Due Diligence & Selecting Managers

IDD vs ODD

Why the separation matters. ODD is conducted by a team independent of the investment team and holds a veto, not a vote. The reason is behavioural: an investment team that has spent months building conviction is poorly placed to walk away over a control weakness. Independent reporting lines, a documented process, and a written veto right are the examinable safeguards.

The asymmetry. Investment risk is compensated — you accept it in exchange for expected return. Operational risk is not compensated at all. Therefore the correct posture is not "price it in" but "avoid it", and a fund with an excellent strategy and a failed ODD is simply not investable at any size.

The ODD checklist

Know the categories and at least two concrete tests within each — essays reward specificity.

Ongoing, not one-off. ODD is repeated: annual refresh, plus triggered reviews on key-person departure, service-provider change, terms amendment, regulatory action, a valuation methodology change, or a performance pattern inconsistent with the strategy.

Red flags, ranked

Order matters — a question asking for "the most serious concern" is testing your ranking, not your ability to list.

1. Self-administration and self-custody. The manager prices its own book, calculates its own NAV, and controls cash movement. This single configuration enabled the largest frauds in the industry's history, and it is the strongest possible signal.
2. Unknown, undersized or affiliated auditor relative to the fund's AUM, a qualified opinion, or frequent turnover of auditor or administrator without a convincing reason.
3. Returns implausible for the strategy. Very high Sharpe, almost no down months, and strong positive serial correlation in a supposedly liquid strategy. Smoothness is the fingerprint of either stale marks or fabricated ones.
4. Opaque or shifting valuation policy for hard-to-value assets; no independent pricing source; methodology changes that coincide conveniently with performance.
5. Secrecy justified as protecting the edge. Refusal to disclose positions even under NDA, to name counterparties, or to permit direct contact with service providers.
6. Style drift. Exposures inconsistent with the mandate, or returns explained by a factor the manager claims not to run. Detect it with factor analysis, not with the manager's commentary.
7. Key-person concentration with no succession plan, no key-person clause, and no evidence the process survives the founder.
8. Liquidity mismatch and a gating history — fund terms more generous than the underlying assets support; previous use of gates or side pockets without clear disclosure.
9. Rapid AUM growth beyond capacity, an incentive structure that rewards asset gathering, or a fee structure with weak alignment (no GP commitment, no high-water mark).
10. Regulatory, litigation or reputational history, undisclosed related-party transactions, and personal-account dealing without pre-clearance.

Valuation & service providers

The fair-value hierarchy. Level 1 — quoted prices in active markets for identical assets. Level 2 — observable inputs other than quoted prices (comparable transactions, quoted prices in inactive markets, yield curves). Level 3 — unobservable inputs, i.e. the manager's model. Any material Level 3 exposure makes valuation governance the central diligence question, because the manager's fee, its track record and its marks all depend on the same judgement.

What good looks like: a written valuation policy approved by a body independent of the investment team; a valuation committee including non-investment members; a third-party valuation agent for Level 3 positions, engaged and paid by the fund rather than by the manager; consistent methodology applied period to period, with documented rationale for any change; and an annual audit that specifically tests valuation.

Administrator. The independent administrator should strike NAV from its own records, not merely reconcile to the manager's file — "shadow accounting" versus "manager-provided NAV" is a distinction worth stating. Confirm the administrator's scope of work directly; some engagements are far narrower than the marketing implies.

Auditor. Check that the audit firm is real, appropriately sized for the fund, and actually engaged; that the opinion is unqualified; that financial statements arrive on time; and that the audited NAV reconciles to investor statements. Historic frauds have used an audit firm with a handful of staff for a multi-billion-dollar fund — a fact discoverable in an afternoon.

Custodian and prime broker. Assets should be held by a qualified custodian, legally segregated where possible, with rehypothecation rights understood and limited. For funds using leverage, understand the financing terms, the margin methodology, and what happens on a downgrade or a market shock.

Performance persistence & selection

The evidence differs by asset class, and the nuance is the answer. For hedge funds, persistence is weak and largely disappears once returns are adjusted for risk factors and once backfill and survivorship bias are removed; what persistence exists is concentrated at short horizons and among poor performers (bad managers reliably stay bad). For private equity, persistence was historically meaningful — particularly in venture, where access to the best deals is itself the edge — but the evidence has weakened as the industry institutionalised, funds grew, and teams turned over. Never answer with a flat "past performance predicts future performance"; state the asset class, the caveats, and the mechanism.

Why raw track records mislead. Short samples give low statistical power; the same manager's numbers are contaminated by market beta in a rising market; a fund's IRR is sensitive to subscription-line use and to timing; and databases suffer selection biases in both directions. The defensible approach combines quantitative decomposition (factor exposures, PME against a like index, dispersion across the manager's own deals) with qualitative assessment (team stability, whether the people who generated the record are still there and still doing the same thing, and whether the strategy's capacity has been exceeded).

Selection alpha and the allocator's own behaviour. Manager selection has wide outcome dispersion in private markets, which is where governance attention pays. But allocators also destroy value systematically by hiring after strong performance and firing after weak performance — the documented pattern where fired managers subsequently outperform their replacements. Defences: written selection and termination criteria set in advance, a minimum evaluation horizon, and a requirement that any termination decision cite a process failure rather than a performance number.

Manager monitoring after hiring: exposures against mandate, attribution against the stated process, team and ownership changes, AUM against capacity, terms and side letters, valuation methodology, and any operational event. Define in advance what constitutes a watch-list trigger and what constitutes termination.

Ongoing monitoring & termination

Diligence does not end at funding. The exam treats monitoring as a continuous re-underwriting of the original thesis, with a pre-agreed set of events that trigger review.

Trigger events that force an immediate review rather than a scheduled one: departure of a named principal, a change of auditor or administrator, a gate or side pocket, a valuation restatement, a regulatory action, a breach of stated risk limits, a strategy or fee change, or a drawdown beyond the underwritten range.

Termination discipline. Distinguish process failure from outcome disappointment — that distinction is the answer to most essay prompts here. Terminate for a broken thesis: the edge is gone, the team that generated it has left, exposures no longer match the mandate, or operational integrity is compromised. Do not terminate solely for a drawdown consistent with the strategy's known risk profile, because selection decisions made on trailing performance are systematically procyclical. Where the concern is capacity or fit rather than integrity, consider redemption over time, a mandate change, or a move to a separately managed account before outright termination — and always document the rationale, since the alternative is a board that relitigates the decision after the next quarter.

Confusion pairs

Practice

Six multiple-choice questions in exam style, with the reasoning — not just the letter.

Constructed-response practice

Write these under time. Each outline is the shape the rubric rewards, not a model answer to memorise.