Trading at the Speed of Light

When you were a kid, did you lie on your back on the lawn and look for shapes in the clouds?

Nanex finds Charlie Brown and unicorns in trading data. Or maybe goblins and Jack the Ripper. IR professionals should know about Nanex. In Boston last month, I asked for a show of hands from IR folks who’d heard the name. No one had.

Nanex is the electronic microscope for markets, zooming in on trades and quotes in thousandths and millionths of a second. They find shapes, patterns. What Nanex calls crop circles.

These are footprints of algorithms. On September 15, 2011, at hours Eastern Time, Nanex discovered that in one second of trading in YHOO, encompassing some 19,000 quotes and 3,000 trade-executions, a number of trades matched at quotes that didn’t exist until 190 milliseconds after the trades occurred. Nanex termed this apparent evidence of time-travel in trades “0.19 fantoseconds.”

Sure, laugh it up. When asked what might limit it, Illinois Institute of Technology HFT expert Ben Van Vliet responded: “The speed of light.”

Really, it’s not yet possible to time-travel in equity trades. We hope. So something else happened. The market relies – we use the term loosely – on the Consolidated Quotation System (CQS) to provide an amalgamated view of trading and top-of-book quoting across the whole national market system. Each quote and trade is time-stamped military-style, HHMMSSCCC, like ours on the Yahoo trades.

But the SEC admitted in its Flash Crash report that most sophisticated traders depend on proprietary feeds from exchanges and market centers, not the CQS.

Wait a minute. Wasn’t the purpose of the national market system to make sure trades occurred at the best price?

Remember the Flash Crash? The SEC blamed trading in the S&P 500 “e-mini” futures contract. Nanex concluded that erroneous time stamps on trades were the cause. Quotes backed up in systems due to extreme levels, and when they received a timestamp, the time was current but the quotes were not, creating massive imbalances.

Arbitrage, milled to its molecules, is buying and selling things for different prices at different times. In the market, one large system, the CQS, provides time-stamped data separately on quotes and trades. Underneath is a massively fragmented market populated with hundreds and maybe thousands of disparate proprietary data streams feeding algorithms that can speed up or slow down in a score of places and five or six asset classes simultaneously.

This is the Medusa your investors face every day. And we’re talking about…say on pay? Dodd-Frank? Social media? I mean, really. It’s getting absurd.