Parallax

Bitcoin research you can check.

Two views, one true measurement.

Astronomers measure the distance to a star by looking at it from two different places and measuring how far it appears to shift. One viewpoint gives you a guess. Two give you a number.

That is how everything here is made. We run our own full archival Bitcoin node, parse the entire chain ourselves from the raw block files, and then measure every finding twice — a second time through code written independently of the first — before it is allowed out the door.


In plain words

What this is

Bitcoin's whole history is public. Anyone can download it. Almost nobody does, because it takes a machine with a spare terabyte, several weeks of grinding, and a great deal of patience to turn 800 gigabytes of raw blocks into something you can actually ask questions of.

We did that. Now we publish what we find, for free — and sell the finished datasets underneath the findings to anyone who would rather not spend a month rebuilding them.

The unusual part is the last step. Every file we publish carries a fingerprint that we commit to the Bitcoin blockchain itself. You can prove that the file you downloaded is byte-for-byte the file we sealed, on the day we sealed it, and that neither we nor anyone else has quietly changed it since. Here is how to check that yourself — you should, and it takes about a minute.

Featured finding — free, in full

The coins that never woke up

In Bitcoin's first years, blocks were mined straight to a raw public key. We took every one of those early coinbase payments and asked a simple question of each: is this exact coin still sitting untouched in the current unspent set?

1,704,022.66597442 BTC

Held in 34,185 early coinbase outputs that have never once been spent — out of 196,479 such outputs in total, spanning block 0 to block 955,609.

Of everything paid out this way in 2009, two thirds of it — 66.13% by value — has still never moved, sixteen years later.

Read the full finding → including the year-by-year table, the method, and the trap that makes most attempts at this measurement quietly wrong.


The standard

Why you shouldn't trust us

Every chain-data vendor sells you numbers you have to take on faith. We would rather you didn't. Four rules govern everything published here:

Measured twice

No finding ships on a single viewpoint. Each is reproduced through an independent code path before it is published, and paired with a control — a test that must fail if our instrument is working correctly.

Our own data

Nothing here comes from someone else's API. Every number is drawn from raw block files parsed on our own archival node. No rented data, no third party to go dark on us.

Re-derivable or it doesn't ship

Every dataset carries the exact query that produced it. If an outsider can't reconstruct a number from what we hand them, we don't publish the number.

Retractions get equal billing

We have published findings and later killed our own. When that happens the correction runs as prominently as the claim did. A record that only ever grows is a marketing document, not a research record.

What we don't do

No price, ever

You will not find a price prediction here, or a trading signal, or a chart with our guess about next quarter drawn on it. We don't sell "on-chain alpha", and we don't label addresses as belonging to exchanges — the published accuracy of the clustering techniques that would let us do that is far too poor to present as fact.

We measure what the chain actually says. What it's worth is somebody else's business.

Datasets

The files behind the findings

Raw Bitcoin data is free — Amazon publishes the whole chain, and so do others. So we don't sell raw data, because it isn't worth anything. What takes the month is the work on top of it: knowing which blocks belong to which era, which measurements are traps, and which coins are genuinely untouched rather than merely appearing so.

See the four datasets →

Current status

The shop is not open yet. The findings are free and complete as published; the datasets are listed with their full schemas and honest limits so you can judge them, but there is no way to buy one today. Nothing here collects your details, and there is no newsletter to join.