Authors: Mark Russinovich
“You know,” Frank said, eyeing his pair of chopsticks dubiously, “we haven’t been spotted yet. At first I kept thinking an alarm’s going to go off, but instead we’ve got the run of the place. I understand why the antivirus programs don’t know we’re there, since we aren’t in their database, but their other automated security programs ought to be spotting our presence. They continuously monitor operation commands and functions. If any company in the world understands how to mine data looking for the smallest hint of something unusual, it would be the Exchange—at least that’s what I thought.”
“So far, we’ve only planted a bit of code, and that looks legitimate. All we’ve really done is take a look.” Jeff smiled. “And we’ve been clever.”
Frank split the chopsticks apart, then tested them in his right hand. “They don’t know we’re there, so we can set up all the offshore accounts we want and move money into them. Of course, it would leave a trail, since computers are keeping tabs on the money, but there’d be nothing to stop us. The trick is leaving nothing behind that points to us personally, then whipsawing the money around the world until it’s impossible to trace.”
“Do you really think that’s possible in this day and age? It seems to me that every digital transaction can be traced.”
“In theory, sure, but if you’re clever about it, create a host of dead ends to mask the money trails, then bury all of them in complexity, you can slow such a search to a crawl. In practice, you can make it never ending. It would take a dedicated team and time, but it can be done. We saw terror groups doing that with the money they raised and stole all the time when I was with the Company. We did sometimes catch it, but we knew what we found was just the tip of the iceberg. The Internet, Jeff, is as close to infinite as anything on Earth. You don’t have to block anyone trying to trace you, even if it were possible; using robo code, you just have to keep stretching the trail ad infinitum. It works out to the same thing.”
“Maybe. Better, though, if the Exchange never knew the money was taken in the first place.”
Frank pursed his lips. “Yes, but how do you do that?”
“Maybe take it directly from clients’ accounts, a bit here, a bit there, keeping in mind that a ‘bit’ in this case is a few hundred thousand, maybe a million at a time. Take a penny of every dollar out of transactions, for example. They might not even notice, and even if a client sees the loss, the Exchange doesn’t.”
“But if enough of them complain, the Exchange will get on it.”
“You conceal the loss within their trading patterns so it doesn’t look as if it’s an Exchange issue. You know bureaucrats, always looking to avoid problems if they can. If you aren’t greedy, all you’re doing is skimming a bit of the cream each time. It might raise a few eyebrows, but there’s no reason—in theory, at least—to cause any serious research. I think that’s the better way to do it. Then you can bury it with electronic false trails like you say. And it’s really only a variation of what the high-frequency traders are already doing, especially those hiding offshore.”
“Good thing we’re honest.” Frank jabbed at his rice with chopsticks. He finally put them down and picked up a fork. “You know,” he said, “I’m thinking about moving my nest egg out of stocks.”
“Why’s that?”
“I don’t like a lot of what I’ve seen, but it’s these high-frequency traders that really get me.”
“What about them?”
“I don’t care if a company finds a way to buy and sell faster. Paying for close physical access isn’t fair, but those with money always have an edge like that. The problem with high-frequency traders is their manipulation of the financial system. And because they’re allowed to hide what they do, no one really knows the extent of the manipulation.”
“I didn’t know you were such an expert.”
“I’m not, but I’m getting there. Actually, I’m reading a book about it. It’s really eye opening.”
“You’re obviously not working hard enough if you have time to read a book.”
“It’s part of research. A vital part, from what I’m seeing.”
The problem, Jeff and Frank had realized from the beginning, was that understanding in detail how the Exchange worked was extraordinarily complicated. The professionals making their fortunes from Wall Street employed their own jargon, in part to convey ideas effectively but also to safeguard their propriety access to the lucrative trading system. Once stripped of the needless complexity, the system wasn’t that incomprehensible.
“Buy low, sell high” was still the lifeblood of trading. Computers and their role in the international market had caused that basic rule to become more complicated than ever, but it remained the essence of the Exchange. When someone wanted to sell, they offered the stock at a specific price. When someone wanted to buy, they listed the price they were willing to pay. Between them was a difference. When one party moved, the transaction took place, not physically off to the side of the trading pit as had occurred at one time, but with nearly unimaginable speed.
Algorithms zipped through the Exchange’s computers, searching for deals within the parameters the programmers had established. When a trade that fulfilled the parameters was found, it was made faster than the blink of an eye, with no human interaction.
The essence of this had always been to stand at the front of the line because there were always more buyers for deals at the right price than there were sellers. The logic was simple enough: More buyers at the listed price drove the price up. The stock available at a desirable price was gone before all the buyers were satisfied. Since getting to the front of the line was essential, the Exchange had a rule—the first to offer to buy was placed ahead of those to follow.
As there was more than one exchange in the world, stocks could be offered for sale or to buy at different prices at the same time. But like water seeking its own level, given time, every stock had but one price. The opportunity came when a delay existed in settling on that common price. In a process known as arbitrage, computers networked around the world reported differences in prices, and algos exploited discrepancies the instant they were discovered. HFTs made money if the difference was an increase in price, and most of them made money by short selling—that is, making money if the price fell. The difference was exploitable either way.
Those opportunities had always existed, but now with computers, they were hunted down as never before, and the chain of transactions took place at unbelievable speeds. This was the red meat for high-frequency traders and as a consequence accounted for a substantial majority of all trading activity, a percentage that grew with each passing month.
HFTs designed and unleashed more sophisticated programs than other trading systems. They paid for proximity to the Exchange’s hub engines to get themselves to the head of the line, beating out more remote competitors. They also possessed a comprehensive understanding of the market’s microstructure. No other traders understood exactly how the trading engines worked, precisely how trades were executed, how orders were prioritized—but HTFs did.
“I don’t know,” Frank said as they finished their meal. “It just seems to me that the stock market doesn’t work any longer, not in any logical way. It’s so complex and fragmented, no one’s got a clear understanding of how it functions. It doesn’t even seem to be about providing a marketplace where people can buy and sell securities. There’s all this other stuff going on all the time. It’s all smoke and mirrors, altered reality, like a video game. What’s scary is that I don’t think anyone understands all the new rules or the full extent and implications of this permissiveness. There used to be just a few kinds of trades and only a couple of places to make them. Now there are more than one hundred types of trades, and if you add the variations, it’s well over that. And there are plenty of places where you can execute them. It’s all intentionally complicated, if you ask me.”
“It’s computers,” Jeff said. “They’re a curse and blessing. They make high-frequency trading possible. And the billions of dollars they’re taking all comes out of the pockets of pension funds, 401(k)s, and regular investors.”
“It’s worse than you think. As far as I’m concerned, these HFTs simply manipulate the market, like I said. Consider this: In an old-fashioned traditional physical trade, a man with a bag of money might stand behind the buyer, demonstrating his interest in buying even more stock if it became available. He never needed to actually bid on the stock to influence the price; simply existing was enough. His presence alone tended to drive up the price. It worked the other way as well.
“These HFTs have perfected a system in which they can appear to be that guy with the big bag of money ready to buy,
but without ever actually executing an order
. The consequence is that the sale or buy price moves, and once it moves in its favor, the HFT programs execute in a millisecond and make a profit. All the while, no one knows if the guy with the big bag of money even exists. In most cases, the HFT never owns the stock it is offering to sell. If things don’t move the way they want—if the difference they spotted disappears before their order is filled—they just cancel it. As a consequence, over half the volume of orders processed by the Exchange are canceled. It’s nothing less than tampering. They aren’t making a legitimate offer to buy, or sell. They are trying to move the price to a point where they can make money. And nothing stands still. The HFTs are constantly inventing new, profitable ways to exploit financial transactions, novel ways only the sophistication—and speed—of their servers make possible.
“Listen to this,” Frank continued. “In May 2010, the financial markets plunged into free fall with no warning. In just minutes, the Dow plummeted almost one thousand points, something like nine percent of its value. It was the biggest single-day drop in history. Nearly one trillion dollars disappeared into digital vapor.”
“If you say so. I must have read about it.”
“We all did, and if you’re like me, it didn’t mean a thing at the time. Some shares fell in value to a single penny, if you can imagine, only to rebound to, say, thirty-five dollars within seconds. And in high-frequency trading, ‘seconds’ is a very long time. It worked both ways. Apple, for one, briefly traded at a hundred thousand dollars a share—can you believe it?—up from around two hundred fifty. It wasn’t alone. Then, within minutes, the stock market righted itself and recovered its losses.
“Trades were taking place so fast, a delay of thirty-six seconds crept into the system. Now, that’s a lifetime in this world. What it meant was that what the computers, even the real traders, were seeing was half a minute away from reality, so they were acting based on dated data. It was like 1929 all over again. They bought and sold, thinking the price was going one way, when actually, it was moving the opposite direction.
“The collapse was so extreme, so profound, it was like watching a pedestrian being struck by a speeding car. Then, as you tried to deal with the horror of what you’ve just witnessed, the victim stood up, brushed himself off, and walked away as if nothing had happened.
“Traders were shocked at what they’d experienced. Everyone wondered if it would start again. But it didn’t. They decided it was an anomaly, so they went back to business as usual right after. The SEC,
Wall Street Journal,
hedge funds, all looked for an explanation without success. All they did was give it a name—Flash Crash.
“Now, nothing like this could happen before computers, and that’s the point. Traders seeing such fluctuations in prices remotely approaching these would have used common sense and not participated. They’d stand down. But the computers applied the logic of their algos and reacted instantaneously, without consideration of the consequences or logic of the situation. Stop and think about this for a minute. If a thirty-five-dollar stock fell to, say, two cents, would you sell? Would you even be taking part in what is clearly a bizarre phenomenon? Of course not. You have common sense, you know something is very wrong. But the computers don’t think; they act. They sell at two cents, they buy at two cents,
if
there’s money to be made. Reality has nothing to do with their world.
“There was a lot of unease, even though the market recovered. They’d seen the pedestrian walk away, but it gave them no comfort. Everybody was really nervous, and the lack of an explanation only made it worse. There was the suspicion, the near certain belief, that high-frequency trading was behind it all. About a half year later, the Security Exchange Commission’s report was finally released. Just one enormous trade had gone terribly wrong, it said. That’s it. And get this, the financial markets weren’t prepared for it! That vulnerability, the SEC said, was because of the aggressive selling by HFTs whose computers had responded automatically to the market’s illogical behavior, exactly as traders had suspected. You get a few HFT algos responding to each other’s movements, and things spiral out of control before anyone realizes what’s going on. It’s not humans trading with each other; it’s computers. Cross-market arbitrageurs, looking to score a quick profit, piled on and drove prices down something like three percent.
“What the HFTs did was issue nonexecutable orders in batches. These were intended to detect early trends or to test latency. Critics claim their purpose was also to clog the exchanges, to create noise, to outmaneuver competitors. Pretty cynical.
“Now, stay with me here. You have responsible high-frequency traders. Their algos are set to pull out when they see erratic behavior. But you also have irresponsible high-frequency traders who are not so smart. So what happened was the smart HFTs left the trading field to the dumb ones. That’s why you had such crazy trading decisions.
“Here’s the big part, in my opinion. All this was caused by a trade of just over four billion dollars on a day in which the volume was two hundred billion. Think about it. What would a larger trade do if it went awry? Four billion’s a lot, but even more is not out of reach. What if the algos causing the disaster failed to correct themselves? What if the exaggerated prices remained fixed for more than a few seconds? And remember, these guys have all been forewarned now. They aren’t likely to be so patient next time. The Exchange keeps telling everyone they’ve put a stop to these computer issues, but there’s no confidence any longer in the integrity of the system. The same thing happened on the Shanghai Composite not much later. It collapsed six percent in just two minutes.