Aipedia: Wealth managers and discretionary fund managers

AI portfolio construction and rebalancing

Aipedia, by the Apex broking team · Last reviewed 2026-10-30

Discretionary managers can use AI to construct and rebalance portfolios, but every trade must stay suitable and within mandate, and the firm is responsible for the tool. One wrong parameter can be applied across hundreds of client portfolios overnight.

Part of: AI in Wealth managers and discretionary fund managers

In short

AI and algorithmic tools now propose allocations and rebalance discretionary portfolios at scale. COBS 9A suitability, the Consumer Duty duty to avoid foreseeable harm and SYSC 8.1.6R on outsourcing all keep responsibility with the firm, and the Data (Use and Access) Act 2025 adds safeguards for solely automated significant decisions. A faulty rebalance can breach many mandates at once. PI usually responds to resulting negligence claims, subject to the wording, but check how it treats costs of correcting trades and whether many complaints aggregate into one claim.

How discretionary firms use AI to build and rebalance portfolios

Portfolio construction has long relied on rules-based systems: target weights, drift tolerances, cash buffers and restrictions coded per mandate. AI adds optimisation models that propose asset allocations, machine-learning tools that suggest trades to bring portfolios back to target at lower cost, and systems that apply those trades across hundreds of accounts at once. Some firms run the tools in-house; many use a vendor's order management or rebalancing platform with AI features switched on.

The appeal is speed and consistency. The risk is the same: whatever logic is wrong is applied consistently to every account it touches, and a discretionary firm does not need client sign-off before the trades are placed.

The duties that apply

Suitability: COBS 9A applies to discretionary management as MiFID business. Every rebalance is an investment decision that must remain suitable for the client and within the agreed mandate, whoever or whatever proposed it.
Consumer Duty: PRIN 2A.2.8R requires firms to avoid causing foreseeable harm, which includes harm from a tool the firm knows can drift out of mandate.
Conduct: COCON 2.1.2R requires due skill, care and diligence from the portfolio managers who approve or oversee automated trades.
Outsourcing and records: SYSC 8.1.6R keeps responsibility with the firm when the rebalancing engine is a vendor's, and SYSC 9.1.1R requires records of what was decided and why.
Automated decisions: section 80 of the Data (Use and Access) Act 2025 (Articles 22A to 22D, in force 5 February 2026) requires safeguards where a solely automated decision has a significant effect on an individual. A fully automated rebalance that materially changes a client's portfolio may fall within this; take data protection advice on your own process.

The FCA's October 2026 good-practice page adds that firms should name an owner for each AI use and monitor live outcomes.

What can go wrong

Common failure points are a mandate restriction that the tool does not read (an ethical exclusion or a ban on a single holding above a set weight), a risk model that underestimates correlation in stressed markets, a mapping error that assigns clients to the wrong risk band, and a data feed fault that makes cash or holdings look different from reality.

Illustrative scenario: a firm updates its rebalancing tool and a parameter for its cautious mandates is entered with the equity ceiling of its balanced mandates. The overnight rebalance raises equity exposure across every cautious portfolio. Nobody checks the trade summary because the tool has run cleanly for years. Markets fall over the following weeks and clients complain that their portfolios were managed outside the agreed risk level.

How PI usually responds

A claim that the firm managed portfolios outside the mandate is a classic professional negligence or breach of contract claim, and PI commonly responds where the policy is silent on AI, subject to its terms and excess. Several points need checking:

Correcting your own error: some policies treat the cost of putting portfolios back, before any client claims, differently from third-party claims. Check whether mitigation costs or trading error costs are covered.
Aggregation: many complaints from one parameter error may be treated as one claim with one excess and one limit, or as many. See aggregation of claims.
Market loss: PI covers loss caused by negligence, not ordinary investment underperformance within the mandate.

More detail is in does PI cover wealth managers' AI errors and algorithmic decisions and PI.

Practical controls

Run a pre-trade check against hard mandate limits that sits outside the AI tool.
Require portfolio manager approval for rebalances above set size or risk-change thresholds.
Test parameter changes on a copy of live portfolios before release, and log who approved them.
Reconcile post-trade risk levels to each mandate the next day.
Keep vendor change notes and your own release records; see AI vendors and outsourcing.
Review signals feeding the optimiser; see AI investment signals.

How Apex helps

Apex arranges PI for wealth managers and discretionary fund managers. A broker can review how your policy treats a single rebalancing error affecting many clients, including aggregation and costs of correcting trades. See the overview at AI in wealth management.

Sources

Frequently asked

Does COBS 9A apply if an algorithm made the rebalancing decision?

Yes. Suitability duties for discretionary management attach to the firm's decisions, however they are generated. The firm must be able to show each portfolio stayed suitable and within the mandate, and that a person with responsibility oversaw the tool.

Is an automated rebalance a solely automated decision under data protection law?

Possibly, if no meaningful human involvement occurs and the decision significantly affects the client. Section 80 of the Data (Use and Access) Act 2025 then requires safeguards. Whether your process qualifies depends on how it works, so take specific data protection advice.

Will PI pay to correct a rebalancing error before clients complain?

It depends on the wording. Some policies cover mitigation costs incurred to avoid a claim, often with insurer consent; others do not. Notify the circumstance promptly and agree the approach with insurers before compensating clients.

Check your PI covers how you use AI

Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start a PI quote Request a callback

Apex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.