Page 2 of 2 Goldman Sachs, the lords of time
By Julian Delasantellis
Slowly, the backstage curtain opens to reveal Goldman's new trick.
"But you can't outrace a computer-based ECN network the same way you can human
specialists, can you?" you might ask. To paraphrase George Bernard Shaw in a
quote made famous by senator Robert F Kennedy, Goldman Sachs dreamt things that
never were, and said "Why not?".
According to press reports, this is how one of Serge and Goldman's "trading
algorithms with low latency requirements responsive to changes in market
conditions" might have worked.
Let's say that you're a big-hitting retail investor, such as the
California state pension fund Calpers, or a big Fidelity mutual fund. Today,
you've got a big hankering to buy a lot of stock, say, 10 million shares of the
QQQQ Nasdaq 100 exchange traded fund (ETF).
This ETF trades an average of about 140 million shares a day, so someone
wanting 10 million at a pop is not all that unlikely. You've got the credit for
the trade, but there's one big problem. If your broker "shows" the market an
order this big, there's just no way that the other market makers on BATS or
another ECN are not going to mark up the price the shares are offered at,
causing you to pay more than you want to on your trade.
What to do? Well, one way to try to fool the market would be to split up your
order among lots of different traders making a market in QQQQ, say, 100 orders
of 100,000 shares each.
The hope here is that the trades will go through underneath the trader's radar.
That's entirely possible; human traders can only process so much information
even when not distracted - it might just work.
It might just work with human traders - not with Serge's algorithms run on
Goldman's computers.
Apparently, what Goldman's computers can do is incredibly sophisticated pattern
recognition that can tell when somebody is trying to sneak in a big order. Not
only that, they've apparently backward engineered the software to recognize
what the market looks like before many types of big (within the context of
about a few seconds to up to 15 minutes of trading) price moves.
Once you know that big orders are coming, and if you have Goldman's powerful
big computers, making guaranteed easy money is a snap. While it may take a
second or two for the multitude of smaller orders to hit the market, Goldman
has already been in the market and gone, buying up most or all of the 10
million share order. When the buy order from the retail customer finally does
hit the market, guess who's there to fill it? There's Goldman, selling the
shares it just picked up maybe a second or two earlier at a lower price for a
tidy little profit, the sum of which apparently went a long way towards making
Goldman's $3.44 billion of earnings last quarter.
Many think that these trading mechanism's are inherently unfair, since, by
their very nature, they disadvantage smaller traders in favor of the trading
power financed by Goldman's wealth. Therefore, are the exchanges and ECNs
working to stop Goldman and the others trading in this manner, so as to restore
a level playing field?
Yeah, right. As the speakeasies said during Prohibition, may I take your hat
and coat, Mr Capone?
Let's say that, over in Galaxy M87, in the constellation of Virgo, they're real
aces at writing high-frequency trading algorithms; still, they're not going to
be able to profit in this market like Goldman has. M87 is 50 million light
years from Earth; any order originating from M87's computers will take 50
million years to reach the exchange. Truly, that's very "high latency".
Here on Earth, firms like Goldman seem to have discovered that, even with
orders moving through the system at the speed of light, the distance from their
computer to the ECNs computer makes a difference. With success or failure in
getting to the market sometimes only a matter of milliseconds, it is only
natural that many of these high-frequency traders such as Goldman have
successfully worked to get their computers as physically close as possible to
the ECN's computers.
As in, like right next to them, what they call "co-location".
One final way that the exchanges support firms doing this type of trading is
not that hard to understand. How different would the result of most American
football matches be if one team had to divulge its whole playbook, including
what plays its offense was about to run, on a fairly regular basis? In exchange
for the liquidity that these high-frequency traders provide the market, the
exchanges allow them to "peek" at other traders' order books, the listing of
what trades, at what prices and at what volumes, the traders will be delivering
onto the marketplace up to a half of a second before they are executed.
For traders with powerful computers and trading algorithms like Goldman's, a
half a second is all they need; sometimes they'll get the job done if they can
see orders but a few microseconds before they're fired off.
The exchanges defend this cozy practice by saying that it helps average traders
by stopping their trades before they would be delivered to the exchange to be
filled at a bad price. Maybe that's true every once in and a while, but, for
many other observers, this smacks heavily of frontrunning, the practice where a
brokerage, seeing a big order come in from one of its customers, trades in the
order's direction in order to profit from the market move that will ensue from
it.
Frontrunning is illegal. Sneak and peeks, like most everything else Wall Street
has developed this decade and beyond, is not. Yesterday's criminal may be
today's technological avatar, but that still leaves the basic act involved a
crime, doesn't it?
With all these new games it has to play, it's no wonder that Goldman wanted
Codefinger Serge to come back to the party with his stolen trading code. If
you're one of those people who hate computers because you've fried every one
you've ever touched, Goldman Sachs has a server in Germany it wants you to
find.
Goldman didn't originate high-frequency trading, and it's not the only shop
doing it now. For years, observers have been astounded at the amount of money
Dr James Simons' Renaissance Technologies hedge fund would regularly make;
frequently, billions of dollars a year. Nobody really knew just what Simons was
doing to earn these riches, but it is now becoming accepted that what he was
doing in the early part of this decade Goldman is doing now. This year,
however, Simons is reporting what he calls a "dismal performance" for
Renaissance, as, apparently, Goldman's massive financial firepower is
arbitraging Simons' profits away .
High-frequency trading's high-priced defenders say that the practice, by hugely
expanding total volume on US stock exchanges, improves the prices that average,
retail investors get on their trades. With up to 70% of trading on US exchanges
now believed to be some sort of high-frequency trading, this is probably true.
A trade for 100 shares of Boeing that previously might have filled at $43 might
now get filled at $42.85. That's an extra $15 in your pocket.
The question is, what, if any price, does society pay for your extra 15 bucks?
One thing that high-frequency trading conclusively proves is just how much the
deck is stacked against small investors. If you're buying and holding your
stocks for retirement 30 years hence, this probably doesn't matter much; but if
you're trying to short-term or day-trade this market, from your den with your
kids' toys in the corner, over your little DSL broadband line, you're going up
against the Goldmans and their massive computer firepower. That makes about as
much sense as betting the mortgage on the prospect of your gymnasium's pick-up
basketball team of insurance agents and furniture salesmen beating the Los
Angeles Lakers. If it ever really was possible to day-trade stocks away from
the floor, it's not now; the whipsaws that high-frequency trading can generate
will murder you.
Some have wondered, with high-frequency trading now sometimes estimated to be
about at 70% of total stock market volume, what would happen if it just
suddenly disappeared? What would happen if, through technical glitches or just
some manner of malfeasance, the big computers located right next to the
exchanges' computers just stopped sending prices to the exchange?
Obviously, bid-ask spreads would widen, perhaps sharply, in its early stages,
and that might be mis-identified as a crash. What if somebody like Serge, as
the SWAT team closes in, tries to save his skin by holding the whole stock
market hostage, threatening to send in code that would permanently crash the
system if taken?
With such complex algorithms, partially or fully understood by so few if any
real people, reconstructing the system and all its busted trades could take
forever.
In his New York Times column, Nobel economics prize winner Paul Krugman,
quoting UCLA economist Jack Hirshleifer, said practices such as high-frequency
trading, while providing private profitability for Goldman and its other
practitioners, contribute nothing but "social uselessness" to society as a
whole. In this, the practice resembles much of what else has come out of warped
laboratories and minds, things like credit default swaps, collateralized debt
obligations and the creative interpretations of value by the ratings agencies,
of the financial system these past few years.
Something like a Tobin Tax, first proposed by Nobel Prize-winning Yale
economist James Tobin, a worldwide levy on short-term financial speculation,
would probably stop high-frequency trading dead in its tracks, but if President
Barack Obama can be charged with socialism just for trying to revive the
economy in precisely the same way his predecessors have, he'd probably be
equated with Robespierre if he ever proposed a Tobin Tax.
What strikes me most about high-frequency trading is the opportunity costs,
what society loses by deploying scarce resources in the fashion it does instead
of towards more productive uses.
Forty years ago last month, I sat with my father in front of our fuzzy
black-and-white TV to watch men walk on the moon. The young president John F
Kennedy had tasked all the nation's considerable technological and scientific
resources to accomplish this one goal before the decade's end, and although he
would not live to see it, and even after the tragic 1967 fire on board Apollo 1
that was thought to put the goal out of reach, in 1969 it happened, to the
astonishment and delight of the entire human race.
Fast forward 40 years. Next year, the Space Shuttle will be retired (if NASA
had any sense, it would be delivered right now to a car dealer for a government
$4,500 cash-for-clunkers credit), and, for the first time in nearly 50 years,
America will have no capability to send a human into space. At present, the
biggest issue in US space policy is the substandard toilets America has
delivered to the International Space Station. They apparently were not
engineered with sufficient robustness of capacity to handle the Russian
cosmonauts' high-roughage diet, currently resulting in frequently the most dire
conditions imaginable up there in space.
The contrast between NASA's crappy space toilets and Goldman's Midas-capable
computers perfectly illustrates the public policy choices America has made
these past few years. Private consumption and investment that benefits only the
few is considered glorious, while public consumption, from small communities
cutting math and science instruction so they can fully fund the school football
team, all the way to paltry national support for basic scientific research, is
forever the hungry orphan with an empty plate always begging for more.
To make the metaphor even more glaring, they actually call the PhDs who come up
with ideas like high-frequency trading "rocket scientists", deserving of
respect and good pay just as long as they stay far, far away from any real
rockets.
Sarah Connor terminated the Terminator in that first movie, but he came back
for two more feature films, a TV series, and another feature film is about to
be released. Just like the Terminators, Goldman Sachs never, ever stops.
Note: 1. This rewrites the original summary of Hempton's blog.
Julian Delasantellis is a management consultant, private investor and
educator in international business in the US state of Washington. He can be
reached at juliandelasantellis@yahoo.com.
(Copyright 2009 Asia Times Online (Holdings) Ltd. All rights reserved. Please
contact us about
sales, syndication and
republishing.)
Head
Office: Unit B, 16/F, Li Dong Building, No. 9 Li Yuen Street East,
Central, Hong Kong Thailand Bureau:
11/13 Petchkasem Road, Hua Hin, Prachuab Kirikhan, Thailand 77110
