Choose ETF data for the question you are testing
“ETF data” is not one interchangeable dataset. Choose the field that answers the question, then check its definition, timing and rights. A fund’s holdings file cannot replace its total-return history.
Match the field to the task
| Input | Can help answer | Does not establish |
|---|---|---|
| Traded market price | Historical quoted/traded value | Reinvested total return by itself |
| NAV | Value of fund assets less liabilities | An investor’s executable fill |
| Distribution event | Cash amount and relevant dates | Complete total return without policy |
| Holdings snapshot | Portfolio composition at its stated date | Holdings known continuously |
| Total-return index | Return with specified reinvestment | Real fills, taxes or investability |
Investor.gov’s ETF overview and NAV definition explain the basic distinction. For a vendor series, its own documentation must establish the actual construction.
Worked example: an ex-distribution date
Use an artificial fund priced at 100 before a distribution. Suppose it pays 2 and its price becomes 98, with no other economic change. A bare price return is -2%; cash plus the final holding equals 100 before costs. This is a teaching identity, not a real fund event.
If a supplied index already reinvests distributions, adding the 2 again double-counts it. If only bare prices are available and distribution treatment is unknown, do not label the experiment full total return. S05’s first contract asks for an already documented total-return sequence; it does not repair arbitrary corporate actions.
Check timing separately
A holdings snapshot describes a portfolio date, while its publication can be later. For a historical decision experiment, preserve both. A current holdings download cannot prove what was known on a past trade date. Similarly, a daily NAV is not a guarantee that all components shared the same tradable close.
Write the data request
Ask for fixed asset identities, one currency and calendar, date range, series basis, distribution policy, source version, permitted use and known gaps. This specification is useful even before buying or downloading anything. The initial CoTrading workflow does not require purchasing new data.
Limitations and verification
An input may be suitable for an explanatory chart yet fail a strict timing test. Personal access does not imply commercial redistribution rights. Validate the actual long CSV and manifest using the preflight checklist; do not infer correctness from a familiar ticker or provider logo.
FAQ
Can a holdings snapshot drive a point-in-time backtest?
Only with documented availability and an appropriate historical construction, not by assuming its portfolio date was public.
Should distributions be added to a total-return index?
No, when they are already included by its documented construction.