Research Note / Operational Due Diligence
How to Verify a Quant Track Record: A Due-Diligence Checklist for Allocators
A working checklist for allocators: which documents to request, how to reconcile statements, how to test a backtest and how to read live evidence before capital scales.
- Mandate focus
- Track record verification and quant fund diligence
- Audience
- Allocators, family offices, diligence committees
Verifying a quant track record is a procedure, not a judgement call. This note sets out the procedure as a working checklist: what to ask for, how to reconcile what arrives, how to test a backtest that has no live record behind it, and how to read live evidence before capital scales. It builds on the companion note on what a verified quant track record means, which defines the term. It is educational, written for professional investors, and is not an offer or solicitation.
- Verify the numbers against sources the manager does not control: administrator statements, broker statements and trade data.
- Reconcile before you analyse. A return series that does not tie to statements is not worth testing.
- Treat a backtest as a hypothesis. Ask for the trial count, the validation method and the data never used.
- Look for live evidence, and compare live fills with what the simulation assumed.
1. Define exactly what is being claimed
Start by splitting the record into segments and labelling each one. Which months are live and which are simulated? Which account, vehicle or entity produced them? Was the team the same throughout, and was the strategy the same strategy being offered today?
Records that blend segments are common: a backtest followed by a short live period, or a track record from a prior employer's vehicle. Neither is disqualifying, but each segment carries a different weight of evidence, and the blend should never be presented as one number.
2. Request the documents
The request list below covers the sources that matter. Ask for each of them at the start. A manager who can produce them quickly has usually built the record to be verified.
| Document | What to ask for | What to check |
|---|---|---|
| Administrator NAV statements or NAV letters | Official statements from an independent administrator, covering the full period claimed, with assets under management confirmed. | Dates, reporting entity and AUM agree with the performance sheet. An audit is preferred where one exists. |
| Prime broker statements | Daily or monthly statements obtained directly from the broker, or exported by you from read-only access. | Month-end equity, deposits and withdrawals tie to the reported return series. |
| Timestamped trade blotter | Every execution with time, price, size and venue, plus the order that generated it. | Trade-level profit and loss sums to the statement. Fill prices sit inside the market at the timestamp. |
| Research file | Data sources and cleaning rules, the number of variants tested, the validation method and the period never used in research. | The trial count is stated and plausible, and the held-out period was genuinely untouched. |
| Fee schedule and account agreements | Management and performance fee terms, the high-water mark and the authority granted to the manager. | Reported returns are net of the fees you would actually pay. The manager has trading authority only. |
| Broker and investor references | Names of counterparties and existing investors who can speak to the account. | References respond independently and describe the same strategy, size and period. |
3. Reconcile the numbers
Reconciliation turns documents into evidence. Work through it in order, and stop at the first break.
- Recompute monthly returns from statement equity, adjusted for deposits and withdrawals, and compare them with the manager's series.
- Confirm the start and end dates, and look for missing months or a period that has quietly been excluded.
- Confirm that fees are taken as the agreement states: management fee on assets, performance fee only on net new profits above a high-water mark.
- Sum trade-level profit and loss from the blotter and compare it with the statement.
- Check that the account size at the time is consistent with the strategy's stated capacity.
Prefer statements you obtain yourself over statements the manager forwards. In a separately managed account the allocator has direct access to the broker's portal and daily statements, which makes this step independent by design.
4. Test the backtest
Where part of the record is simulated, the question is how many chances the research had to find a flattering result. Ask for the number of variants tried, the provenance of the data, and the validation method. Walk-forward testing alone gives a single path. Purged and embargoed cross-validation, set out in combinatorial purged cross-validation, gives a distribution, and a Deflated Sharpe Ratio adjusts for the trials.
Then ask what the research has never seen. A strategy validated only on the data used to design it has not been tested. The note on the blind out-of-sample air-gap describes one way to keep a window of history untouched, and placebo testing describes how to check that a signal disappears when its source is removed.
5. Check execution realism
A simulation that fills every order at the midpoint overstates what a strategy can earn. Compare simulated fills with timestamped live fills, and look at slippage, latency and fees. The mechanics are covered in microstructure feasibility. For futures strategies, also check the denominator: returns measured on capital that margin requirements would have consumed are inflated. The note on managed futures and CTA strategies explains the margin constraint.
6. Grade the evidence
A simple ladder helps a committee record where a program stands. It is an illustrative working scale, not an industry standard, and a higher level does not imply a better strategy, only better evidence.
| Level | Evidence | What it supports |
|---|---|---|
| Level 0 | Claim | A figure in a deck with no source behind it. |
| Level 1 | Backtest | A simulation under stated assumptions. Informative about logic, silent about fills. |
| Level 2 | Administrator NAV | An independent party confirms the vehicle's value and assets over a period. |
| Level 3 | Broker statements | The clearing broker's own records support the equity curve for one account. |
| Level 4 | Tick-level trade match | Timestamped executions reconcile to statements and show realistic slippage. |
| Level 5 | Live pilot | Real capital observed before scaling, with live fills compared against the simulation. |
7. Check operations and custody
A verified record does not make a program safe to fund. Confirm who holds the assets, what authority the manager has, who the regulator and counterparties are, and which risk limits are enforced by systems rather than by policy. An allocator should be able to see that the manager can trade the account but cannot withdraw, transfer or encumber capital. The structural comparison is in the Cayman master-feeder note, and the broader audit list in the 38-point forensic due diligence framework.
8. Red flags worth escalating
- Performance quoted without a source the manager does not control.
- A backtest presented without the number of variants tried.
- Returns measured on capital the strategy could not have deployed once margin is counted.
- A track record from a different vehicle, team or strategy than the one being offered.
- Statements supplied only as manager-edited spreadsheets or screenshots.
- No stated capacity, or a claim that capacity is unlimited.
9. How Qlumina applies the process
Qlumina, a BVI FSC approved investment manager, admits a program to the Apex Marketplace only after a quantitative review, an operational review and a live pilot allocation. Manager evaluation includes algorithmic verification of timestamped fills, with checks for overfitting, hidden regime bias and tail risk, and portfolio allocations and risk parameters require human fiduciary sign-off. Where references are available, they are required and verified.
The diligence record stays attached to the program, and program statistics unlock after investor verification. Managers can see what is examined on the portfolio managers page and in the diligence FAQ. Allocators can browse the Apex Marketplace or the program catalogue, and read how evidence and fees are explained in the investor guide. For the network side, see how allocators find and connect with systematic managers.
Frequently asked questions
What does a verified quant track record mean?
A verified quant track record is one whose figures can be traced to sources the manager does not control, such as administrator NAV statements, direct prime broker statements and timestamped trade data. A return stream that exists only in the manager's own spreadsheet is a claim rather than a record.
Is a backtest a track record?
No. A backtest shows what a model would have done under stated assumptions. It does not show real fills, and it is exposed to overfitting and look-ahead bias. A track record starts where the simulation ends, with real capital in a real account.
Do all managers need audited track records?
Audits are strongly preferred but not always available, particularly for smaller funds. In that case official NAV letters from an independent administrator that also confirm assets under management are an accepted alternative, supported by broker statements and references.
How long should a live pilot last?
There is no universal answer. For strategies with limited capacity, Qlumina's diligence process often uses a pilot of 90 to 180 days with real capital to compare live performance with the historical record or backtest.
What does a quant fund diligence platform do?
A quant fund diligence platform collects evidence from source, keeps it attached to each program and asks the same structured questions of every manager, so that programs can be compared on the same basis instead of through different decks.
Reconcile first, then test, then watch it live.
A track record earns trust in a fixed order: the numbers tie to independent sources, the backtest survives questions about how it was built, and live capital behaves as the simulation said it would. Skipping a step moves the risk from the manager's process to your own capital.
Investing involves substantial risk of loss. The value of investments and the income derived from them can fall as well as rise, and investors may not recover the amount originally invested. Past performance is no guarantee of future returns. This note is for professional investors only and is not financial, legal, tax or investment advice. See the risk disclosure.
Diligence Attached
See programs admitted on evidence
Every program on the Apex Marketplace carries its diligence record. Professional investors can browse the live programs and request access.
Research notes are published for professional readers. Not an offer or solicitation. Risk disclosure
