Trade Execution Cost, Speed Inversely Related Intraday

May 7, 2008 ( - A new statistical analysis of trading data during the trading day from the NASDAQ Stock Market shows an inverse relationship between two important considerations for traders: execution speed and price impact.

Two university researchers who studied   5.2 million NASDAQ stock orders (9 billion shares) placed over a nearly four-year period said the optimum time to execute trades depends on the trader’s preferences and the type of order being put into the market.

Researchers Ryan Garvey of Duquesne University and Fei Wu of Massey University College of Business, who authored the study for the Social Science Research Network, said their examination of orders originating from 2,979 brokerage accounts throughout the U.S., over a near four-year period was different than prior efforts because it tracked “execution quality” trends during a trading day. Prior research has focused on the trading trends in different market venues.

“Traders will have different preferences for execution speed, execution cost, and other dimensions of execution quality based on their different trading strategies and objectives,” the researchers said. “A starting point in considering when to trade is for traders to identify what dimension(s) of execution quality is most important to them.”

The key finding was an offsetting relationship between speed and execution. For example, orders submitted midday take significantly longer to execute than orders submitted around the market open and close, the authors said. However, midday orders execute with lower execution costs.

According to Garvey and Wu, when more informed traders are in the market around the open and close of trading, more price volatility occurs and execution cost rises. However, execution speed slows around the open and close because informed traders have a preference for trading quickly in order to profit from their private information.

Order size is lowest, price impacts are highest, and traders submit more marketable orders around the open and close of trading, the researchers said. Holding everything else equal and controlling for various market, stock, and trader characteristics, the study found that the execution cost for marketable orders in the 9:30 to 10:00 a.m. period is 8.76% higher than that for the midday benchmark period 12:31 to 1:00 p.m. The study also found that the execution cost for the 3:31 to 4:00 p.m. period is 2.26% higher than that for the midday benchmark period.

On the other hand,   Garvey and Wu found,midday orders take longer to execute. For example, the 9:30 to 10:00 a.m. period has an average order execution speed 5.45% lower than the midday benchmark period and the 3:31 to 4:00 p.m. period has an average order execution speed 12.44% lower than the midday benchmark period.

Aggregate trading measures such as the number of orders, trades, volume, and dollar value in the 9:30 to 10:00 a.m. period are more than twice that of the 12:31 to 1:00 p.m. period. Trading activity steadily declines from the morning to midday periods and then increases leading up to the close.

"Execution speed and execution cost are inherently integrated and both measures of execution quality are important to traders," the researchers said. "Faster execution speed, among other things, allows traders to more quickly react to changing market conditions and to resolve uncertainties concerning execution price."

In total, ther data consists of over 10 billion executed shares with a trade value exceeding $101 billion. There are over 6 million orders and 8 million trades. The orders were submitted by 3,258 accounts.

The research report is available  here .