Order flow trading is the practice of reading executed trades and resting orders to infer what large participants are doing, and positioning accordingly. Instead of asking where price has been, it asks who is buying, who is selling, and how aggressively.
What order flow shows
The aggressor side of each trade — whether it hit the bid or lifted the offer. Imbalance between bids and asks at nearby levels. Where large resting orders sit and whether they get pulled as price approaches. Cumulative volume delta, the running difference between aggressive buying and selling.
Read together, these show whether a move is being driven by conviction or by thin liquidity.
What it doesn’t show
Intent. A large sell that hits the tape might be a whale exiting, a market maker hedging, or a liquidation. The flow is visible; the reason isn’t, and most order-flow narratives are stories told after the fact.
Where it works better
Fully on-chain order books. On a centralised exchange you see what the operator shows you. On Hyperliquid every order, cancellation and fill is public, so the flow is complete and verifiable rather than sampled through an API.
The systematic version
Order flow is a family of features, not a strategy. Book imbalance, trade aggression and liquidity depth are inputs a model can use alongside price and volatility. The difference between a tape reader and a systematic approach is that the model quantifies what the flow predicts instead of narrating it.