How LensVest reaches a conclusion
A plain-English account of what evidence we use, how we test it, and why a result is only ever as strong as its weakest important piece of evidence.
Read the evidence before the mechanics
Every LensVest report starts with the concepts that decide how strongly an ETF story can be stated. The technical audit trail is still available, but these boundaries come first.
01
Evidence status
The evidence status is the ceiling for the public claim. If the checked evidence is fragile, proxy-limited, or research-only, the page keeps the wording inside that boundary instead of raising it into a stronger claim.
02
Proxy history
Proxy or index history can add context before an ETF existed, but it is not the same as actual product history. LensVest keeps that distinction visible so readers know which part of the story comes from the fund itself and which part is only supporting context.
03
Rolling windows
Rolling windows test whether a claim depends on one flattering start and end date. The same comparison is checked across many overlapping periods so endpoint luck does not carry the whole conclusion.
04
France/PEA friction
France and PEA checks include wrapper, contribution, whole-share, broker, and fee assumptions when the reviewed inputs support them. Missing or unknown local details remain evidence boundaries rather than universal broker or tax coverage.
Accountability is in the review record
LensVest explains who owns the result through the records it keeps: checked inputs, visible limits, deterministic output, and manual scope review before any commercial step.
Checked inputs
The exact claim, instruments, country, wrapper, contribution pattern, broker profile, fees, and source basis are checked against what the deterministic review can actually support.
Recorded assumptions and dates
Assumptions, source dates, and review dates stay attached to the report so readers can see when the evidence was checked and which setup the conclusion belongs to.
Unavailable evidence stays visible
Missing history, unsupported broker details, or incomplete source coverage stay visible as evidence boundaries. They are not hidden behind a stronger conclusion.
Deterministic output
The same checked inputs and assumptions produce the same verdict wording, evidence boundaries, and reader sections.
Evidence boundaries
Evidence status, safer wording, and supportability meaning stay tied to the checked evidence. The published page does not upgrade or reinterpret them.
Human scope review
A reviewer checks supportability and evidence boundaries before any paid scope, checkout, or delivery promise. If the evidence cannot support the requested scope, the scope is narrowed or declined.
Supportability review comes first.
A request can become paid work only after LensVest checks the supportability and evidence boundary. That review happens before paid scope, checkout, or delivery promise.
The evidence we use
We distinguish clearly between real fund history and stand-ins. The kind of evidence is always part of the result.
Actual ETF history
Live, post-inception prices from the real fund.
Listed-fund proxy history
A comparable listed fund used as a stand-in, clearly labelled.
Index proxy history
The underlying index before the fund existed.
Synthetic estimates
Modelled series where no listed data exists, flagged as modelled.
User-provided data
Data you supply, used as given and attributed to you.
The tests we run
The same claim is examined across periods, regimes, and realistic execution — not a single best-case window.
Rolling-window tests
Rolling three-year and five-year periods, not one flattering full-period figure.
Crisis-window tests
Behaviour during major market-stress periods present in the available data.
Monthly contribution simulation
Regular contributions, integer shares, and residual cash, where they matter.
Broker and fee assumptions
Stated, conservative assumptions for fees and friction — never silently omitted.
“A result can only be as strong as its weakest important evidence.”
If the data behind a claim is thin, proxy-based, or period-dependent, the conclusion is held back accordingly — and we say why.
Source dates and review dates are recorded
Every input carries the date its data was sourced and the date the review was produced, so a conclusion can be placed in time.
Unavailable tests stay visibly unavailable
If a test cannot be run — missing history, insufficient data — it is shown as unavailable rather than silently dropped.
The result is deterministic
The final result is based on recorded inputs, assumptions, source versions, and test outcomes. The same inputs produce the same conclusion.
Reproducibility
Because inputs and assumptions are recorded, a review can be reproduced and audited rather than taken on trust.
Reference library
What you can read today, and what we're working on next. We will not publish numbers or articles the evidence does not support.
Public report library
Free LensVest evidence checks showing verdicts, evidence boundaries, and safer wording from reviewed public reports.
Planned research
PUST vs CW8 since inception
Live history vs index-proxy extension, across rolling windows and crises.
Leveraged ETF backtest accuracy
How proxy and synthetic history change a leveraged-ETF conclusion.
How to verify an ETF backtest
A checklist for inception dates, proxy data, fees, and period dependence.
Want a specific claim covered next? LensVest checks supportability and evidence boundaries before paid scope, checkout, or delivery promise.
Request a ReportWant this applied to your claim?
Request a Report and LensVest first checks supportability and evidence boundaries before paid scope, checkout, or delivery promise.