Why the same bitcoin trades at different prices

When the same asset trades at two prices, there is a profit available to anyone who buys at the lower one and sells at the higher. That activity is arbitrage, and it is the force that keeps venue prices close. An arbitrageur buying on the cheap venue pushes its price up; selling on the expensive venue pushes that price down. The two quotes converge, and the gap narrows until the profit no longer covers the cost of executing the trade.

The gap is never eliminated, because the cost of closing it is never zero. Moving bitcoin between venues takes time, and during that time the price can move against the position. Capital has to be pre-positioned on both venues, which ties up funds. Trading fees are paid twice. Withdrawal limits and verification requirements slow the transfer. And the size that can be arbitraged is limited by the depth available on the thinner venue, which is usually the one with the anomalous price. Each of these costs sets a floor beneath which the gap is not worth closing, and that floor is why a spread of a few basis points persists even in a well-functioning market.

The mechanism also fails in specific circumstances. During a sharp move, the venues can diverge widely for minutes at a time, because arbitrageurs are unwilling to take on the risk of a position that may move before they can close it. During a venue outage, the price on the affected exchange can detach entirely from the rest of the market. And in the early years, when transfers were slow and venues were few, the gaps were far wider than they are today. The narrowing of the spread over time is one of the clearest measures of the market's maturation.

Every price series on this site is drawn from a stated set of venues, and the provenance line beneath each table names them. That matters because a series built from one venue and a series built from a volume-weighted average of many will differ, sometimes materially, in the early years. The differences are largest in the 2010 to 2014 period, when venues were few, spreads were wide, and the concept of a market price was considerably looser than it is now.

A reader comparing figures across sources should expect small discrepancies and should not treat them as errors. Two reputable series can disagree on the closing price for a given day because they sampled different venues at different moments. What matters is that each series states its basis, which is why the provenance line appears beneath every table on this site rather than only on the data sources page.

The practical implication for a reader is that any single quoted price is an approximation of a market that has no single price. The approximation is good enough for the purposes this site serves — describing the shape of a multi-year record — and it is not good enough to settle a dispute about the exact level on a particular afternoon. The liquidity page explains why the gaps widen when depth thins, and the trading volume page covers the reliability of the venue figures themselves.