How much of the reported volume is real, and whose number is "the" price?

This is the lesson a market-data company is obliged to write. Crypto data has three specific defects that do not appear in exchange-listed equity data, and every one of them will silently corrupt an analysis that assumes otherwise.

1. Reference prices are recipes, not observations

Because there is no consolidated tape, a single "price of bitcoin" has to be constructed, and every construction is a published methodology with choices in it:

• Which venues are eligible — and what disqualifies one
• The observation window — an instantaneous snapshot, or a volume-weighted average over the last hour
• Weighting — by volume, equally, or by a median across venues
• Outlier handling — trimming, partition medians, and rules for a venue that goes offline mid-window

Regulated benchmark rates used for futures settlement, index-provider rates used by funds, and the internal index prices that derivatives venues use for marking and liquidation are all different recipes. They give different numbers for the same instant. None is wrong; they answer slightly different questions. The mistake is assuming there is a single canonical number underneath them.

This has teeth. The index price a venue uses for liquidations determines whose position closes. Two venues, two indices, one market move, different survivors.

Some of these recipes are not only quoted; they are traded at. A benchmark with a strike time, such as CME CF's Bitcoin Reference Rate at 16:00 London and its New York variant at 16:00 New York, is the price at which CME futures settle and at which most US spot bitcoin ETFs strike their daily NAV, per each fund's prospectus. Creations and redemptions in those funds, and the market makers hedging them, execute against that window, so a measurable share of institutional flow clusters into one hour of a market that trades all 24. For anyone hedging or benchmarking against these products, the daily close is that fix, not UTC midnight.

2. Reported volume includes volume that never happened

Exchange-reported volume is self-reported, and the incentives to inflate it are direct: ranking on aggregator sites, attracting listings and listing fees, and appearing liquid enough to be worth routing to. Wash trading — trades between accounts under common control, producing volume without transferring risk — is the mechanism.

A widely cited 2019 analysis submitted to the US SEC examined roughly 81 exchanges reporting about $6bn a day of bitcoin spot volume and concluded that only a small fraction of it — on the order of a few per cent — was economically real. Methodologies and venues have improved since, and the headline number should not be treated as current. The mechanism is what persists, and it is worth knowing the tells:

• Volume that is enormous relative to visible order-book depth
• Volume with no price impact — real flow moves prices, wash flow does not
• Trade-size distributions that are suspiciously uniform, or lacking the round-number clustering that human and algorithmic traders produce
• Trade timing that is too regular
• Spreads that stay wide despite "heavy" turnover

For your own work the practical rule is severe and simple: treat venue-reported volume as an unverified claim. Rank venues by depth and by price impact, which are much harder to fake than a volume counter.

3. Dead tokens leave the dataset

Tens of thousands of tokens have been issued. A very large share of them no longer trade at all — abandoned, exploited, delisted or simply ignored. Data aggregators quietly remove them.

So any statistic computed over "the tokens currently listed" — an average return, a median drawdown, a success rate — has already excluded every complete failure. The Foundations course gave you the question that catches this: who is missing from this sample? Crypto is where it bites hardest, because the mortality rate is high and the removal is silent. A historical study of altcoin performance built on a current listings file is not a study of altcoin performance; it is a study of survivors.

The three-question checklist

Before trusting any crypto figure: which venues, which methodology, and which assets have been dropped? None of this is a reason to dismiss crypto data — it is what using it competently requires.

In the data

Price and turnover for bitcoin over the same year, one under the other:

Distrust the volume bars on principle. They are a dollar figure aggregated from venue-reported numbers, while "volume" on a futures chart counts contracts, so the label tells you nothing about the unit. And even the number of crypto pairs a provider covers is a claim rather than a measurement: two descriptions of the same dataset, taken the same day, have given different counts.

Try it now

1. Pick out the five tallest volume bars on the chart above, then look at what the price did on those same five dates.
2. Real flow tends to move price. Measure each of those five days on the price chart and write the moves beside the volumes. A series where the largest volumes sit on unremarkable days deserves a second look. Ask what is being counted before you conclude anything about who is doing it.
3. Then read one published index methodology end to end, for example CF Benchmarks' methodology for the CME CF Bitcoin Reference Rate, which is public on its own site — venue eligibility, window, weighting, outlier rules — and write down which choice you think matters most. That is the number behind every headline price you have ever quoted.