(Kitco News) - It was the “flash crash” of May 2010 that woke Eric Scott Hunsader to the idea that high-frequency trading might have a hand in trading disruptions.
The “flash crash” caused the Dow Jones Industrial Average to fall 9%, but the index recovered those losses within minutes. Since then, there have been other trading glitches, whether they affect individual stocks or exchanges, which market participants blame on high-frequency trading.
Because of the “flash crash,” Hunsader, owner of market technology firm Nanex, has taken to social media and other forums to publicize what he sees as a development that’s overwhelmed current market capabilities -- allowing certain trading systems to get an unfair trading advantage and creating an atmosphere that may ultimately harm markets in the long-term.
High-frequency trading has its fans, saying the improvement in trading technology has brought down costs and allowed retail participants to use software previously unavailable to them. Further, they say, there’s always been some sort of complaint about unfairness in trading, whether its high-frequency trading now, or in the days of the trading pits, someone being taller than another person. Mishaps have always occurred in trading, whether it was dealing with out trades during the days of pit trading or glitches in electronic trade, proponents of high-frequency trading add.
Another issue, Hunsader says, is that high-frequency trading, also known as algorithmic trading, not only doesn’t allow other traders to take advantage of price discrepancies because they happen so quickly, but that by putting out so many quotes and then pulling them immediately, it clogs the market with the equivalent of trading spam.
While the bulk of what Hunsader sees as high-frequency trading causing trading disruptions is done in the equity and equity options market, algorithmic trading has also caused price spikes in the futures market, he said.
Some of those quick price spikes, particularly ones that caused prices to temporarily fall, caught the eye of gold traders, many of whom are quick to call foul when gold prices fall.
FINANCIAL MARKET SOFTWARE DEVELOPER
Hunsader is no recent upstart and he’s not anti-technology. Hunsader got his start back in 1986 by creating an automated trading system to trade Standard & Poor’s 500 stock index futures. His firm, Nanex, streams whole market data feeds from exchanges.
“That’s our primary business. The research that we do, we don’t monetize, we simply put it out there because of how bad things have gotten in the market in hopes that it can still be saved,” he said.
Nanex archives all of the market data feeds – Hunsader said they have nearly 10 trillion quotes and trades stored – so when the “flash crash” happened he and his team wanted to find out more about the event.
Their data go back to 2003, and Hunsader said they started to notice changes in trading patterns and high quote-to-trade ratios.
“I guess it dawned on us that a lot of the message traffic we were processing weren’t trades, they were quotes, put in there and cancelled immediately. It kind of ruins your day when you realize that 95% of the hard drive space, network space, you name it, is gone to this silly information. We were getting less good information and it was a lot more expensive to deal with,” he said.
Hunsader said high-frequency trading usually falls into two categories: one, where an algorithmic system floods the market with quotes to see if anyone bites, and then quickly cancels them. Many of those quotes may only last milliseconds, he said. The second type trading is for a system to flood quotes in an effort to deflect what it is actually doing.
He dated the rise of these types of algorithmic trading to mid-2007, not long after the Securities and Exchange Commission’s passage of Regulation NMS in 2005.
At least one SEC regulator, Luis Aguilar, suggested in December revisiting the regulation to see if the rules have led to unintended consequences, as critics like Hunsader charge.
Related Stories:
- Traders To Be Mindful Of Weather Impact On U.S. Payrolls Data
- Gold After The Lunar New Year; Jeff Christian Looks Ahead
Because there are several equity and equity options exchanges, with many of those exchanges listing the same stocks, it makes this type of trading much more of an issue than it does in futures markets where there generally is only one exchange trading one commodity, such as gold, he said.
But it does happen, he added. He said some of the Comex “stop logic” events, where trade is halted for a few seconds or a few minutes, were likely because of an algorithmic trade. The idea behind the stop logic events, he said, is when a big order to buy or sell comes in and may move the market far out the current trading range.
To clarify, not all “stop logic” events are caused by algorithmic trading. Circuit breakers are in place to keep order in the markets and have been for years, even before high-frequency trading debuted.
When those spikes happened, some gold market watchers would cry manipulation because prices fell. But Hunsader said algorithmic trading doesn’t cause prices to only fall. “We saw a few of those moves on big up moves,” he added.
He was quick to point out that he’s not against price swings, but what the high-frequency trading does is limit how quickly others can react to the price aberrations because they move so fast.
“When the market moves so fast, say gold plunged $20 and came back in a second, you can’t participate in that. People would love to buy it there, but it doesn’t last long enough. You can have these big wild moves that trigger circuit breakers, but if more people participate you wouldn’t have that. You’d have people be willing to buy it $5 down, $10 down,” he said.
And comparing this type of trading to spam is not unfair, he said. “It’s the same kind of model; it doesn’t cost them anything, but everyone else has to process it downstream. It’s harder for the average trader to see what’s going on. They can’t participate in the sub-second moves,” he said.
DOES IT MATTER?
Some academic research papers show that high-frequency trading only affects the very short-term trading patterns, but doesn’t affect long-term trading patterns or fundamental price trends.
Hunsader agreed to a point.
“It (matters) during that period of time (it occurs). I think you wouldn’t even notice that it happened the next day, except for the people who got caught,” he said.
And those people might change their trading behavior because of that. “They might not participate anymore; maybe they’ll see what bitcoin is up to instead,” he said.
It’s the lack of participation that concerns him as fewer market participants mean a less robust market.
Because of all this extra information, Hunsader said it makes it not only more costly to store archived information, but it also makes it expensive for academics and others who study historical data since they have to buy much more information to do their work.
“The historical data price is too high. The number of people in academics who can look at flash crash data is significantly smaller than it used to be. So you don’t have eyes looking at this data to tell us what this means,” he said.
Read the latest news in gold and precious metals markets at Kitco News.
By Debbie
Follow me on Twitter at @dcarlsonkitco
