Fund Fees, Expenses and Return History

Fee tables, expense examples and standardised return histories for US registered funds, taken from the XBRL riding on their prospectus filings: management fee, 12b-1 fee, other expenses, gross and net expense ratio, sales loads, portfolio turnover, and the 1-, 5- and 10-year average annual returns including the after-tax and benchmark rows. One row per share class per filing.

edgar_fund_risk_return - the dataset name to pass to the Obscura API.

What one row means

One **share class's fee table and standardised return history** as a US registered fund states it in one prospectus filing - the Risk/Return (`rr:`) inline-XBRL facts riding on a `485BPOS` or a `497`. Management fee, 12b-1 fee, other expenses, acquired-fund fees, gross and net expense ratio, the contractual waiver, sales loads, the SEC's $10,000 expense example at four horizons, the annual-return bar chart, best/worst quarter, and the 1/5/10-year and since-inception average annual returns including the two after-tax rows. Per share class, per filing, dated. **This is what makes `485BPOS` worth claiming.** The form's PROSE is 863 KB per filing → 62 GB across the corpus, past the exporter's practical ceiling and analytically almost worthless. The XBRL riding on the same filing is a few hundred facts. No prospectus text is stored on this table, and none ever should be: ~35 of the instance's 64 `rr:` tags are `…TextBlock` / `…Heading` / `…Narrative` prose containers (`RiskTextBlock` occurs 227 times in one sampled instance) and every one of them is dropped. # SERIES-LEVEL FACTS ARE NOT COPIED ONTO CLASS ROWS `PortfolioTurnoverRate`, `RiskReturnHeading` and the benchmark returns carry only the series dimension - MEASURED on both sampled instances. They live on a **class-less row** (`class_id = ''`), never fanned across the classes. Copying one portfolio-turnover figure onto four class rows would look like four measurements of the same thing, and averaging them would be wrong. The migration CHECKs it rather than trusting the writer. # Not every prospectus carries XBRL, and `497K` never does A `497K` summary prospectus carries no `rr:` facts at all (measured: zero on `0001193125-24-131423`), and a `497` may or may not - `0001137439-24-001011` (26 KB) has none, `0001213900-24-039977` (6.7 MB) has 3 series and 6 classes of them. A filing with no instance produces **zero rows, not null rows**, and is recorded done in the shared `edgar_doc_processed` marker so it is never re-examined. See `docs/datasets/edgar_fund_risk_return.md`.

One row per One row per (filing, series, share class) Risk/Return fact set, keyed by (accession, series_id, class_id). class_id is the EMPTY STRING on the SERIES-LEVEL row, which is where the facts whose XBRL context names only dei:LegalEntityAxis live - portfolio turnover, the risk/return heading and the benchmark returns. Facts whose context also names rr:ProspectusShareClassAxis live on that class's row. A combined prospectus covering 30 funds produces 30 series rows plus one row per share class. The key is the ACCESSION and not the archive path, for the same per-CIK fan-out reason as edgar_fund_series_classes..

Point-in-time availability

Every Obscura dataset carries available_date: the calendar day the publisher made the row available, day-of, with no session rounding. It is the one column a backtest filters on, and it means the same thing on every dataset in the catalog.

For edgar_fund_risk_return: filing_join_occurred: available_date is the EDGAR DISSEMINATION day from the bronze object's own day partition; occurred_on is rr:ProspectusDate, the day the prospectus and therefore this fee schedule TAKES EFFECT. It is frequently in the FUTURE relative to publication - T. Rowe Price 0001741773-24-001700 was disseminated 2024-04-25 and its rr:ProspectusDate is 2024-05-01 - so occurred_is_forecast carries that and the usual occurred_on <= available_date monotonicity does not apply to those rows. Both dates are stated by the publisher (one in the daily index, one as a tagged fact), so both are hard: evidence class A + B, CONFIRMED.

Refresh cadence

Obscura refreshes edgar_fund_risk_return every 6 hours - the most frequent scheduled job that re-collects or re-exports it. This is Obscura's own pipeline cadence, not the upstream publisher's release schedule; when a row became public is recorded per row in available_date.

Schema - 51 columns

The full public column list for edgar_fund_risk_return, with the meaning of every field. The same schema is served unauthenticated at https://api.obscura.trade/v1/catalog/edgar_fund_risk_return.

Access edgar_fund_risk_return

Two delivery paths, one identifier. Both require an Obscura account and an active subscription; the catalog entry and the schema above are public.

import obscura

client = obscura.Client("obs_live_…")

df = client.query(
    dataset="edgar_fund_risk_return",
    symbols=["NVDA", "AAPL"],
    start="2024-01-01",
)

Create a free account Browse all 122 datasets

Frequently asked questions

Related datasets