Order Flow

Order-flow analysis examines trades tick by tick, separating aggressive buys (lifting offers) from aggressive sells (hitting bids) and accumulating the imbalance into net pressure — commonly cumulative volume delta (CVD). It is the base dataset of market-microstructure analysis.

Order flow is most valuable for spotting divergence: price making a new high while cumulative aggressive buying fails to confirm means the rally is driven by sellers stepping away rather than buyers stepping in, and such advances typically lack durability. Conversely, price falling while CVD flattens hints at exhausted selling. The pitfall is inconsistent methodology across venues — platforms differ in how they label the taker side, so aggregating CVD across exchanges without normalising definitions can produce divergences that are pure data-stitching artefacts.

Volume and market-structure metrics describe the size and distribution of trading activity. Methodologies differ sharply between venues (wash-trading filters, USD conversion), so confirm the accounting basis before comparing platforms side by side.

Methodology decides whether a CVD chart is comparable across sources. Exchange-labelled taker direction is the gold standard; where it is absent, the tick rule, which classifies each trade by whether it printed above or below the previous price, approximates it with known error in fast markets. Perpetual and spot CVD deserve separate panels, because spot delta reflects fully funded flow while perp delta is dominated by leverage, and their divergence is itself a signal: perp-led rallies with flat spot delta are the fragile kind. Cumulative series also need explicit reset conventions, whether session, daily or rolling windows, since an unanchored CVD's absolute level means nothing and only its slope and divergences do. One subtlety connects this page to the liquidations feed: forced liquidation orders execute as aggressive flow, so a burst of apparent aggressive selling during a crash is often the exchange's engine closing longs rather than traders choosing to sell, and filtering those prints materially changes the read.

The live chart for this metric is still being wired up. Global volume, venue share and perp-vs-spot pages already carry real data.