Point-in-Time Accuracy
Financial backtests fail when data is available too early — you use information that wasn't publicly known at the time. Valuein timestamps every fact with accepted_at: the exact moment the SEC accepted the filing. Filter by it and your backtest is safe.
Why Most Databases Introduce Look-Ahead Bias
Most financial databases store data as it exists today, not as it was known historically. A fiscal year 2021 annual report filed in March 2022 is often backdated to December 31, 2021 — making it appear as if that data was available before it was. Backtests built on this data are invalid.
Timeline for a 2022 10-K Filing
Fiscal period ends. Results for the full year are computed internally. The public knows nothing yet.
Company submits the 10-K to SEC EDGAR. Still not indexed in full — processing occurs over hours.
SEC accepts and timestamps the filing. This is the earliest moment any investor could have seen this data. Filter by this.
The hidden trap
If a data vendor stores this filing against fiscal_year = 2022 with no timestamp, a backtest that says "use 2022 annual data as of Jan 1, 2023" will include it — but the filing wasn't accepted until Feb 15, 2023. Your simulated portfolio used information that didn't exist yet. This introduces look-ahead bias and inflates backtest performance.
The Three Key Fields
accepted_atThe exact UTC timestamp the SEC accepted the filing that disclosed this fact. Each fact row inherits accepted_at from its parent filing on indexing, so the filter works identically on either table. This is your PIT anchor — use it exclusively for backtest-safe queries. It represents the earliest moment any investor could have read this data.
Always filter: WHERE accepted_at <= your_date
filing_dateThe date the SEC received the filing. Very close to accepted_at but lacks the exact time component. Suitable for rough date-range filtering but accepted_at is more precise for PIT analysis.
Safe for range filtering, less precise than accepted_at
report_dateThe fiscal period end date (e.g. December 31 for a calendar-year company). This is NOT a PIT field — using it as a filter introduces look-ahead bias because the data wasn't known until the filing date weeks or months later.
For display purposes only — never use as a PIT filter
Wrong vs. Right Queries
The difference between a biased and a valid backtest often comes down to a single WHERE clause.
-- WRONG: look-ahead bias introduced
-- This returns data as if you knew it on Jan 1 2022,
-- but 10-K filings for fiscal year 2021 weren't published
-- until Feb–March 2022. You're using future information.
SELECT
r.symbol,
fa.numeric_value / 1e9 AS revenue_billions
FROM references r
JOIN filing f ON f.entity_id = r.cik
JOIN fact fa ON fa.accession_id = f.accession_id
WHERE fa.standard_concept = 'Revenues'
AND f.fiscal_year = 2021 -- WRONG: fiscal year is NOT when data was known
AND f.form_type = '10-K'
ORDER BY revenue_billions DESC;-- RIGHT: point-in-time safe using accepted_at
-- Only returns data that was publicly available on 2022-01-01.
-- If a company filed its 2020 10-K late (e.g. Feb 2022),
-- it will NOT appear in this query -- correct behavior.
SELECT
r.symbol,
fa.numeric_value / 1e9 AS revenue_billions,
fa.accepted_at -- visible timestamp
FROM references r
JOIN filing f ON f.entity_id = r.cik
JOIN fact fa ON fa.accession_id = f.accession_id
WHERE fa.standard_concept = 'Revenues'
AND f.form_type = '10-K'
AND fa.accepted_at <= '2022-01-01' -- RIGHT: PIT filter
ORDER BY revenue_billions DESC;Survivorship Bias
Look-ahead bias is temporal — using today's data in the past. Survivorship bias is structural — only analyzing companies that still exist today. Both inflate backtest returns and both are invisible unless your dataset is specifically built to prevent them.
Delisted companies
Valuein tracks all entities including those that were delisted, acquired, or went bankrupt. The Pro and Institutional tiers include 19,000+ entities — active and inactive.
Historical index membership
The index_membership table records exact effective_date / removal_date for each company in each index, with [) interval semantics. A 2010 S&P500 backtest uses the 2010 constituents, not today's.
PIT universe construction
Use get_pit_universe(as_of_date) to reconstruct the exact investable universe on any historical date — free of additions that happened after.
-- Build a survivorship-bias-free universe for March 2020
-- This returns exactly who was in the S&P500 on that date --
-- before COVID additions/removals, before failures, before mergers.
SELECT
cik,
ticker,
name,
sector
FROM get_pit_universe(
as_of_date => '2020-03-01',
index => 'SP500'
);
-- WRONG alternative (survivorship bias):
-- Using the current S&P500 list for 2020 data excludes
-- companies that were dropped and includes companies that
-- didn't exist in the index yet.PIT in the Python SDK
Every SDK method that returns time-series data accepts an as_of_date parameter. Pass it to transparently filter by accepted_at.
from valuein_sdk import ValueinClient, ValueinError
try:
with ValueinClient() as client:
# PIT-safe: only data known as of the backtest date
df = client.run_query("""
SELECT r.symbol, fa.fiscal_year,
fa.numeric_value / 1e9 AS revenue_bn,
fa.accepted_at
FROM fact fa
JOIN references r ON fa.entity_id = r.cik
WHERE r.symbol = 'AAPL'
AND fa.standard_concept = 'TotalRevenue'
AND fa.fiscal_period = 'FY'
AND fa.accepted_at <= '2023-01-01' -- PIT filter
ORDER BY fa.fiscal_year DESC
LIMIT 10
""")
# All rows have accepted_at <= 2023-01-01
print(df[["fiscal_year", "revenue_bn", "accepted_at"]])
except ValueinError as e:
print(f"Error: {e}")Frequently Asked Questions
Ready to build a PIT-safe backtest?
Start with the free sample tier — all PIT fields are included. Upgrade for full S&P500 history or the complete 19,000+ ticker universe.