Which exchanges can be trusted with trading volume data and why?

What are the trading volume data of cryptocurrency exchanges to be trusted?

No 24-hour trading volume data published by any cryptocurrency exchange has been independently audited. This number is reported by the exchanges themselves and filtered by the websites that display the data-and the filtering rules of platforms such as CoinMarketCap, CoinGecko, Coin Metrics, and Nomics vary significantly, so that when you check different websites on the same exchange, the transaction volumes may be very different.

Why only look at the superficial trading volume?

Trading volume is easy to be artificially created. According to Jon Geenty, a data scientist at Coin Metrics, exchanges "often falsely report transaction volume data in order to climb on ranking websites." CoinMarketCap's own methodology document mentions similar issues, noting that transactions that are free of fees or provide high trading incentives are "susceptible to fraudulent transactions, resulting in artificially inflated reported transaction volumes" and therefore exclude them from the calculation. CoinGecko's Trust Scoring Methodology page expresses the same view, but does not give a fixed formula: It points out that exchanges in the cryptocurrency market may misreport their data, so trust scores are not ranked directly based on superficial numbers, but rather look at deeper information.

How CoinMarketCap filters transaction volume

Based on its supporting documentation (no release date was specified), CoinMarketCap takes the exchange's self-reported 24-hour trading volume and then converts it to U.S. dollars using a reference price. Two of these rules are crucial to determining whether the numbers listed are exaggerated. First, it excludes trading pairs that are free of fees or provide significant trading incentives, on the grounds that subsidized or free of fees are a high incidence area for fraudulent trading. Second, instead of summarizing all trading pairs listed by an exchange, only the top 25 trading pairs are averaged, so that exchanges cannot move up the rankings with a large number of inactive trading pairs. Neither rule tells readers how much of the current trading volume on any particular exchange is excluded-the document describes the methodology rather than the specific results for each exchange.

CoinGecko's trust scoring mechanism

CoinGecko views reported trading volume as an input rather than a primary signal. This is explained in two separate documents. Its trust scoring methodology page describes that scoring will comprehensively consider liquidity (i.e. consistency of transaction volume and depth of order book), as well as cybersecurity history, regulatory status, past events and whether the exchange has issued certificates of reserve. If the data reported by the exchange deviates significantly from the reference benchmark, penalties will be imposed. A separate document (CoinGecko's Volume Methodology page) covers more specific mechanical rules: It excludes trading pairs that are blacklisted due to inconsistent data, or that are not updated for more than three hours, and uses volume-weighted average prices based on a set of trusted exchanges to calculate trading volume. The page also points out that in May 2026, CoinGecko rebuilt the trust score and replaced network traffic comparisons with direct assessments of transaction volume and orderbook depth-which is useful for comparing trust scores obtained before and after the change.

Two earlier frameworks and their findings

Two attempts to rate exchange trading volumes have demonstrated on paper the possible large gap between reported and "real" trading volumes, but both results have been years since.

In 2020, Coin Metrics established a system called the "Trusted Trading Volume Framework". As of July 1, 2020, it had certified only 13 exchanges, including Binance, Bitfinex, Bitstamp, Coinbase, Gemini and Kraken. The framework evaluates each exchange through three tests: whether its price data correlates 80% or more with a set of established exchanges; whether its reported trading volume matches its website traffic; and a qualitative score that includes characteristics such as KYC requirements and regulatory status. Based on this, Coin Metrics believes that as of July 1, 2020, the market's real 24-hour trading volume was approximately US$13.25 billion, while exchanges reported trading volume was close to US$113 billion-the real figure is "approximately one-tenth of the reported figure." One exchange, OKEx, failed all three tests;BiBox, ZB.com, LBank and HitBTC each failed two tests.

A year ago, analytics firm Nomics launched its "Transparent Trading Volume" service, rating exchanges A to D based on the extent to which they publish raw trading data. According to media reports, Coinbase Pro, Kraken and Binance can receive A or A+ ratings by publishing complete trading histories; while a D rating means that exchanges only provide market-level data-that is, daily trading volumes and prices, with no underlying details. The same article also cited a report by Bitwise Asset Management that found that 95% of trading volumes on unregulated exchanges were false. The figure comes from media paraphrasing of the Bitwise report rather than a direct review by the author of this article.

What the regulatory authorities say

On November 16, 2023, the International Organization of Securities Commissions issued 18 policy recommendations for the cryptocurrency and digital asset markets, covering conflicts of interest, market abuse, customer asset custody and information disclosure. The recommendations are intended to provide guidance for regulators in developing rules for crypto asset service providers-they address how exchanges should self-govern and disclose trading arrangements, rather than how data aggregators should calculate 24-hour trading volumes. The report did not propose its own trading volume verification criteria.

Common misconceptions

A common mistake is to treat the "24-hour transaction volume" numbers on aggregated websites as a definite, comparable fact, just like readers would view a company's audit revenue data. But this is not the case. This number is provided by the exchange and is processed through one aggregator's filtering rules, which may differ from the rules applied by another aggregator on the same day for the same exchange. Numbers that seem large on one website may seem ordinary on another simply because the transaction pairs excluded and the weighting method used are different.

What is not covered in this article

This article explains the fundamentals of the main volume filtering methods. It does not tell you whether current trading volume data for any particular exchange is true-this requires applying each framework to the current data, which the evidence gathered here does not do.

Materials from Coin Metrics and Nomics are outdated. Coin Metrics 'results as of July 1, 2020 were reported by the media on July 28, 2020;Nomics' rating and Bitwise's findings came from an article on August 28, 2019. There is nothing in the evidence in this article to indicate whether the two frameworks are still being maintained, or whether the above-mentioned exchanges still maintain the same rating today.

The descriptions of CoinGecko and CoinMarketCap in this article are from partial summaries of their published methodology pages, are not complete documents, and neither does have a release date-therefore, thresholds and marginal situations beyond the above-cited (such as the exact scoring weights behind CoinGecko's trust scores) are not covered.

95% of the figures attributed to Bitwise come from second-hand sources, which are media paraphrases of Bitwise reports. This article does not verify against Bitwise's original report.

None of the four frameworks above are independent financial audits. Each framework is a methodology designed and run by a data provider that has not been verified by external auditors (like a reserve certificate)-so the high rating given by any framework reflects the provider's own standards, not approval by regulators or auditors.

Sources of information

Each of the above facts is attributed to one of the reports mentioned above. If there are differences, they have been clearly pointed out in the article.