US regulators ask DOE lab to study flash-crash forecasting tool

US regulators enlist Department of Energy computing power to analyse metric that aims to predict catastrophic market events

crystal-ball2

A US Department of Energy (DOE)-funded research lab has been analysing a tool designed to forecast flash crashes, with a view to developing preventive regulation. The Lawrence Berkeley National Laboratory – which belongs to the DOE but is managed by the University of California – was introduced to the tool by a US market regulator.

The lab conducted research assessing the ability of the volume-weighted probability of informed trading (VPIN) metric to predict sharp price drops in index and single stock markets, such as that experienced by the Dow Jones Industrial Average on May 6, 2010, when it dropped 9.2% in a matter of minutes before rebounding.

While the researchers stop short of recommending the implementation of  the metric – which measures the imbalance between buy and sell orders relative to historical levels – as a trigger mechanism for regulatory measures such as circuit breakers, they see it as a first step towards developing a more complete version.

"A US regulator approached us informally and told me about the VPIN metric. They suggested we'd be good guys to analyse it as a potential regulatory tool because we have all this computing power," says David Leinweber, head of the lab's Center for Innovative Financial Technology (CIFT). "It's too early to make a specific recommendation on tying regulation to it, but it's a really interesting approach. VPIN shows you can get a quantitative hold on toxic flow – at least to first order."

A source close to the regulator in question says the research is being watched closely – but some scepticism remains over the metric, first published in January this year by Marcos Lopez de Prado of Tudor investments, and Maureen O'Hara and David Easley of Cornell University.

"It's interesting," the source says, "but what I'm interested in is: how will that look on different days? Is this providing more information than we know already? Does it allow me to spot the aggressor when a toxic order flow arrives? A statistical measure might not tell you about the causality of a crash."

The most obvious regulatory use of the metric would be to trigger a stop-loss mechanism forcing participants to cease trading if the VPIN hits historically high levels. But the CIFT's Leinweber sees this as simplistic, and suggests a more flexible traffic light system that would merely limit trading volumes in a given period. However, the regulatory source is sceptical. "It's not immediately clear that speed is necessarily the problem, or that slowing down will help. Perhaps the answer is actually to speed up trading. We need to look at it more closely," he says.

Such metrics could be used in tomorrow's over-the-counter derivatives markets as well as in equities – some market participants fear the regulatory push to have OTC derivatives trade on largely electronic platforms will be used as a trojan horse by high-frequency traders, potentially making the market vulnerable to an equity-style flash crash. Although there is still debate about the links between high-frequency trading and the 2010 flash crash, there's a growing consensus that the OTC market will increasingly be traded by algorithms.

Leinweber agrees the VPIN is not the finished article. "One big area where we need clarity is the false positives. How often does it say there's a crisis coming when there isn't? It's one thing to shut a market down once something bad has happened – everyone understands that. But will they understand when everything seems OK and you stop them trading?" he says.

"You could expand it to look at the whole limit order book, rather than just the bid/offer spread. It would be good to account for message flows, the role of short-life quotes, and other factors."

The laboratory – commonly known as the Berkeley Lab – was founded in 1931 and contributed to the Allied war effort, including the Manhattan project that developed the first atomic bomb. The CIFT was created in 2008 as a dedicated resource to analyse financial data, led by Leinweber.

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@risk.net or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact info@risk.net to find out more.

Most read articles loading...

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here