The 2010 ‘flash crash’ and subsequent market-technology disruptions have underscored the need for market data that regulators can use for forensic purposes. Data printed to the consolidated tape isn’t real-time, nor is it detailed enough to reconstruct precisely what happened at any given point in time.

While the tape does provide data on the price and size of the best bid and best offer for each stock on each exchange, it does not provide information on orders placed below the best bid or above the best offer.

That has led the Securities and Exchange Commission to deploy MIDAS (Market Information Data Analytics System), which combines advanced technology with empirical data to promote better understanding of markets.

“The flash crash of May 2010 and the subsequent report by the SEC and CFTC entailed a lot of headaches because it was difficult to get the data,” said Charles Collver, financial economist at the SEC’s Office of Analytics and Research, in a conference call on Wednesday sponsored by the Security Traders Association.

The actual analysis involved many weeks of data sourcing, event reconstruction, plus actual analytics once the data was in place.

“At one point in time or another, it involved dozens of people including analysts, consultants, economists, lawyers and many others from both agencies working together,” Collver said. “Today, that very same heavy lifting, where you get the data and put in in a single place where it can be accessed and manipulated and understood, can now take a matter of a few days maybe or even less [using MIDAS].”

MIDAS, which was built for the SEC by Tradeworx and went live in 2013, contains a historical database of all orders printed to the proprietary feeds. “We get the orders that are printed, so that means we receive all orders that are intended for display, and we receive all of the modified or cancellation orders associated with displayed orders,” Collver said. “We also see when an order interacts with an order on the book that results in an execution. For the most part, that means we see lit orders and almost all trades.”

There are a range of MIDAS users across the agency, including the divisions of trading and markets, economics and risk analysis, enforcement, and compliance and inspections, among others. “My own office has several dedicated MIDAS users, who are not only expert programmers and analysts, but also possess considerable knowledge of our financial markets and the associated data,” said Collver.

One type of analysis that MIDAS supports is event reconstruction, where analysts will use the MIDAS system and the tools that are associated with it to investigate and understand one-off events in the market.

An example is a mini flash crash, in which a particular stock or symbol will drop three percent or five percent in a matter of a seconds, then pop right back up, shoot up for a few seconds, and then bounce right back to its original price.

The Office of Analytics and Research uses MIDAS to generate a variety of metrics and associated analysis to help the SEC better understand how modern markets work, and to inform policy decisions.
Some of the initial research efforts involved a better understanding of message traffic, order lifetimes, odd lot trading, and dark trading on exchange, said Collver. Current efforts focus on small cap liquidity, sweep order trades, and a more complete understanding of the NBBO.

“Going forward, I think we going to be developing an ever increasing list of metrics, based on our internal research and requests that we get,” said Gregg Berman, associate director of the Office of Analytics and Research.

Featured image via tmc_photos/Dollar Photo Club