Trading Glitch Still Performance Mystery

At 2:28 PM on Thursday, May 5 the Dow Jones Industrial Average was 10,622; twenty minutes later it was 9977, a drop of 645 points, or slightly over 6%, in twenty minutes – a truly incredible performance,especially if you were short.    Real companies like Accenture and Exelon fell so hard they  were trading like penny stocks. In fact they were actually traded at a penny. It wasn't just the Dow component stocks, or the NYSE, all exchanges were effected. Hard to say exactly how much of the loss was due to the mystery glitch.  Markets were trading lower on fears of Greek default.  The first few feet of plunge down that cliff might have been normal markets at work, but at some point, something very strange took over.  It is clear that markets temporarily lost $Trillions in value, although by the close the losses on the day were small enough so it seems plausible that markets had recovered entirely from the glitch and had only “normal” losses.

What happened?  Early speculation pointed to the “fat finger”, essentially a typographical error in a sell order taken literally by a computer (Star Trek lives) that just kept selling.  Maybe a little too much Cinqo de Mayo tequilla made the fat finger a little fatter. Turns out fat finger pointing isn't polite, or accurate.  No one can find the fat finger trade.  To this observer, it also seems unlikely that a single error on one order in any one security would have triggered the breadth of Thursday's carnage – but see below on the performance of exchange and  market interactions, just in case.

Conspiracy bloggers were out in force, with two variations which can be combined for your reading pleasure.  A) Someone figured out how to game the system to make a fortune – remember, the markets recovered but securities were bought and sold at the glitch induced bottom and some made a bundle.  B)  Someone (no names mentioned, but PRC is only initials) was practicing their “Death to the Universe” hacking skills or C) a terrorist  hacker wanted to make a few $ Billion while demonstrating a virtuoso hacking performance.  All sound theories, surely, making for an exciting post:  Unfortunately,  totally unsupported by evidence to date.  Making a buck isn't quite as easy as it sounds – the exchanges will cancel trades that  are associated with erroneous or unusual trading activity.  NASDQ has announced that it will cancel trades in over 250 stocks in which the trade price was at least 60% less, or more, than the price at 2:40.  Other exchanges are taking similar measures.  Even so, the panic created big profits for gold bugs, people who went into the glitch short (including puts, short side ETF's etc) and closed out their positions in the trough, people who established new long positions that were not canceled.  Fortunes were lost and made on Thursday afternoon.

The theory of the moment (Saturday, May 8, 4PM) is that when the  old NYSE (which handles less than 20% of trade volume) switched to slow mode (essentially a safety mechanism, a quick human look at the real market before execution) on some executions, sell orders were routed to other exchanges, but there was no comparable depth of buy side orders, hence computerized order matching drove prices down fast, down to a penny in extreme cases.  The problem may have been exacerbated by another safety mechanism – at least one computerized trading program, and possibly more, shut down due to extreme volatility, drying up  potential buy side order volume.

The appropriate Congressional Committee has already called a hearing for Tuesday.  The theory of the moment already has some shouting that all the exchanges must coordinate their safety mechanisms to avoid counter-productive performance.  All well and good, but won't we need another hearing when the SEC and the Exchanges have actually had time to figure out what happened, including who really tucked away some big profits (not necessarily wrongfully – just useful information when you want to know what really happened).  Even if the theory of the moment is correct, what safety mechanism has the  best performance and will it work even better if every exchange employs it in exactly the same way?  This is urgent, markets don't need fear of a technical glitch adding to their fundamental shakiness right now.  It's also important enough to get it right.

Photo Credit: Frostnova