Author : Prashant Rishi , IIM Lucknow
It was in late 2000, when the NYSE
decided to quote prices of stocks in decimals of a dollar, as opposed to a
fixed list of fractions. The event (called
decimalization) sowed the seeds of what is today popularly known as High
Frequency Trading or HFT. Stated simply, HFT is trading in stocks by computers,
with minimal human assistance. Carried out by super computers of major
investment banks & hedge funds, high frequency trades range in time from
less than a second to a few hours. Today, it is estimated that majority (~60%)
of all equity trading in NYSE is done by trading algorithms. Although predominantly
into equity, HFT firms have started moving into other asset classes, like
derivatives, FX and fixed income instruments.
Figure 1: Asset classes traded by HFT firms
The obvious advantage that
computers offer in trading assets is speed of processing information and
executing trades. Add to it other advantages like low cost, high execution
consistency & anonymity and you begin to understand why High Frequency
Trading is so popular among all trading desks.
Figure 2: Why funds prefer High Frequency
Trading
Generally, trading algorithms are
built on complex mathematics and statistical modeling. They are designed by Quants
(as Math PhDs are known in Wall Street lingo). The hedge funds, who own these
algorithms, protect them with as much zeal as Google protects its proprietary search
algorithm or Coke protects its secret soft-drink ingredient. Most algorithms typically
employ “flat” strategy, ie trading positions are closed within the same day. Profit
with one such milliseconds-long trade is sometimes only a few pennies, but it
is the massive trade volume that drives the total daily profits, which are in
several thousands of dollars.
Players & Strategies
In the US equity markets, some of the
highest volume high-frequency traders include proprietary trading desks of
firms like Goldman Sachs, Knight Capital Group, Getco LLC & Citadel LLC.
There are 4 basic strategies employed
by almost every HFT firm:
Figure 3: Players in HFT space (US Equities)
Traditional market making involves
placing limit orders to buy & sell in order to earn the bid-ask spread. But
for an HFT firm, the bid-ask spread is not the only source of money. Since market
makers provide additional liquidity to the market by being counterparty to
incoming market orders, they get rebates from exchanges for quotes that lead to
execution. So, if an HFT’s bid (buy order) of $15 for XYZ shares is matched, it
might immediately post an offer (sell order) for the same price, hoping to
capture two rebates while breaking even on the spread. Building up such market
making strategies typically involves precise modeling of the target market
structure & trading volumes using stochastic control techniques.
Ticker
Tape Trading
To appreciate ticker tape trading, it
is essential to understand the concept of “co-location”. Co-location is a
system wherein a stock exchange allows large hedge funds and i-banks to place
their computers near its own data terminals, in exchange for rental income.
Proximity to the stock exchange’s data centre ensures that any market movement
(read the ticker tape) is detected by these computers before general public.
Pre-designed algorithms can thus detect any trend in the prices, and carry out
their own trades seconds before the general public even knows about the prices,
and reacts to them. To realize the importance of a few seconds in computing
terms, consider the case of Lotus Capital Management LP of New York. Earlier
this year, it realized that a competitor was beating it to a trade it had
programmed by exactly 3 microseconds, day after day. The loss meant Lotus was
forfeiting about $1,000 in daily revenue on that particular trading strategy.
Subsequently, that trading strategy was discarded since firm did not have the
infrastructure to speed up the execution by 3 microseconds.
Event
Arbitrage
Event Arbitrage is very similar to
Ticker Tape Trading, except that the item of interest here is the news feed. Most
HFT traders employ a class of algorithms to deal with each possible kind of
corporate event (including earnings reports, earnings outlook, mergers and
acquisitions, and analyst rating changes), and convert news into positive or
negative trading signals. An example would be a very simple algorithm that
would read words like “profit”, “confidence”, “beats expectations”, “good
quarter” from a Reuters news flash, and would start buying the stock before
general public have a chance to even finish reading the news. The trick is to
be the one who makes the move first: to be the one who has the fastest news
feed, the fastest information extraction algorithms and the fastest execution.
Statistical Arbitrage strategies aim
to make money by exploiting statistical mispricing of securities, like
deviations in interest rate parity in forex markets. Carried out over prices of
over hundreds of securities at a time, it is possible to detect such mispricing
using extensive data mining & complex mathematical techniques. The
arbitrage strategies hinge on the possibility that assets would obey their historical
statistical relationships with each other in long run.
The Dark Side of HFT
There is another side of the story. High
Frequency Trading is in the midst of a raging debate. Consider ticker tape
trading as described above. A person who is privy to market prices before other
players is called an insider trader, but if it is only a question of few
seconds, the boundaries of law start to blur. Any firm with enough cash to buy
high-tech infrastructure & pay rents to a stock exchange can enjoy the free
lunch of being few seconds ahead of the market. HFT is, thus, accused by its
critics to be a legal form of insider trading.
Now, consider market making. HFTs are
in no obligation to provide liquidity to the markets. They do so to serve their
own profit purpose (bid-ask spreads and rebates from exchanges). However,
during periods of high volatility, these market making algorithms stop
immediately, leading to an almost instantaneous erosion of liquidity. A perfect
example of this phenomenon was Dow Jones Flash Crash on May 6, 2010, when DJIA
plunged 900 points (9%) in 5 minutes, only to recover within next 10 minutes. A
July, 2011 report by the IOSCO concluded that "the usage of HFT technology
was also clearly a contributing factor in the flash crash event of May 6,
2010." Since then, many mutual funds have moved significant portions of
their money out of US equity markets, and are considering other asset classes.
They say that the US stock markets have been reduced to computerized gambling
houses where algorithms devise microsecond-length trading strategies. All long-term
valuation of business fundamentals seems to have lost its meaning.
And it’s not just equity. In February
2010, a trading algorithm owned by Infinium Capital Management ran amok and
caused worldwide surge in oil prices by USD 1. The company currently faces
civil charges for causing a global mayhem.
Of course, advocates of HFT (read
hedge funds and investment banks) are quick to dismiss this criticism. They
point that they provide the much-needed liquidity to the market, and hence
improve efficiency of the markets. While regulators are vying to bring High
Frequency Trading into the ambit of rules, there is undoubtedly a powerful
lobby opposing this.
Figure 4: The Dow Jones Flash Crash of 2006
SEC recently passed a legislation
banning the use of naked sponsored access, which allowed firms to trade
directly on an exchange using a broker’s infrastructure without pre-trade risk
controls. Similarly, IIROC, Canada’s financial regulator, has proposed new
tariffs that would charge trading desks per message, rather than per executed
trade. If these costs are passed down by trading venues to their members, it
would have a marked impact on the execution fees paid by HFTs. What now remains
to be seen is will these regulations prove effective in tightening the actions
of HFT firms, or will the exodus of long-term investors from the US equity
markets continue unabated.
0 comments:
Post a Comment