not always the full cost. Underlying ETFs or unit trusts have their own ongoing expense ratios, and the customer may face other transaction, currency or account charges depending on the service. Even modest annual fees can affect long-term compounding.
Show the platform fee, weighted underlying fund costs and material additional charges in comparable terms. For example, a purely illustrative 0.35% platform charge plus 0.15% weighted fund expense equals 0.50% per annum before other possible costs; on S$10,000 that simple starting-value illustration is about S$50. The actual fee calculation depends on balances and provider terms.
9. Compare performance on an appropriate consistent basis
A robo-adviser may show model portfolio histories, backtests and actual customer-account returns. These are not interchangeable. A favourable hypothetical backtest cannot be sold as if a customer really experienced it. Currency conversion, platform fees and investment contributions can also affect the comparison.
State the observed period, benchmark, asset mix, source and net or gross basis. Use the correct treatment of investor deposits and withdrawals. The Investment Performance Measurement Gold Standard explains why account-balance growth can differ from actual investment returns.
10. When a client changes goals, the recommendation must be revisited
A person may need to withdraw funds earlier because of family needs or a changed property purchase date. Their original risk questionnaire and asset allocation may no longer fit. An app that continues automatic investing without recognising the new horizon can leave the investor exposed to an avoidable mismatch.
Offer a controlled process to update circumstances and receive a revised recommendation or appropriate warning under the service’s remit. Record consent to material portfolio changes where required. Do not silently treat a profile change as authority to trade every holding without explaining what the new instruction means.
11. Know when a human adviser or specialist is more appropriate
Digital advice can be useful for straightforward goals, regular investments and comparatively standard asset allocations. Complex estates, business ownership, insurance needs or substantial foreign assets may require different expertise. The existence of a robo-adviser does not remove the value of qualified human judgement.
MoneySense distinguishes typical robo-adviser use cases from more complicated situations requiring broader advice. A gold-standard service recognises when its structured questionnaire is insufficient and offers a legitimate referral or clear limitation rather than producing a recommendation beyond its competence.
12. Protect account access and financial profile data
Robo-advisers may collect details about savings, investment holdings, income, dependants or financial commitments. This information can be sensitive even when the portfolio trades only publicly available funds. Unauthorised changes to withdrawal destinations or risk profiles can have consequences beyond a simple password problem.
Apply appropriate authentication, access logging, data minimisation and controls over sensitive changes. Explain which service providers handle information and for what purpose. A user should have a trustworthy route to recover access after losing a device without handing credentials to an impersonator. Secure onboarding and safe later account management must work together.
13. Withdrawal and transfer requests need realistic liquidity information
An investment portfolio is not the same as a bank savings account. Selling a fund can require a trading session, settlement or redemption period and possibly currency conversion before usable cash reaches the client. A platform should not promise instant withdrawal merely because its dashboard updates instantly.
Show the eligible amount, sale status, fees and realistic steps until cash reaches the nominated account. Verify account details and explain restrictions. When a withdrawal fails, establish whether the issue lies in the portfolio trade, cash settlement, banking route or customer instruction. Financial completion must be more important than a cheerful redemption message.
14. Stress-test portfolios and customer journeys during a market fall
Suppose stock markets fall sharply and many investors simultaneously review or sell holdings. The algorithm may recommend rebalancing while trading venues are volatile, prices are delayed and customer support is overloaded. The provider needs to ensure the model and operations work as intended under unusually difficult conditions.
Test market scenarios, stale data, failed orders, incorrect account valuations and heavy withdrawal traffic. Communicate honestly about risk and what remains uncertain. A platform cannot promise to avoid losses but can preserve accurate records, functioning controls and support that remains reachable when demand rises.
15. A fictional home buyer receives the wrong risk profile
A customer tells an app that a housing deposit will be needed in three years, but an overly simple questionnaire gives the same score as a thirty-year retirement investor. The recommended portfolio contains an equity allocation the customer is not comfortable risking before the purchase. The client sees a reassuring risk label but not its assumptions.
A stronger service checks contradictory answers, makes the time horizon prominent and records why the recommendation fits the user’s actual purpose. A qualified alternative may be needed if the system cannot determine suitability. The example is fictional, not a specific investing recommendation.
16. The gold standard is responsible automation with visible ownership
The service should measure unsuitable recommendation exceptions, unexplained model changes, incorrect risk-profile mapping, fund-fee disclosure errors, stale prices, failed rebalancing, delayed withdrawals and unresolved complaints. Subscription growth or customer app engagement is not enough to establish quality.
Review real client outcomes within appropriate privacy controls. An algorithm may produce thousands of recommendations, but a regulated business remains accountable for the advice and transactions it provides. Gold-standard robo-advisory means meaningful suitability, verifiable execution, honest reporting and an accessible repair path—not algorithmic certainty.
Robo-adviser questions
Do robo-advisers guarantee positive returns?
No. They invest in financial markets, so values may rise or fall. Diversification and rebalancing do not remove investment risk.
Are robo-advisers regulated in Singapore?
Providers undertaking regulated robo-advisory activities require the appropriate MAS regulatory authorisation; verify the actual firm in official registers.
Does the platform fee include every investment cost?
Not necessarily. Underlying funds and transactions may incur additional costs.
Where are the securities held?
Through the relevant custody or account arrangements, which the firm should disclose and explain.
Is a robo-adviser suitable for complex estate and family planning?
It may be insufficient. A qualified human adviser may be appropriate for multifaceted financial circumstances.
Robo-advisory governance audit
- Verify the platform’s actual MAS-authorised activities.
- Test whether the customer risk-profile questionnaire detects inconsistent answers.
- Reconstruct the algorithm’s mapping from client facts to portfolio selection.
- Review change control and validation for allocation or model revisions.
- Explain all platform, fund, trading and foreign-exchange costs.
- Reconcile trades, custody, valuations and actual withdrawals.
- Provide human-led review where the automated process cannot determine a suitable outcome.
Current Singapore sources and eduKate ecosystem links
- MoneySense: Robo-Advisers versus Traditional Advisers (September 2026).
- MoneySense: Financial Advisory Process and Needs Assessment.
- MoneySense: Authorised Financial Advisers and Robo-Advisers.
- MAS: Digital Advisory Framework and Regulatory Sandbox Distinction.
- MAS Financial Institutions Directory.
- eduKate: Asset Management.
- eduKate: Investment Performance Measurement.
- eduKate: Securities Custody.
Financial-services educational content only; not a personal investment recommendation. Actual portfolio risk, regulatory status, fees, liquidity and custody protections depend on the provider and product.