> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ostrum.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Where the data comes from

> Everything in the dataset was contributed by someone with first-hand knowledge

Ostrum's data comes from one place: contributors. Every entry in the dataset
was submitted by someone with first-hand knowledge of it. No scraping, no
bought-in data, no synthetic filler.

Here's the thing about markets like these: no individual ever holds enough
observations to see one clearly. Pooled, the contributors collectively hold
a better dataset than any incumbent.

## What a contribution looks like

A contribution is one thing you know first-hand. In the launch markets, for
example:

* The terms of a market-making agreement: loan size, option strikes, tenor,
  fees
* An exchange listing quote: venue, fee structure, deposit requirements,
  timing
* A funding round: valuation, structure, investor terms

Intake is a short guided conversation. You describe what you know, Ostrum
extracts the structured fields, asks about anything missing, and shows you
exactly what will be recorded before you submit anything.

<Info>
  Exact values never enter the dataset. Figures are converted to buckets and
  ratios at submission and the raw numbers are destroyed. Precise enough to
  be useful, coarse enough that no single contribution is identifiable.
</Info>

## Keeping it real

A market for information is only worth anything if the information is real.
Two mechanisms keep it that way.

**The verification system.** Every submission runs through it. The system
tests validity against public sources, internal data, and signs of
fraudulent or manipulated documents, and supporting proof is purged once it
reaches a verdict. Some of what the system checks is published here; some
is deliberately withheld to protect it against exploitation and data
poisoning.

**Corroboration.** Independent contributors converging on the same facts is
the strongest signal the market has. Corroborating submissions confirm
earlier ones and earn in their own right.

Contributions that fail verification are excised and earn nothing. Fabricating
simply doesn't pay.

## Why contribute?

Because first-hand knowledge depreciates either way. Held privately, it
decays toward worthless as the market moves. Contributed, it pays you
upfront and keeps paying while it stays relevant. The mechanics are in
[How you earn](/tokenomics/earning).