Order Flow

Order flow is the stream of buy and sell orders arriving in a market over time. In its narrower technical sense it is signed: each trade is classified by whether the buyer or the seller initiated it, and the balance between the two is taken as a measure of net buying or selling pressure. Order flow is the mechanism through which the views and needs of traders are translated into prices.

The concept is central to market microstructure, the study of how trading takes place. Market makers watch order flow to judge whether they are trading with informed counterparties, short-term traders use it to anticipate price moves, and brokers treat their clients' order flow as a commercial asset. The same term is used in all three contexts.

A trade has a buyer and a seller, so the number of shares bought always equals the number sold. What distinguishes the sides is who was willing to pay to trade immediately. A trade is buyer-initiated when a buyer's market order or aggressive limit order lifts the offer, and seller-initiated when a seller hits the bid. Where the data do not record the aggressor, it is inferred: a trade above the mid price is treated as a buy and one below as a sell.

A common summary is the order flow imbalance over an interval:

• $V_B$ is the volume of buyer-initiated trades in the interval.
• $V_S$ is the volume of seller-initiated trades.

The measure runs from -1, all selling, to +1, all buying. If 60,000 shares trade on buyer-initiated orders and 40,000 on seller-initiated orders in five minutes, the imbalance is $(60{,}000 - 40{,}000) / 100{,}000 = 0.2$. Related measures use changes in the bid size and ask size at the best quotes, capturing orders that are placed and cancelled as well as those that trade. Computing them requires tick-by-tick data.

Economic theory explains the link through information. In models developed in the 1980s by Albert Kyle and by Lawrence Glosten and Paul Milgrom, some traders know more than the market maker about the value of the asset. The market maker cannot tell who they are but knows that informed traders buy when the price is too low and sell when it is too high. A run of buy orders is therefore evidence that the asset is worth more, and the market maker raises quotes in response. Prices move in the direction of the order flow, and the size of the move per unit of flow measures how illiquid the market is.

This risk of trading against better-informed counterparties is adverse selection, and it is one reason a bid-ask spread exists. The process by which trading reveals information and incorporates it into prices is price discovery. Empirical studies support the theory: in foreign exchange, for example, research by Martin Evans and Richard Lyons found that order flow explained a large share of daily exchange rate changes that economic fundamentals could not.

• Market making. Liquidity providers monitor the imbalance of incoming orders and widen or shift their quotes when flow becomes one-sided, since one-sided flow often precedes a price move.
• Short-term prediction. High-frequency trading strategies use order flow and order book imbalance to forecast price changes over the next seconds.
• Execution. Algorithms that work large orders track prevailing flow to decide when to trade aggressively and when to wait, with the aim of reducing market impact.
• Discretionary trading. Some traders read the sequence of trades and quotes directly, a practice once called tape reading, using displays of traded volume at each price.

In a commercial sense, order flow refers to the orders a broker receives from its clients. Orders from retail investors are, on average, less informed than those from institutions, so trading against them is less risky. Wholesale market makers are willing to pay brokers to receive these orders, an arrangement called payment for order flow, and execute them against their own inventory, usually at prices slightly better than the public quote. The practice funds commission-free retail trading in the United States. It is criticised for the conflict of interest it creates for brokers, and it is restricted or banned in several other jurisdictions.

Order flow is informative only on average and over short horizons. Much of it is noise: trades made for liquidity, hedging or index-tracking reasons that carry no view on value. Classification of trades is imperfect, and a large share of volume now occurs in dark pools and other venues where it is reported late or without detail.

Because others watch order flow, it can be faked. Spoofing places orders that are not intended to trade in order to create a false impression of buying or selling interest, and it is illegal. Knowledge of pending client orders can also be misused through front-running, in which a firm trades ahead of an order it knows is coming.