Execution Strategy

An execution strategy is the method used to carry out a trading decision once it has been made. It sets how quickly an order is completed, how it is divided into smaller pieces, which trading venues receive those pieces, and what types of order are used. The decision to buy a million shares is an investment decision. How to buy them without pushing the price up is a question of execution strategy.

The aim is to keep the total cost of trading low. For small orders in liquid markets the choice hardly matters. For large orders it can outweigh the commission many times over. Execution strategies are now mostly implemented as algorithms supplied by brokers or built by investors, and they form the part of algorithmic trading concerned with how to trade, as distinct from what to trade.

Every execution strategy balances two costs. Trading quickly demands liquidity from the market at once, and a large order moves the price against itself. This is market impact. Trading slowly reduces impact, but leaves the order exposed for longer to price changes that have nothing to do with it. This is timing risk.

A patient strategy has low expected cost and high uncertainty. An urgent one has high expected cost and low uncertainty. The right balance depends on the size of the order relative to normal volume, on the asset's volatility and spread, and on how fast the information behind the trade is expected to lose its value. This framework was formalized by Robert Almgren and Neil Chriss around 2000 and underlies most modern execution algorithms.

• VWAP. The order is spread through the day in proportion to the expected pattern of market volume, aiming to match the volume-weighted average price.
• TWAP. The order is cut into equal slices sent at regular intervals, aiming at the time-weighted average price.
• Percentage of volume. The algorithm trades a fixed share of actual market volume, for example 10 percent, speeding up and slowing down with the market.
• Implementation shortfall. Also called arrival price. The algorithm trades faster at the start to stay close to the price when the order arrived, and adapts its pace to conditions.
• Liquidity seeking. The algorithm searches displayed and hidden venues, including dark pools, and trades when size is available.
• Passive and hidden orders. A resting limit order earns the spread instead of paying it, and an iceberg order shows only part of its size.

Within any of these, a routing component decides where each slice goes. Order routing logic compares prices, fees and the chance of a fill across venues.

The standard measure of execution cost is implementation shortfall, a concept introduced by André Perold in 1988. For a buy order, expressed in basis points:

• $P_{\text{exec}}$ is the average price at which the order was filled.
• $P_{\text{arrival}}$ is the market price when the order was handed over for execution.

For a sell order the sign is reversed. A fund sends an order to buy 100,000 shares when the price is $50.00. The algorithm completes it at an average of $50.06. The shortfall is (50.06 − 50.00) ÷ 50.00 × 10,000 = 12 basis points. In money, the cost is 0.06 × 100,000 = $6,000, before commission.

Systematic study of such figures is transaction cost analysis. It compares fills with benchmarks such as arrival price, VWAP and the closing price, and helps investors choose between algorithms and brokers. The benchmark shapes behavior: an algorithm judged against VWAP will track volume faithfully even when trading faster would have been cheaper.

Order size is usually stated as a fraction of average daily trading volume. An order of 1 percent of a day's volume can be completed quickly with little impact. An order of 30 percent needs to be spread over the day or several days, or negotiated as a block trade. Urgency matters as much. A trade based on news that is about to become public calls for speed. An index fund adjusting to a change in its benchmark wants the closing price and will concentrate its trading in the closing auction.

Liquidity conditions also guide the choice. A wide bid-ask spread favors passive orders, and a thin book favors smaller slices. Many algorithms adjust in real time, trading more when the price moves in the order's favor.

A predictable strategy can be detected. Regular slices at fixed intervals leave a pattern that other traders may identify and trade ahead of, so algorithms randomize sizes and timing. Passive strategies risk not completing, leaving part of the order to be filled later at a worse price. Sending orders to many venues can reveal the trader's intentions, an effect called information leakage.

Brokers that execute for clients owe a duty of best execution, which requires them to take reasonable or sufficient steps to obtain the best result, taking account of price, cost, speed and likelihood of execution. Firms must be able to show how their execution strategies and venue choices meet that duty and must monitor the results.