Operational and regulatory data on listed companies — car registrations, bank call reports, nuclear output, credit-card delinquencies, clinical trial slippage — mapped to tickers and dated by both when it happened and when you could first have known it.
What alternative data is, at ORTEX
Alternative data is information about a company that does not come from the company.
It comes from the regulator that supervises it, the register it has to file with, or the trust
it funds itself through. So it arrives on the publisher's schedule rather than the company's —
and usually well before the quarterly report that will eventually confirm it.
Three examples, all of them public, none of them from an investor-relations department:
- A nuclear operator does not announce an outage. The regulator publishes every US reactor's
output every morning. - A bank does not pre-announce a bad quarter. Its call report lands with the FDIC about 30
days after quarter end. - A lender's credit-card trust reports delinquencies and charge-offs monthly, months before
the lender itself reports earnings.
A surprising amount of the highest-value data in markets is free, structured and public.
What has been missing is somebody collecting it, mapping it to instruments, and serving it in
one shape. That is what these endpoints are.
Two clocks on every reading
Every value ORTEX publishes carries two dates, and the difference between them is the point.
| Field | Meaning |
|---|---|
periodDate | What the number is about — the month, quarter or day it measures |
publishedAt | When it first became knowable — when the publisher released it |
A Form 10-D describing July is filed in mid-August. A backtest keyed on periodDate therefore
reads that document six weeks before it existed. Join on publishedAt for anything where
look-ahead bias would matter; from_date and to_date filter on periodDate, because that
is the axis you plot against.
The gap between the two dates is your information edge. It is also the easiest way to fool
yourself, which is why both are on every row rather than one being computed away.
The endpoints
Per-dataset endpoints — start here
Each dataset ORTEX publishes has its own endpoint, named for what it measures. Call one for a
listing and you get every market ORTEX counts that company in; call it with no filters and you
get the whole dataset.
Company-level — GET /api/v11/stock/{exchange_symbol}/{ticker}/altdata/{dataset}/
| Dataset | What it measures |
|---|---|
vehicle_registrations | New cars registered by manufacturer, from 30 national licensing authorities |
clinical_trials | Trial status, phase, enrolment and — the one that moves a stock — the primary completion date |
consumer_credit | Card and auto-loan delinquencies, charge-offs and excess spread, from lenders' securitisation trusts |
bank_fundamentals | Bank balance sheets per institution: FDIC call reports for every insured US bank, and the monthly APRA (Australia) and OSFI (Canada) returns |
share_buybacks | Repurchases as actually executed and filed, not as announced |
medical_devices | FDA 510(k) clearances and recalls by manufacturer |
nuclear_output | Daily operating power of every US commercial reactor |
monthly_revenue | Monthly revenue disclosures from Taiwan-listed issuers |
air_traffic | Airline passengers, capacity, cargo, fuel and punctuality, and airport traffic, from twelve aviation authorities (UK, Spain, France, Türkiye, Eurostat, US, Canada, Mexico, Brazil, India, Australia, Taiwan) |
insurance_premiums | Monthly premium written by every Indian insurer, life and non-life (IRDAI) |
telecom_subscribers | Subscribers per operator and mobile number portability, India (TRAI) and Spain (CNMC) |
prescription_volumes | US Medicaid prescriptions, units and dollars reimbursed per drug manufacturer |
fda_regulatory_events | FDA Complete Response Letters, warning letters and first generic approvals, per company |
oil_gas_production | Oil and gas production per operator: US Gulf of Mexico (BSEE) and every Norwegian field (Sodir) |
mine_production | Copper and molybdenum production per Chilean mine (Cochilco) |
power_generation | Electricity generated per Brazilian plant and per owner (ONS) |
mortgage_originations | VA-guaranteed home loans per lender: purchase, cash-out and streamline refinance |
futures_broker_financials | Net capital and customer segregated funds per US futures commission merchant (CFTC) |
Market-level — GET /api/v11/market/altdata/{dataset}/
These describe a country or a whole market, so there is no ticker in the path. Read them as
the context the company-level datasets are measured against.
| Dataset | What it measures |
|---|---|
labor_market | Payrolls, unemployment, participation, earnings, vacancies and jobless claims across 12 markets |
inflation | CPI, CPIH and HICP — headline and core — across 12 markets |
economic_output | Industrial production, monthly GDP, manufacturing output and retail sales volume |
vehicle_registrations | The denominators for manufacturer volumes: Canadian national new-vehicle sales, and the US registered fleet by state, class and fuel |
air_traffic | Japanese air traffic and capacity for the whole market, domestic and international (MLIT) |
healthcare_enrollment | US Medicare enrolment and Medicare Advantage penetration |
freight_capacity | US trucking capacity: carriers active, inactive and net |
gaming_revenue | Macau gross gaming revenue, monthly, and quarterly by game type (DICJ) |
visitor_arrivals | Hong Kong daily arrivals and Singapore monthly visitor arrivals by source market |
treasury_cash | The US Daily Treasury Statement: TGA balance, withheld payroll taxes and daily flows |
Filtering
Every per-dataset endpoint takes the same three optional filters, and they combine:
| Parameter | Selects |
|---|---|
market | One market — DE, GB, US-CA, EA, EUROPE, … Each endpoint's page lists the ones it covers |
series | One measure within the dataset |
source | One publisher |
Omit all three and you get every series in the dataset, for every market. Each row names its
own market, seriesSlug and sourceSlug, so a single request is enough to chart the lot.
A series slug is unique per publisher, not globally
unemployment_rateis published under that exact slug by the US Bureau of Labor Statistics,
the UK Office for National Statistics and Statistics Canada. Where a slug names more than one
series in a dataset, passing it alone returns400 invalidSerieslisting the alternatives —
pair it withmarketorsource. ORTEX would rather refuse than hand you one country's
print labelled as another's.
The catalog, and the generic endpoints
- Alt Data catalog — every series ORTEX publishes, with
its cadence, unit, theme tags and how far ahead of the reported figure it typically arrives.
Discovery is effectively free, so start here if you are exploring. - Observations and Signals (
/altdata/observations/,/altdata/signals/) are the
generic per-series endpoints, driven by aseriesparameter. They still work and still serve
the same data; the per-dataset endpoints above exist because most people want a dataset, not
a series.
Signals: the comparable form
Raw values are frequently not comparable. Receivables of $10.9bn next to $4.9bn say nothing
without each book's size, and a 5.3% charge-off rate means something different on a subprime
book than on a prime one.
A signal is ORTEX's computation over a series, and it states what the value was compared
against:
| Field | Reading it |
|---|---|
zscore | Distance from the comparison set. Below 1 is noise, 2 is notable, 3 or more is rare |
percentile | Rank within the comparison set, 0–100 |
crossSection | own_history — is this high for this company? all — is this high for this series right now? The two answer different questions |
higherIsBetter | Whether a high reading is the bullish one. High delinquency is not; high excess spread is |
Reading the coverage honestly
Why a company you expect may be missing
Three reasons, and none of them is a bug.
- The dataset is a window, not a census. Several publishers release a rolling window rather
than a full history — the clinical-trials feed is seven days of changed trials, so a large
sponsor can be absent for weeks simply because nothing moved. - The filer is not the listed company. Registers name a marque, a bank charter or a
securitisation trust. ORTEX maps those to the listed parent where the link is defensible and
leaves them unmapped where it is not — a wrong mapping silently attributes another company's
data to your ticker. - The population is genuinely narrow. Most FDIC charters belong to no listed company at
all. No amount of mapping changes that.
Mapping is by exposure, not just by name: a reactor series maps to the utility that runs
it, a securitisation trust to the lender that sponsors it, a marque's registrations to its
parent.
How much history there is
Several of these publishers overwrite their own pages and keep no archive. Where that is so,
the series begins on the day ORTEX started capturing it and lengthens from there — nobody,
including the publisher, can supply the days before that. The US nuclear file is the clearest
case: the regulator publishes a rolling 365 days and the older ones exist nowhere else.
Everything ORTEX fetches is archived on capture, so a series only ever grows.
What ORTEX may republish
Public-sector publications are served as the publisher wrote them. A small number of sources
are derived from a company's or a trade body's own website, and their raw values are not ours
to redistribute — those are excluded from the per-dataset endpoints, and naming one explicitly
returns 403 seriesNotAvailable. Their derived signals are available, because a percentile
computed by ORTEX is ORTEX's to publish where the publisher's own numbers are not. The
catalog's rawAvailable flag tells you which is which.
A source whose raw values are withheld in full is not given a per-dataset endpoint at all,
rather than one that could only ever return an empty result; it stays in the catalog.
Two further sources are in the catalog but served by no endpoint, because their publishers'
terms do not allow commercial use of the data or of anything derived from it: EUROCONTROL's
daily airport flights and the German airports association's (ADV) airport traffic.