3 / 4 · 5 min
Who leads a move and who catches up
In every instrument there is a venue where price changes first and venues that repeat it. Which one leads depends on the coin, the hour of day and who is trading right now.
When a market rises it looks as if all exchanges move at once. On a scale of seconds they do not: price first changes where the demand arrived, and the rest are pulled along by those who profit from the gap. The interval is small, from tenths of a second to seconds, but it exists, and the question 'who was first' has an answer.
What the answer depends on
- The instrument: bitcoin has one leader; a small coin is led by the exchange where its main turnover is.
- The hour: the Asian session and the American one are traded by different people on different venues.
- The kind of event: a listing announcement moves the listing exchange first; macro news moves large derivatives venues first.
- Market state: in quiet hours the venue with the largest turnover leads; in a run, the one with the thinnest book does.
Mistaking your own latency for an exchange's leadership
The main trap of the topic, and it is technical. Data from different venues reaches you with different delay: one feed is fast, another slow. If you compare by the time data arrived, you will inevitably conclude that the exchange with the best connection 'moves first'. That is a statement about your channel, not about the market. The only honest way is to compare by the timestamp the exchange itself applied, and that is exactly why we keep both stamps.
Why this matters in practice
Two uses, both applied. First: if you trade on a following venue, the leader's move is an accomplished fact you learn about late; building an entry on it is already too late. Second: a divergence from the leader is a signal to check whether a feed has broken. If every venue moved and one stands still, the question is not about the market but about that venue's data.
What this looks like in our data
We do not compute leadership as an automatic number and say so honestly: such a measure needs careful statistics per coin, and a single figure saying 'the leader is Binance' would be a lie for half the instruments. But the data for observing it is there: prices from every venue where a coin trades arrive at the same time and sit side by side. Large moves on a liquid coin are visible on all venues at once; on a small coin the picture differs — there is usually one exchange with real turnover and several that simply repeat its price with a lag.
Agreement between venues beats a single spike
The practical reason to keep several sources: simultaneous aggression on several exchanges weighs more than the same volume on one. A spike on a single venue is most often somebody's large order hitting a thin book. A spike on four at once is about the coin, not the venue. Densities read the same way: a wall standing on several exchanges at the same level means more than a lone one.
Prices and turnover across every venue where an instrument trades — on its page; the state and freshness of each source — in the about section.
State of the sourcesCatch the order
Pick a coin trading on at least four venues and wait for a noticeable move. Write down which venue's price left first and by how many seconds it preceded the rest. Repeat five times at different hours. You will either see a stable leader or become convinced it changes — and both answers are useful.
On one exchange a coin rose 4% in a minute, on the other five by 0.2%. What do you check first?
That venue's turnover and book: most likely somebody ate a thin book with a large order and price will return to the rest within a minute. The second thing to check is whether the other five feeds are alive: if they have not updated, the move may be real while the data is frozen. The difference between 'one exchange flew' and 'five exchanges lagged' is settled by freshness, not by price.