Ask what bitcoin costs and you get a number. Ask where that number came from and the answer is usually an aggregator, which took it from a set of exchanges, which serve different customers in different jurisdictions under different rules. The single number is a convenience. Underneath it sits a market that is fragmented by geography far more than the charts suggest.

For anyone doing serious analysis rather than reading a ticker, that fragmentation is not noise. It is the signal.
Contents
Three ways geography shows up in the numbers
The first is persistent price divergence between regional venues. The best documented example is the premium that has appeared repeatedly on Korean exchanges, where capital controls and the practical difficulty of moving fiat in and out prevent arbitrage from closing the gap. It is not a data error. It is a real price, in a real market, that a global average quietly averages away.
The second is availability. Exchanges restrict which assets and which products they offer by the customer’s jurisdiction, so the listing page itself differs by country. Derivatives, leverage limits, staking products and entire tokens appear and disappear depending on where the request comes from. An analyst checking whether an asset is broadly accessible from one location is answering a narrower question than they think.
The third is liquidity distribution. Depth for the same pair varies enormously between venues, and a headline price backed by thin books is not comparable to one backed by deep ones. Any figure quoted without the venue and the depth behind it is close to decorative.
Building a data set that survives review
- Record the venue and the timestamp with every quote, since a price without both is not reproducible
- Prefer the exchange’s own public API over an aggregator when the question is about a specific market
- Capture order book depth alongside the last traded price, because the two together are the actual answer
- Note the jurisdiction restrictions in force at collection time, as listing availability changes with regulation and is rarely announced
- Sample regularly rather than opportunistically, because divergences are only meaningful as a series
- Keep the raw responses, since a re-parse is cheap and a re-collection of a moment that has passed is impossible
The practical obstacle to all of this is rate limiting. Public market endpoints cap requests per address, and a study that polls several dozen venues on a short interval will hit those caps within minutes. What follows is worse than an outright failure: the collection continues with gaps, the gaps look like market behaviour, and a chart gets built on them.
The usual remedy is to distribute the requests, and since this is public endpoint polling at volume rather than anything that needs to resemble a consumer connection, datacenter proxy servers are the cheap and fast option. Pair that with a completeness check on every run, comparing the number of records you received against the number you expected, so a partial collection is caught rather than published.
The line worth not crossing
Collecting public market data is one thing. Opening or operating an account on a venue that does not serve your jurisdiction is another, and it breaches the exchange’s terms in every case, usually in the same paragraph that covers identity verification.
Enforcement is not theoretical: accounts get frozen with balances inside them, and withdrawal then requires exactly the verification the arrangement was avoiding.
The honest framing is that regional differences are something to measure, not something to trade around. Knowing that an asset is unavailable in a major market is itself a finding, and frequently a more interesting one than the price.
What this changes about how you read research
When a report quotes a price, look for the venue. When it claims an asset is available, ask where. When it shows a divergence, check whether the two series were collected at the same moment or merely on the same day, because in this market a few minutes is a long time.
None of this requires expensive tooling. It requires writing down where the number came from, which remains the least glamorous and most reliable improvement available to crypto analysis.
FAQ
Why do exchanges in different countries show different prices?
Local supply and demand, plus friction on moving fiat across borders, prevent arbitrage from closing the gap. Capital controls make the effect persistent rather than momentary.
Why does an exchange list different assets for different users?
Regulatory requirements differ by jurisdiction, so venues restrict assets and products based on where the customer is. The listing page is assembled accordingly.
Is aggregated price data good enough?
For a general view, yes. For any question about a specific market, an aggregate averages away the thing you are trying to measure.
What is the most common flaw in collected market data?
Silent gaps caused by rate limiting. The series continues, the missing records look like market behaviour, and nobody notices unless completeness is checked on every run.
