Stuff that Quote!

I posted a couple weeks ago on creepy botnets placing tens of thousands of orders on the nation’s electronic exchanges.

Both Reuters and the Wall Street Journal now report that the phenomenon— which now seems to be called “quote stuffing”—  is under investigation by the SEC and the CFTC.   Quote stuffing is now suspected of being behind the Flash Crash of May 6, when the Dow lost ten percent of its value in only a few minutes.

The quote stuffers place tens of thousands of buy or sell orders on the markets in just a few seconds, then immediately cancel the orders.

The Nanex study uses market graphics and playful names to illustrate quote stuffing, arguing that high-frequency trading firms do this to flood the marketplace with bogus orders to distract rival trading firms.

Investors could make trades under the false impression that those orders were legitimate, only to see liquidity disappear and the market move against them when the orders are canceled — all in the blink of an eye.

Trader’s Narrative explains how this worked on May 6:

According to Nanex ((The data vendor who discovered the phenomenon)), the fault lies with delayed NYSE quotes. In a recent report on their website, Nanex claims that the NYSE started crossing quotes at 2:42 PM on May 6th. Since the quotes were timestamped with the time they exited the queue, instead of when they were created, the algorithms had no idea they were stale. This created an arbitrage mirage because it appeared that the NYSE bid was slightly higher than the competing exchanges. Naturally high frequency trading systems jumped at the chance and created a vicious cycle:

In summary, quotes from NYSE began to queue, but because they were time stamped after exiting the queue, the delay was undetectable to systems processing those quotes. On 05/06/2010 the delay was enough to cause the NYSE bid to be just slightly higher than the lowest offer price from competing exchanges, but small enough that is was difficult to detect (See Part 3, The Evidence). This caused sell order flow to route to NYSE — thus removing any buying power that existed on other exchanges. When these sell orders arrived at NYSE, the actual bid price was lower because new lower quotes were still waiting to exit a queue for dissemination.

This situation led to orders executing against whatever buy orders existed in the NYSE designated market maker (DMM) order book. When an order is executed, the trade is reported to a different system (CTS) than quotes (CQS). Since trade report traffic is much smaller than quote traffic, there is rarely any queueing or delay.

Because many of the stocks involved were high capitalization bellwether stocks and represented a wide range of industries, and because quotes and trades from the NYSE are given higher credibility in many HFT systems, when the results of these trades were published, the HFT systems detected the sudden price drop and automatically went short, betting on capturing the developing downward momentum. This caused a short term feed-back loop to develop and panic ensued.

The Trader’s Narrative article has cute graphs an’ everything.  Nanex’s conclusions?

Competition between HFT systems today has reached the point where microseconds matter. Any edge one has to process information faster than a competitor makes all the difference in this game. If you could generate a large number of quotes that your competitors have to process, but you can ignore since you generated them, you gain valuable processing time. This is an extremely disturbing development, because as more HFT systems start doing this, it is only a matter of time before quote-stuffing shuts down the entire market from congestion.

In fact, Nanex claims this happened twice before the Flash Crash, but no one noticed.

Back in my youth, I used to play historical games created by companies like SPI and Avalon Hill.  I would cheerfully march my little cardboard counters over hexagon-covered gameboards, rolling dice and consulting rulebooks and charts while engaged in something that claimed to be an analog of the Battle of Kursk or the Normandy Invasion.

Because, though eccentric, I was somewhat sane, I knew that what I was doing was in fact entirely unlike the experience of actually participating in the Normandy Invasion.  (More in the way of beer and Fritos, for one thing.)  There were those players who rather thought that participation in Avalon Hill games turned them into actual military geniuses, and these were players that I did my best to avoid.

Computer programs acting as high frequency traders are very much like those deluded gamers.  Though they invest vast sums on the electronic exchanges— amounting at last count to 73% of all trades—  they actually know nothing about goods and services, cost of production, labor services, capital, productive efficiency, supply, demand, efficiency of management, industrial organization, demand for labor, externalities, value, and other factors that professional traders have to understand. They knew considerably less about economics than I knew about the actual Normandy Invasion.

What the bots have instead is very, very detailed information about past behavior of the market, and sophisticated algorithms that tell them when to intervene.  When something moves on the market, they can reference the previous times that the market moved in that direction, and anticipate what happens next.

Because these systems control huge amounts of capital, for the most part they act to normalize the market.  They shove it in the direction that it “ought” to move, according to history and their own lights.

What they can’t do is react rationally to an unprecedented situation.  (Russia’s bankruptcy?  North Korea nuking Japan?  Black swans attacking American computers with EMP?)  They’re bound to misinterpret it and very likely react in the wrong direction.  And when their own information is corrupted, they go absolutely mad, as apparently they did on May 6.

(Of course, if you want to know where this kind of thing can lead, you should hie yourself to the nearest bookstore and pick up This Is Not a Game.  On the whole, it’s a far more sound investment than you’re likely to find on Wall Street.)

Nanex, the company that exposed the practice, offers three solutions to the problem:

1. Quote and trade data must be time stamped by the exchanges at the time it is generated. This will ensure delays can be detected by everyone.

2. Quote-stuffing should be banned.

3. Add a simple 50 millisecond quote expiration rule: a quote must remain active until it is executed or 50ms elapses. If the quote is part of the NBBO, it may be improved (higher bid or lower offer price) at any time without waiting for the expiration period.

#2, while a good idea, is very difficult to do in practice.  How do you tell whether a canceled quote has been “stuffed” or whether it was canceled for some other reason?

#3 is a fine idea, slowing down the trades, even by 50 milliseconds, in order to gain everyone a little time and head off a panic.

I prefer the system which I believe is in place in Britain, placing a tiny, tiny tax on every offer, which would make it very expensive to make thirty thousand orders that you have no intention of executing.

The SEC is also investigating “sub-penny pricing,” in which “a sub-penny order sitting in a dark pool, where prices are not publicly displayed, can trade before orders in displayed markets, possibly giving a false impression of the degree of buying or selling demand.”

Christ, spare me.