On this page
A CAIA® Level II candidate evaluating cyber risk in investment operations should trace each critical activity through the manager and its service providers, then ask what happens if access, data or instructions are compromised. Strong defences on the manager’s own network do not establish that an administrator can restore records or that a payment change will be independently verified. The useful judgement is a specific control gap and its possible operational consequence, supported by evidence rather than a broad claim that a firm is “secure”.
Start with the investment operation, not the technology label
Cyber risk becomes material when a failure interrupts a process that the fund needs to value assets, settle trades, move cash or report to investors. A manager might have robust controls around its portfolio system while its administrator holds the authoritative investor register. An attacker need not alter a security price to create harm; corrupt records, unavailable systems or misdirected instructions can stop the operation. For an allocator, the first question is therefore which process could fail, who performs it, and what information or authority each party holds. That gives a map for questioning controls without pretending to conduct a technical penetration test.
The CAIA Association Level II companion explicitly places cybersecurity for fund managers inside risk measurement and asks candidates to evaluate vulnerabilities, preparedness and evidence that security functions occur regularly. That context matters: cybersecurity is an operational due-diligence problem, not a separate checklist of computer products. It invites a comparison between the promises in a policy and what the manager can demonstrate through ownership, tests, exception records and follow-up. A candidate should be able to explain why the same control can be adequate for one workflow yet irrelevant to another.
Map the trust boundary
Draw a simple chain from the investor or counterparty to the manager, the fund administrator, the custodian and any outsourced technology provider. At each hand-off identify who originates an instruction, who can change it, who approves it and who records the outcome. The organisation operating a system is not necessarily the organisation bearing the disruption. If an administrator processes a redemption instruction, its email security and verification procedure may matter more to that transaction than the manager’s password policy. Conversely, the manager may need the ability to suspend a workflow and reconcile the administrator’s records after an incident.
Consider three separate failure modes. A confidentiality failure exposes positions or investor data. An integrity failure changes a record or payment instruction. An availability failure prevents a system from operating when it is needed. One event can affect all three, but the best control question differs. Encryption of stored data addresses some confidentiality exposure; it does not prove that a false wire instruction will be challenged or that clean records can be restored. Identifying the failure mode prevents a candidate from matching a plausible security word to the wrong risk.
What evidence makes a control credible?
Ask who is accountable for the process, what the written procedure requires, how often it is exercised and what happened when an exception occurred. A manager may provide a policy, but an untested policy gives little evidence about its operation under pressure. Useful evidence could include a recent exercise record, a documented review of privileged access, a change log for payment details, a reconciliation record or a service-provider review with outstanding issues and owners. The exact document depends on the risk being evaluated. The point is to connect an observable action to the vulnerable workflow, not simply count documents.
For a provider with access to investor instructions, due diligence should test the boundary between its controls and the manager’s. Ask how the manager learns of a provider incident, whether it can independently confirm changed payment details, who can halt transfers and what alternative records it can use. An assurance report may help frame questions, but its period and scope must match the service and data at issue. It cannot by itself show that the manager’s own approval step was followed yesterday. A contract can assign duties, yet the allocator still needs to understand how those duties are monitored in practice.
The SEC’s Regulation S-P announcement provides a useful regulatory example, with a defined scope. It says covered institutions, including registered investment advisers and investment companies, must maintain written incident-response procedures for unauthorised access to or use of customer information. That is a specific customer-information rule for covered US institutions, not a universal duty imposed on every alternative fund worldwide. A candidate can use it to illustrate why detection, response and recovery are distinct from prevention while still checking the entity and jurisdiction before making a compliance claim.
An illustrative diligence decision
Suppose an invented fund manager has £500 million of assets and uses an administrator to maintain investor records. The manager demonstrates multifactor access to its own systems and an annual security review. The administrator sends a message requesting that a planned £2 million investor distribution go to a revised bank account. The payment team can make the edit from the email alone, and nobody calls a previously verified contact. These figures are illustrative and describe no real fund or incident. The analyst’s task is to locate the weak control, not calculate a market-risk statistic from the asset value.
Here the critical risk is instruction integrity at the hand-off. The manager’s own access control may be sound, but the payment process accepts a changed destination without an independent check. A sensible finding is that the bank-detail change procedure lacks verification through a previously established channel. The follow-up is evidence of the actual approval and callback process, including how exceptions are recorded, not a generic request for another firewall description. If the administrator’s account were compromised, an attacker could exploit the process even though the manager’s login controls never failed.
A second question is what happens after the suspicious message is detected. Who pauses the distribution, who contacts the administrator, who checks whether other instructions changed, and which record is authoritative? These are response and recovery questions. A manager that can describe prevention but cannot identify decision owners may lose valuable time while each party assumes another is acting. An allocator need not declare the manager uninvestable on the basis of this example. The defensible conclusion is narrower: a material gap has been identified and must be evidenced, remediated or explicitly weighed within the operational assessment.
Common errors in an exam answer
The first error is to treat cybersecurity as a yes-or-no property of the manager. A useful answer identifies an asset, threat path, control and consequence. The second is to infer that outsourced work means outsourced responsibility. The provider may operate the system, but the manager still depends on the result and needs a way to oversee the relationship. The third is to substitute a control in one category for a different one: secure access does not prove payment authenticity, and backups do not stop information leakage. State precisely what each control can and cannot address.
Another error is to cite a proposed regulation as if it were binding. The SEC’s June 2025 page states that its broad adviser and fund cybersecurity proposal was withdrawn. A current analysis should therefore distinguish that withdrawn proposal from the separately adopted Regulation S-P amendments. The distinction is especially important when a scenario does not state the manager’s registration status or jurisdiction. Begin with the operational analysis; add a legal point only where its scope and current status are supported.
A compact method for the candidate
When reading a case, write down the critical function, the party that performs it, the system or instruction crossing the boundary, the failure mode, and the evidence you would request. Then ask whether the stated control actually interrupts that path. If a case highlights a vendor, include the manager’s oversight and the route for escalation. If it highlights an incident, separate preventing access from discovering it, containing it and restoring operations. This method turns a long narrative into a testable control assessment without relying on fashionable terminology.
The CAIA programme page places this material in the wider course, while the methodology page describes the study approach. For revision, make the takeaway a short finding: “The administrator can alter payment details without independent confirmation, so a compromised message can redirect cash; request evidence of callback and approval controls.” That sentence names the exposure, mechanism and evidence. It is more useful than claiming the manager is cyber-secure or cyber-insecure without specifying what has been tested.
Sources
- 2026 Level II Companion | CAIA Association · caia.org · Retrieved
- SEC Adopts Rule Amendments to Regulation S-P to Enhance Protection of Customer Information | SEC · sec.gov · Retrieved
- Cybersecurity Risk Management for Investment Advisers, Registered Investment Companies, and Business Development Companies | SEC · sec.gov · Retrieved



