CFA Level II Exam · Guidance for Standard V: Investment Analysis, Recommendations, and Actions

Quantitative Models, Group Research and Investment Decisions under Standard V(A)

Updated 7 October 2026 · Fact-checked

Under Standard V(A), you must have a reasonable basis for any recommendation or action. With quantitative models, you must understand the inputs, assumptions and limits and test them. In group research, if you believe the view lacks a reasonable basis, you should dissent and ask that your name be removed. You may rely on the group view only if you judge it sound.

Understand Quantitative Models, Group Research and Investment Decisions

Standard V(A) says you must exercise diligence, independence and thoroughness, and have a reasonable basis and support for any investment analysis, recommendation or action. The Standard does not care whether the work came from your own spreadsheet, a vendor model or a team report. You remain responsible for what you act on.

Quantitative models include valuation models, screens, risk models and credit scoring tools. Before relying on one, you should understand what it does, what data it uses and what assumptions sit under it. You should know where it fails. For example, a model built on a calm market history may understate risk in a crisis. You should test the model, including stress tests and scenarios outside the normal range. A model output is not a reasonable basis just because it came from software.

This is different from relying on secondary or third-party research. There, you judge the provider: its reputation, its assumptions, its rigor and its track record. With a quantitative model, you look at the mechanics: the inputs, the logic and the limits. In both cases you should make a reasonable effort to check. Review is needed at the start and again from time to time. The more you rely on it and the more important the decision, the more checking you should do.

Group research is produced by a team. Not every member has to agree with every point, and a member is not required to dissent from a view they accept. But if you believe the group view lacks a reasonable basis, you should dissent and ask that your name be removed from the report. Continuing to be associated with a view you believe lacks a reasonable basis would be improper. A member who in good faith relies on a group view is generally fine, provided the member has no reason to doubt it and the group followed a diligent process. If you know the work is flawed, you cannot hide behind the group.

Firms should also have sound investment decision processes. Written procedures help show that decisions were reasoned and not reckless. In the exam, ask whether the person did the checking that fits their role, or whether they accepted an output without asking how it works.

Communication is a separate duty. Under Standard V(B), you must tell clients about the limitations of a model or process that matter to a recommendation. This sits alongside diligence under V(A). It does not replace it.

Key formulas to remember

Core duty: Reasonable basis = diligence + independence + thoroughness before any recommendation or action; Applies to own work, models, group views and third-party research alike.
Quantitative model rule: Understand inputs + assumptions + limitations; test, including stress and scenarios; You must be able to explain why you trust the output.
Group research rule: Believe the view lacks a reasonable basis → dissent and ask that your name be removed; You are not required to dissent from a view you accept. Staying associated with a view you believe unsupported is improper.
Reliance on third parties: Judge the provider; periodically review the process; Contrast with models, where you examine the mechanics.

How to solve Quantitative Models, Group Research and Investment Decisions questions

Use this order for any vignette on models, group research or decision processes.

1Identify the source of the recommendation: own analysis, a quantitative model, group research or an external provider.
2Find what the analyst actually did: did they review inputs, assumptions and limits, or simply use the output?
3Check for warning signs in the vignette: untested assumptions, data from a narrow period, a known flaw ignored, or pressure to sign.
4Match the source to the right duty: model means understand and test; group means dissent and ask to be dissociated if you believe the view lacks a reasonable basis; third party means judge the provider.
5Decide whether the diligence was proportionate to the importance of the decision and the role of the person.
6Name Standard V(A) and, if the case touches it, communication under V(B), then choose the option that shows reasonable basis.

Quickest way: Three-question check

When to use it: When time is short and the three answer options differ mainly on whether the analyst did enough checking.

Did the analyst understand and test what they relied on?
If it is a group view the analyst believes lacks a reasonable basis, did the analyst dissent and ask to remove their name?
Pick the option that adds checking, testing or dissent. Reject options that rely only on the output, the reputation or the group. Disclosure of model limitations under V(B) is a good addition, but disclosure alone does not fix a missing reasonable basis.

Common mistakes in Quantitative Models, Group Research and Investment Decisions

Treating a model output as a reasonable basis on its own
Software feels objective and precise.
Fix: Ask whether the analyst understood the inputs and limits and tested it. Without that, the basis is not reasonable.
Saying a violation occurs only because the model was wrong
Students judge by outcome.
Fix: Judge the process. A diligent analyst can still be wrong; a careless analyst can be lucky.
Believing group members must support the group view
Team loyalty seems to override personal views.
Fix: A member who believes the view lacks a reasonable basis should dissent and ask that their name be removed.
Confusing model diligence with third-party research diligence
Both involve relying on others' work.
Fix: For models, examine mechanics and assumptions. For third-party research, judge the provider's process and reputation.
Assuming a member of a group who relies on it in good faith always violates the Standard
Students overstate personal responsibility.
Fix: Good-faith reliance is generally acceptable when there is no reason to doubt the work and the group process was diligent.
Thinking checking is needed once only
Students see due diligence as a one-time task.
Fix: Models and providers should be reviewed periodically, since markets and data change.

Worked examples

Example 1

Vignette: Mira, an analyst at a global asset manager, uses a vendor credit-scoring model to rate bond issuers. The model was built on ten years of data from a benign credit period. Mira knows this but runs it unchanged, because the vendor is well known and the output is quick to produce. She recommends several holdings based on the scores. Question 1: Has Mira most likely violated Standard V(A)? Question 2: What should she have done?

Show the solution

Source: a quantitative model, so the duty is to understand inputs, assumptions and limits and to test them.
Mira knew the model was trained only on a benign period, which is a limitation she ignored.
She relied on the vendor's reputation, which is the test for third-party research, not enough for a model.
She did not run stress tests or scenarios for weaker credit conditions.

Answer: Q1: Yes, she most likely violated Standard V(A) because she lacked a reasonable basis. Q2: She should have tested the model under stressed conditions and understood its limits. She should then have adjusted the output as needed and communicated the model's limitations under Standard V(B) when making the recommendation.

Example 2

Vignette: Daniel is on a five-person team that produces a research report on a consumer company. After review, he believes the revenue forecast uses an unsupported growth rate. The team lead says the group view is final and asks all members to sign. Question 1: What should Daniel do under the Standards? Question 2: Would Daniel violate Standard V(A) if he signs while believing the view lacks a reasonable basis?

Show the solution

Source: group research, and Daniel believes the view lacks a reasonable basis.
Members are not required to dissent from views they accept, but Daniel does not accept this one.
Daniel should dissent and ask that his name be removed from the report.
If he signs while believing the forecast has no support, he knowingly stays associated with a view without a reasonable basis, which is improper.

Answer: Q1: He should dissent and ask that his name be removed from the report. Q2: Yes, signing while believing the view lacks a reasonable basis would be a violation of Standard V(A).

Exam tips

Read for the source of the recommendation first. The right rule depends on whether it is a model, a group or a third party.
Look for the words that signal weak diligence: only, unchanged, without testing, because the vendor is reputable.
The correct answer often adds action: test, stress, review, dissent or remove the name. Disclosure of model limitations under V(B) is still required alongside diligence, but disclosure alone does not cure a lack of reasonable basis.
Judge process, not outcome. A profitable result does not cure a lack of reasonable basis.
Group items rarely ask you to ignore the group; they test whether you should dissent and ask to be dissociated when you believe the view is unsupported, and whether reliance was in good faith.

Quantitative Models, Group Research and Investment Decisions in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Quantitative Models, Group Research and Investment Decisions: frequently asked questions

What does Standard V(A) require for quantitative models?

You must understand what the model does, its inputs, assumptions and limits. You should test it, including with stress tests and scenarios. Only then can its output support a recommendation.

Do I have to dissent from a group research report?

If you believe the group view lacks a reasonable basis, you should dissent and ask that your name be removed so you are not associated with it. You are not required to dissent from a view you accept.

How is model diligence different from secondary research diligence?

With a model, you examine the mechanics: inputs, assumptions and limits. With secondary or third-party research, you assess the provider and its process, and review it periodically.

Am I responsible if a group member made an error I did not know about?

Generally not, if you relied on the group in good faith, had no reason to doubt the work and the group followed a diligent process. Once you know of a flaw, you must act on it.