How to read a price, and why two sites disagree
A headline price is a constructed average of many venues, and the construction choices explain nearly every discrepancy a reader notices.
A price is a record of one trade, on one venue, at one moment. There is no central exchange for digital assets and no consolidated tape, so every headline figure is an aggregate built by someone who chose which venues to include and how to weight them. Two reputable sites showing different numbers are usually both correct about slightly different things.
What a single venue's price actually reports
On a centralized exchange, an order book holds resting bids and offers. The last traded price is history: it says what someone paid, not what anyone can pay now. The live picture is the best bid, the best offer, the spread between them, and market depth showing how much size sits at each level on either side.
On a decentralized exchange there is often no book at all. An automated market maker quotes from a formula against pooled reserves, so the price is a deterministic function of the pool's balances and every trade moves it along a curve. What a taker receives depends on trade size relative to the pool, which is price impact, and this is distinct from slippage, the difference between the quote seen and the fill received because other transactions landed first.
Both venue types therefore report a price valid only for a small trade. A quoted figure is not a price at which any quantity can transact, and for thinly traded assets the gap between the two is the entire story. This is also why the same asset can show a different price on two venues at the same instant without either being wrong: they are separate pools of liquidity connected only by whoever is willing to move capital between them.
How an aggregate is built
An aggregator collects trades or quotes from many venues and combines them, typically volume-weighted over a short window, so that deep venues dominate and a single small print cannot move the headline. Serious methodologies then add several filters: excluding venues that fail a liquidity or integrity screen, trimming outliers, discarding stale quotes, and treating pairs quoted in a stablecoin differently from pairs quoted in fiat currency. Arbitrage keeps venues close together most of the time, which is why differences are usually small, and why they widen exactly when conditions are stressed and arbitrage capital is constrained or slow to move.
Index construction is a further variant used where a price must be defensible rather than merely indicative. Reference rates behind regulated products are typically computed over a fixed daily window across named venues, with published rules for outliers and for what happens when a venue fails to report. Such a rate can differ from any spot screen at the same moment and is still the correct figure for its purpose.
Six reasons two sites show different numbers
| Cause | What it does to the number | How to check |
|---|---|---|
| Different venue set | Including or excluding a large regional exchange shifts the weighted average | Read each site's venue inclusion rules |
| Different weighting | Volume-weighted, depth-weighted and median constructions diverge in thin markets | Compare against a single venue's own book |
| Snapshot timing | Figures taken seconds apart differ during fast moves; some tables refresh on a schedule | Look for the as-of timestamp on the page |
| Quote currency | A pair quoted in a stablecoin inherits any deviation from its peg, and fiat conversion uses a separate rate | Check the peg deviation on the quote asset |
| Fabricated volume | Inflated volume raises a venue's weight unless it is screened out | Compare reported volume against depth and turnover |
| Ticker collisions | Two different tokens share a symbol and get mapped to the same row | Verify the contract address rather than the ticker |
The fifth row deserves its own note. Wash trading, in which the same party is on both sides of a trade, has been documented on a range of venues and inflates the field that most naive aggregation methods rely on. A venue reporting large volume with a thin order book is describing two things that cannot both be true, and that inconsistency is the practical detection method.
Derivatives add a further layer often mistaken for a spot disagreement. A perpetual future trades at its own price, tethered to spot by periodic funding payments rather than by an expiry, and exchanges compute a separate mark price from an index specifically to avoid liquidating positions on a single venue's wick. A screen showing a perpetual is not showing the spot market, and during dislocations the two can differ noticeably.
What the number does not contain
A price carries no information about how much could transact near it. Two assets at identical prices with identical market capitalizations can have books differing by orders of magnitude, and the shallower one will move much further on the same size. Reported volume is a partial proxy but is the field most vulnerable to fabrication; turnover, which relates volume to market capitalization, is more comparable across assets, and a value implausibly high relative to peers is a reason to inspect the venue breakdown rather than to conclude anything about interest.
A price also carries no information about who set it. Where a small number of holders control most of the supply, the market price reflects marginal trades among a thin float, and applying it to the entire supply produces a figure no process could realize.
Reading percentage changes carefully
Change columns such as 24-hour change depend on the reference point chosen, whether a rolling window or a fixed daily close in some time zone, and the two conventions can disagree materially on a volatile day. Comparing a rolling figure on one site against a session-based figure on another compares different quantities. Volatility measures depend likewise on sampling interval and annualization convention, both of which differ between providers and are rarely displayed next to the number.
The construction rules used here are written out on the methodology page, and the venues and vendors behind each field are listed under data sources. The next lesson walks that pipeline in detail, source layer by source layer.