The Non-Trader’s Guide to High Frequency Trading
January 05, 2016
by Kathleen Zhou, W’17
Not much has been written for the Non-Trader, who does not work in a government financial regulatory agency or in the Street or has taken classes where professors drop three-letter acronyms like they’re hot. Consider this the layman’s guide to High Frequency Trading, written by someone who has an interest, WiFi, spare time, and supervisors who’ve pointed me in some interesting directions.
We start with the basics
High Frequency Trading (“HFT”) is the use of complex algorithms to transact, track, and execute a large quantity of orders in a matter of milliseconds.
Michael Kearns, a professor in Penn’s Computer and Information Sciences department, draws further distinction between so-called “passive” HFT and “aggressive” HFT. Passive HFT involves setting limits (e.g. “sell if below x price” and “buy if above y price”), exclusively waiting for the order to be fulfilled. Aggressive HFT involves actually crossing the bid ask spread (a fancy way of denoting a firm, instead of placing orders, is forced to transact at the market’s prices, buying at x or selling at y) and consequently paying the order’s execution costs.
Why does speed matter?
When our friendly neighborhood trader in Philadelphia places an order to buy 200 shares of Amazon, bidding $524.00 a share, the computer then drops that trade in one of eighteen national securities exchanges, not to mention the thirty-odd alternate trading systems (“ATS” or “darkpool”). Our trader’s bid will join a line of bids, and will sit in line until the higher bids are fulfilled and a seller’s ask matches our trader’s bid. Being first could mean our trader gets a better deal. So that’s why HFT firms are eyeing NATO microwave towers; speed is king, and shaving a few more milliseconds could result in more profits.
But! It is obvious the time between an order being placed and an order landing in an exchange can never hit zero. And reducing time by fractions of milliseconds becomes more and more expensive as returns become smaller and smaller. In this humble writer’s opinion, it’s akin to an asymptote at zero. So what will our trader do now?
Your Pretty Little Algorithm
What is perhaps more important than just sheer speed is (increasingly complex) algorithms. If our friendly neighborhood trader invests in high-tech algos, our trader can track trends and price movements before they happen and position orders accordingly. A sound algorithm is capable of using “reinforcement learning” to optimize revenue, predicting price movements from an order book or bids and asks, and “exploring” dark pools to determine the best execution. For an HFT firm, an algorithm is essential. Say a seller of Amazon is currently asking $523.99 a share and a buyer is currently bidding $523.97 a share. Our clever HFT firm’s algorithm could examine these bids and decide Amazon should be selling at $524.02 a share, according to historical data and the current status of the order book, and so ‘cuts’ by bidding $523.98, putting the firm ahead of the original buyer. Once Amazon’s stock price rises to $524.02 a share, our clever HFT firm then turns around and sells its shares, pocketing a pretty four cents a share. A speed advantage and predictive algorithm serves to help the firm gain an advantage.
In equities (a type of security “representing an ownership interest”), algorithms account for 50 percent of trading volume. In other securities (such as treasuries, commodities, futures, etc.) algos make up a similar percentage of trading volume.
On Flash Boys
Seeing as this writer has referenced the above book twice in the opening paragraph, the writer feels obliged to give some background. Michael Lewis, former bond-salesman at Salomon Brothers turned writer and Street critic, published Flash Boys to criticize HFT and alternate trading systems (“ATS”) as exploitive and unfair. The focus of Lewis’ argument is that HFT firms, in their ability to be first and predict price movements before they happen, have some sort of unfair advantage. A choice selection from Flash Boys:
It used to be that when [Brad Katsuyama’s] trading screens showed 10,000 shares of Intel offered at $22 a share, it meant that he could buy 10,000 shares of Intel for $22 a share. … By the spring of 2007, however, when [Katsuyama] pushed the button to complete a trade, the offers would vanish. In his seven years as a trader, he had always been able to look at the screens on his desk and see the stock market. Now the market as it appeared on his screens was an illusion.
The book chronicles Katsuyama as he founds what became IEX, an ATS that adds 350 microseconds to each order to remove the advantage of speed and complex predictive algorithms that firms with HFT firms employ.
Arguments against Flash Boys
Virtu Financial doesn’t call itself a HFT firm. It calls itself a “leading electronic trading firm” and a “market maker.” Market makers supply the holy grail of traders: liquidity. It is both the technical definition outlined in Investopedia (“the degree to which an asset or security can be bought or sold in the market without affecting the asset’s price.”) and, in this humble writer’s opinion, a shorthand to denote risk (a liquid asset is easy to sell and easy to hedge your bets with) as well as the market’s health (a liquid market means lots of transactions and smooth sailing). a Market makers hold shares in a given security and guarantee buy and sell prices for a certain number of these shares in order to facilitate trading. The NASDAQ, one of the largest exchanges in the US, has more than 500 member firms which act as market makers.
HFT firms like Virtu Financial, in their ability to guarantee shares at certain prices and execute orders in mere milliseconds, allows traders to place and exit positions before even blinking. When major market shifts happen (such as the inevitable raise of interest rates or ITG announcing a huge SEC fine) traders can adjust their portfolios as needed.
Does HFT impact me?
Michael Kearns would be skeptical. In his paper “Empirical Limitations on High Frequency Trading Profitability,” he suggests profitability is limited to $3.4 billion (almost two thirds of which would be lost to trading fees) for the entire HFT firm population. In 2008, he determined profits were bounded up to $2.1 billion. For scale, the entire annual U.S. trading volume is worth $52 trillion, meaning at most aggressive HFT profits would constitute 0.05% of trading volume. Kearns’ study examined only aggressive HFT firms (remember, the ones that are forced to execute trades at the market’s prices instead of setting limits) using data from 2008-2009.
But remember our clever HFT firm, the one that ‘cut’ in line by bidding a higher price on Amazon shares? The example I described is often cited as a legal form of front-running, in which our clever HFT firm has, in effect, jumped ahead of the queue by offering a more competitive price.This can certainly make it harder for individual investors without the same level of sophistication to buy the shares of Amazon our clever HFT bought first, and practically impossible to compete on a micro-level, millisecond to millisecond scale. Without the same technology—the high speed fibers, NATO microwave towers, lasers—an individual investor, or a small firm, must turn to other strategies and other approaches to make a profit.
What does this mean for smaller agents, like unsophisticated investors and small firms? Will HFT firms essentially stop pushing for more speed advantages? What impact will HFT and algo trading firms have on the future?
Both high frequency trading and algorithmic trading are both newer technologies. The full extent of their impact on both liquidity and the players in the financial markets remain to be seen. That newness, that potential for profound change (or not) is what makes them such a topic of discussion in The Wall Street Journal and in this post.
 See Michael Lewis’ Flash Boys, published March 31, 2015.
 In its April IPO filings, Virtu Financial claimed out of four years of trading (approximately 1,278 days) it incurred exactly one day of losses. (An interesting note: Virtu Financial also does not call itself an HFT firm, preferring to emphasize its technology roots.) Chief Executive Officer Doug Cifu later regretted trumpeting Virtu’s track record; this writer suspects trotting out a near-perfect trading record just as a book criticizing HFT firms for having an unfair advantage might have backfired. But hey! Any publicity is good publicity~
 The U.S. Securities & Exchange Commission (“SEC”), the U.S. Commodities Futures Trading Commission (“CFTC”), and the Financial Industry Regulatory Authority (“FINRA”), to name a few of the agencies tasked with regulating the financial market. This humble writer interned in the Division of Trading & Markets at the SEC.
 “High Frequency Trading.” Investopedia.com. Web. 9 Aug. 2015. http://www.investopedia.com/terms/h/high-frequency-trading.asp (Buckle down, it’s going to get a lot more complicated than this)
 Michael Kearns, Alex Kulesza, and Yuriy Nevmyvaka. “Empirical Limitations on High Frequency Trading Profitability.” 2010. PDF. http://web.eecs.umich.edu/~kulesza/pubs/hft_jot10.pdf Pages 1-2.
 Intro to Economics: in perfect competition, profits are zero. This makes sense; if a HFT firm makes millions then other firms (or new entrants) are interested in joining the same industry. Not to say the HFT industry is perfectly competitive—your average Joe (or this writer) can’t buy a radio tower or a giant fiber optic line—but the rule, in this case, means that competitors will also increase their speed and any single HFT firm’s competitive edge will be reduced.
 Kearns, Michael and Nevmyvaka, Yuriy. “Machine Learning for Market Microstructure and High Frequency Trading.” 2013. PDF. https://www.cis.upenn.edu/~mkearns/papers/KearnsNevmyvakaHFTRiskBooks.pdf Pages 1-2.
 A four cent profit isn’t exactly much to write home about, but let’s say this was 200 shares. A respectable $8 profit. Let’s multiply this by hundreds, thousands, millions of transactions, and suddenly the 4 cent profit is looking quite respectable indeed.
 “Equity.” Investopedia.com. Web. 10 Aug. 2015. http://www.investopedia.com/terms/e/equity.asp
 Salomon Brothers went under in 2003 under a storm of scandals. It’s the subject of Lewis’ first book Liar’s Poker.
 In 2014, Virtu reported it made money every trading day.
 Front running, which is entirely illegal, is the practice of trading on information before it becomes fully public
If you are ever enticed by brokerages advertising a one-second execution time, I’d suggest taking a more fundamentals-type approach. In one millisecond hundreds of orders have already been placed, canceled, and fulfilled.
Student Blog Disclaimer
The views expressed on the Student Blog are the author’s opinions and don’t necessarily represent the Penn Wharton Public Policy Initiative’s strategies, recommendations, or opinions.