An execution algorithm needs price and time limits

Graduate · enrolled learners

This lesson opens with Waddell and Reed and the Flash Crash, 6 May 2010.

What happened

A large asset manager used an automated execution algorithm to sell about 4.1 billion dollars of stock-index futures to hedge a position, and the algorithm was configured to target a fixed percentage of trading volume, around nine percent, with no limit on price and no limit on time. In an already-nervous and thinning market the algorithm kept selling in proportion to volume, and because its own selling helped raise volume, it sold faster as prices fell, feeding a spiral. Its selling was a major contributor to the Flash Crash, in which major indices fell about nine percent and then recovered within minutes. The algorithm was not broken and no key was stolen; it did exactly what it was told, and what it was told, sell at nine percent of volume regardless of price or time, was an execution instruction with no brakes. The decision to trade may have been sound; the way the order was worked, relentlessly and without limits, was the failure.

The decision point

How an order is executed, worked into the market over time and price, is a risk decision in its own right, separate from the decision to trade at all, and an execution algorithm given no price limit and no time limit will keep trading no matter how far the market moves against everyone, including the trader running it. Waddell and Reed is the case: a percentage-of-volume sell algorithm with no price or time bound kept selling into a falling, thinning market and helped drive the Flash Crash, because its instruction, match nine percent of volume, contained nothing that could tell it to slow down or stop as prices collapsed. This is a foundational execution lesson: the choice of how to work an order, market or limit, how fast, over what horizon, with what caps, determines the market impact and the risk of the execution, so an execution algorithm is not a neutral pipe that carries out a decision but a strategy of its own that needs its own limits. An algorithm told only what to achieve, a target size or a share of volume, and not what it must never do, trade through a price, keep going past a time, will pursue the target off a cliff. So the discipline is to build limits into every automated execution: a price beyond which it will not trade, a time or horizon after which it stops, and a cap on how aggressive it can be, so that the execution has brakes the market itself does not provide, because Waddell and Reed showed that an execution instruction with no brakes will, faithfully and catastrophically, chase the market all the way down.

Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.

It is free. We do not sell the list and there is nothing to buy at the end of it.