Look up the price of Bitcoin on three different sites at the same moment and you will get three slightly different numbers. This regularly confuses people, and occasionally alarms them. The explanation is straightforward once you know how a crypto price is actually produced, and it is worth knowing because the same mechanics explain a much more serious phenomenon: prices that are wrong in ways that cost people money.
There is no official price
Stock exchanges are centralised. A share of a listed company trades on a primary venue with a closing auction and an authoritative print. There is a single price of record, and everyone quotes it.
Cryptocurrency has no equivalent. Bitcoin trades continuously on hundreds of independent venues in dozens of countries, plus a large and growing volume on decentralised exchanges that are not companies at all. Each venue runs its own order book, and each order book has its own price at any given instant. There is no closing bell, no consolidated tape, and no authority whose number is the number.
So the price you see is not the price. It is a calculation, performed by whoever is publishing it, over the venues they can see.
How aggregated prices are built
Data providers — the companies that supply prices to sites like this one — construct a reference price roughly as follows.
They connect to as many exchanges as they can, usually by API and often over a live socket connection, and receive each venue's last trade price and order book. For a widely traded asset that might be well over a hundred sources.
They then filter. Venues with implausible volume, unreliable data, or trading patterns consistent with wash trading are excluded or discounted, because including them would let a small dishonest exchange drag a global price around. Providers maintain their own criteria for this, and those criteria are one of the things that distinguishes them.
What survives is combined into a weighted average, with weight determined mainly by real trading volume. A venue handling a billion dollars a day carries far more influence than one handling fifty thousand, which is why a single outlier exchange quoting a wildly different price barely moves the published figure.
Finally, prices quoted in other currencies are converted. Much of the world's crypto volume is denominated in stablecoins rather than dollars, and a portion trades against euros, yen, won and other national currencies. Producing a single dollar price requires converting all of that, using stablecoin and foreign exchange rates that themselves move.
Why sites disagree
Once you see the pipeline, the discrepancies stop being mysterious.
Different venue sets. Two providers tracking different exchanges are averaging different markets. For Bitcoin the effect is tiny. For a mid-sized asset listed on twelve exchanges, where one provider covers ten and another covers eight, it can be visible.
Different weighting and filtering. One provider might exclude a venue another includes, or weight by trade count rather than notional volume.
Different update intervals. Prices are recalculated on a cycle — sometimes seconds, sometimes a minute or more — and caching adds further delay at each layer between the source and your screen. In a fast-moving market, two sites a few seconds out of step will show different prices for entirely mundane reasons.
Different currency conversion. A site quoting in euros applies its own FX rate, and rates differ between sources.
For a major asset these differences typically amount to a fraction of a percent. If you are seeing more than that between two reputable sources, the asset is probably thinly traded, and that is itself the more important observation.
Where prices are genuinely unreliable
The differences above are noise. Three situations produce prices that are actively misleading.
Very thin markets
A price is a record of the last trade. If the last trade in an asset happened four hours ago, the price on your screen is four hours old regardless of how live the site looks. Small tokens can sit for long periods with almost no activity, and a single modest order will then move the quoted price by double digits. The number is technically accurate and practically meaningless: you could not buy or sell any real size near it.
Volume is the check. A price attached to negligible volume should be read as an estimate of where the asset might trade, not as a price you can transact at.
Manufactured volume
Because volume drives weighting, rankings and perceived legitimacy, there is a persistent incentive to fake it. Wash trading — an exchange or a market participant trading with itself — has been documented extensively across the industry, and it inflates both volume and, indirectly, the credibility of a venue's prices.
Reputable data providers filter for it, but filtering is imperfect and adversarial. When an asset shows most of its volume on venues you have never encountered, treat the price with corresponding caution.
Depegged and fragmented markets
Occasionally an asset trades at genuinely different prices in different places for real reasons — capital controls, a withdrawal freeze at a major exchange, or a stablecoin losing its peg on one venue before others. In those moments an average across venues describes a market that does not exist as a single market, and the number nobody can trade at is the average itself.
What "live" actually means
Sites describe prices as live or real-time. In practice there is a chain: the exchange records a trade, the data provider ingests and recalculates, a caching layer stores the result, and a site renders it. Each stage adds latency, usually small.
For reading the market, this is irrelevant. A price a few seconds stale tells you what you need to know. For trading, it is not irrelevant at all: the price you execute at is set by the order book of the venue you are trading on at the instant your order arrives, and that will differ from any aggregated figure. If you are placing an order, the exchange's own book is the only price that governs what you pay.
Practical takeaways
Compare like with like. Two sites showing marginally different prices for a large asset is normal and not a sign that either is broken. Check volume before trusting a price, because a price without trading behind it is an estimate. Expect wider divergence in small assets, and treat that divergence as information about liquidity rather than about data quality. And when you are actually transacting, ignore the aggregate and read the exchange's order book, because that is what will fill your order.
Where Coinvilo's data comes from
Our market data — prices, market capitalisation, volume, supply figures, historical charts and exchange listings — is supplied by CoinGecko, which aggregates and weights across a large set of venues and applies its own filtering for unreliable volume. We publish what that source reports without adjusting or delaying it. Where a figure looks wrong, we would rather hear about it: you can reach us on the contact page, and we will check it and pass it upstream where necessary.
This article is educational and is not financial, investment or tax advice. Market data can contain errors or be delayed. Verify anything important against the exchange you intend to use before acting on it.