How to read a crypto market session

Start with the session clock and BTC spine

Choose Sydney, Asia, London or USA by selecting its name or timeline bar. The clock opens that market's latest available book for the displayed trading day. USA combines the New York cash window with the aligned Chicago/CME institutional window, while preserving both sources on its market anchors. Every book follows the exchange-local opening and daylight-saving changes, while visible countdowns and Session Rail labels use your device's local time. Before a selected market opens, the latest completed book stays visible with a clear countdown to the new session. The chart records BTC in fixed five-minute buckets and shows one hour before and after a finalized session as context. The open, close, final metrics and event markers stay fixed to the stored session window. ETH and SOL remain in the evidence without crowding the main reading surface.

• Price spine BTC open, current value, return and range through the selected session.
• BTC session line A compact real-price path behind the session rows. The clock ends at the last scheduled close, so empty time after all sessions does not compress the useful market window.
• Market anchor A named open, overlap, close or settlement recorded in its local timezone.
• Coverage The number of stored samples and the sources actually available for the book.

One chart, one strip of events

Live opens with a 24-hour session clock, then one market card: the candles, the hour ruler under them, and the session's events as a strip below. The strip runs oldest to newest, the same direction as the axis above it. The three newest events stay open and everything older is a narrow spine you can click to open in place; the arrows at each end appear only while there is something that way. A row of event markers used to sit between the chart and the strip; it was removed in September 2026 because it said in 8px squares what the strip under it says in words. On a phone the events are a vertical list under the chart, split by hour, with the two newest open. Desktop traders can still choose 1m, 5m, 15m or 1h, and the chart still pans and zooms. Below the card, Market direction, Exchange flow, Recent pattern, Asset impact and Liquidation risk read the same prepared evidence as before. Distance from usual describes how far a number is from this session's normal level, not the probability of a price move.

• Liquidation depth The ladder is a depth profile, not a list of the biggest levels. It walks equal 0.5% bands out to 2% each side of the anchor price and gives every band one row: the heaviest real price level inside it, how much was forced to close in that band, and a HEAVY mark when a band holds a quarter of everything in view. Anything further out than 2% collapses into one overflow row per side. A band with nothing in it keeps its row and shows a dash. These are realized liquidations recorded during the session, not a forecast of where liquidations will happen, so they can only exist at a price the session actually traded: a session with a 1.2% range can never fill a 2% ladder. The session's own high and low are drawn into the ladder as dashed lines, and the bands beyond them are marked as price the session never reached, which is different from no forced closures happening there.
• Last 10 sessions The dark Recent pattern card compares this session with the same session on the ten trading days before it. Usual is the median of those ten, and the chart below plots them: one column per session off a centre zero line, oldest on the left, this session as the last column, and the median drawn across all of them as a dashed line. Where this session lands against that line is the comparison, so a claim like calmer than usual can be checked against the sessions that produced usual.

Tell a stored session from a live one

Crypto trades around the clock but these sessions follow the opening hours of the exchanges they are named after, so at weekends and overnight no session is open. The page then shows the last completed one, and it says so: the header carries a status chip, the session name, its trading date, a clear note that visible times use your device timezone, a STORED or live freshness marker beside the price, and a notice naming the day on screen and when the next session starts. Read the price in that header as the close of the session named above it, not as the current market. For the same reason the liquidation ladder labels its anchor SESSION CLOSE while a stored book is open, and BTC NOW only while the price is streaming.

• Post-session hour A session does not freeze at the bell. For an hour after the close the book stays open as POST-SESSION: the result is already fixed, and what the chart adds is the hour that followed it, the mirror of the pre-session context before the open. The book is stored to the archive when that hour ends.
• Archive The Archive tab opens the record for one session at a time: how many of its sessions closed positive, the median it sets, the distribution behind that median, the day chain it sits in and every session as a row you can open. Off-hours this is the part of the page with data.
• Round trip A deposit and a withdrawal of the same asset at the same venue, minutes apart and the same size, are folded into one entry marked NETS TO ZERO. Its inspector keeps the same aggregate view and never reclassifies one raw leg as directional.
• Locked event Some timeline events are on paid plans. The headline still shows the amount, asset, venue and direction; a single plan block explains what the required tier adds. Creating a free account is not presented as access to paid exchange evidence.

Read changes as chapters, not as dashboard tiles

The chronology adds an episode only when a configured threshold or a real market event is reached. Related updates of the same kind, asset, venue and direction inside a short window are grouped so repeated transfers do not become separate conclusions. A grouped transfer episode states gross movement, net exchange flow and the supported BTC context; its inspector counts direct evidence links. Inflow and Outflow always describe where funds moved, while Bullish, Bearish and Risk describe the separate BTC interpretation. Open an episode to inspect the same grouped facts shown on the timeline. Missing source coverage appears as unavailable instead of a zero, and paid futures facts use the same entitlement as the Perps workspace.

Use the archive as market memory

The archive answers for one session cohort at a time. The prior states how many of that session's finalized books closed positive, with its median, best, worst and how often the next session in the chain extended the move. Outcome distribution buckets every one of those returns and marks the median and the last close inside the shape, so you can see whether the edge is a small directional tilt or a fat tail. Handoff odds read the real chronological chain: after an up Sydney, what the next session did, over the handoffs on record. The session chain puts one trading day per row with every session across it, names the session that owned that day's high to low range, and marks CEX flow in that session's own sigma. Who owns the tape ranks the sessions by their share of the daily range over the last 90 days. Every row of the table opens that stored book. Each number is aggregated on the session loop's schedule from finalized books only, so the archive never recalculates market history when you open it, and a session with too thin a sample says so instead of showing a shaped number.

• The prior Wins against total for the cohort, its median return, its best and worst finalized session and how often the next session in the chain closed up after it.
• Session chain One row per trading day, one column per session, coloured by return. The violet tick under a return is that session's CEX net flow measured in its own sigma, drawn only from about one sigma out.
• Range owner The session with the widest high to low share of that day, and its share of the whole day's session range.
• Verdict The stored classification of that session: range expansion, trend up, trend down or quiet. It is the same threshold applied at finalization, not a fresh judgement.

Treat the final verdict as a grounded read

The final read combines BTC return and range with stored CEX net, DEX net, liquidation totals, qualified-event count and key public chapters. The AI version can use only those normalized facts and public chapter references. A validator rejects unknown numbers, unsupported causal language, contradictory flow direction and missing evidence links. If generation fails, the same multi-source deterministic read remains visible. It is descriptive evidence, not a trading signal or proof that a market open caused a price move.