Page 1 of 2 Goldman Sachs, the lords of time
By Julian Delasantellis
Just for the moment, let's pretend that James Cameron's 1984 The Terminator
was being made for the first time today, and, instead of the evil robots
emerging from the fatally misguided foundries of Cyberdyne Systems, they came
from the dark laboratories of Goldman Sachs.
From out of the future, a warrior is sent back in time to warn the present.
"You still don't get it, do you? They'll find your money!! That's what they do!
That's all they do! You can't stop them! They'll wait for you! They'll reach
down into your bank account and tear its f*&^*^g balance out!"
"Now just wait one cotton-picking moment!" I hear many of you
saying. A few weeks ago, in my review of Matt Taibbi's Rolling Stone-published
anti-Goldman Sachs screed, I demurred, advancing an argument that, although
Goldman certainly proudly wielded the sharpest elbows in investment banking,
they were not the eternal center of a global nefarious conspiracy of plutocrats
that has been making the lives of average people miserable since the 1920s.
good but not that bad, Asia Times Online, June 9, 2009).
I received a number of thoughtful and well considered e-mails over my
disagreement with Taibbi, many making helpful and constructive suggestions on
the order of "Up against the wall and die, you capitalist sellout rat pig ... "
Therefore, one may consider this report on Goldman an attempt to return to the
side of the angels. It's not - well, maybe it is. It's just that, as another
chalk outline is drawn on the mean streets of international finance, once again
Goldman's prints are found on the scene.
A few weeks ago, Goldman surprised the markets yet again, reporting
second-quarter earnings at US$3.44 billion, much higher than expected, and,
following similarly impressive results from the first quarter, almost back to
the profit levels the bank, back when it was a buccaneering investment bank
rather than the supposedly more sedate commercial bank it is now, was earning
during the easy money days of 2007. Goldman's report was particularly curious
in that two of its main Wall Street competitors, Bank of America and Citigroup,
were able to report no such cornucopia.
Where were the profits coming from? Certainly not from intermediation, the
bringing together of borrowers and lenders that is supposedly the core function
of banking. There's just not enough economic activity currently going on to
justify that much profit from standard intermediation. John Hempton of the
Clusterstock blog suggests that Goldman is helping sovereign wealth fund (SWF)
purchases of US and British debt securities, but it is doubtful that the sharp
operators running the SWFs would be leaving this many crumbs on the table. 
Goldman's chief financial officer, David Viniar, attributes the outstanding
results to "basic blocking and tackling", American football cliches for hard
work at the business fundamentals. As the first 24-hour news cycle following
the release came to an end, most observers, having generated no real
alternative, were forced to accept the official story.
Then people remembered the bizarre spy caper story that ran the previous week,
a story that we'll call, for lack of a better word, Codefinger.
On July 4, US Federal Bureau of Investigation (FBI) agents arrested one Serge
Aleynikov, up until early June a well-paid Goldman Sachs vice president for
equity strategy quantitative analysis, or "quant", who worked for the previous
two years on, as his Linkedin page put it,
Lead development of a distributed real-time co-located high-frequency
trading platform. The main objective was to engineer a very low latency
(microseconds) event-driven market data processing, strategy, and order
submission engine. The system was obtaining multicast market data from Nasdaq,
Arca/NYSE, CME [Chicago Mercantile Exchange] and running trading algorithms
with low latency requirements responsive to changes in market conditions.
The FBI affidavit took the above gobbledygook and, straining it through the
cheesecloth of the law, came up with the crime. From the charging affidavit of
The defendant, unlawfully, willingly, and knowingly, without
authorization, copied ... downloaded ... communicated and conveyed, a trade
secret ... a product that is produced and placed in interstate and foreign
commerce. With the intent to convey that trade secret to the economic benefit
of someone other than the owner thereof ... ALEYNIKOV, while in New York,
copied without authorization, proprietary computer code belonging to a
financial institution in the United States and uploaded the code to a server in
A few canny observers took his Linkedin description of
what he was doing at Goldman, combined it with the affidavit's claim that he
was stealing from Goldman, put the two together, concluded that whatever
Goldman was doing with its "very low latency (microseconds) event-driven market
data processing, strategy, and order submission engine", which Aleynikov and
others call "high frequency trading", it had some value because apparently
Aleynikov was stealing it with the obvious intent to sell or give it to
A few noted that he had started his education at Moscow's Institute of
Transportation Engineering. Good God! Was he the spy going back to the cold, a
deep-cover agent in the service of Russia's FSB, the successor to the KGB?
And what did it say about how Russia compares the relative importance of
Goldman Sachs and the US government that, having this supposed well-trained and
resourceful operative, they allegedly chose to put him at Goldman instead of at
the Central Intelligence Agency or the Pentagon?
But just what was he doing? What was Goldman's secret? What was high-frequency
Most people know just enough about the workings of the US stock exchanges to
really put their fortunes at risk in it. Part of that is knowing the precise
mechanics of just how a stock trade is executed.
At least for the New York Stock Exchange, many people probably think that
trades are executed through the floor-based "specialist system". This was the
process illustrated in Oliver Stone's 1987 movie Wall Street; in it,
when a trade was to be executed, a representative from the brokerage of the
buyer or seller received a message from the brokerage, and walked the floor
until he was in the actual physical presence of a floor trader called a
"specialist", a representative of a firm that has promised the exchange that it
will "make a market", that is, buy from retail sellers and sell to retail
buyers, in the stock in question.
This they do by always maintaining and declaring a so-called "two-way" price,
at which they are willing to buy a stock from the a public that might want to
sell, and a higher sell price that they will use to sell stock to those who
want to buy. Thus, by "making a market" in the stock, by buying low and selling
high all day long, the specialist firm earns its operating profit, frequently a
very nice little operating profit.
But the specialist system was never adopted for the NASDAQ, and, for about 97%
of the trades currently on the NYSE, it doesn't work that way either; these
days, the actual floor of the stock exchange serves not much purpose other than
being television news' Potemkin village for modern-day capitalism. The real
action is happening nowhere near the floor, inside the data servers of what are
called "electronic communications networks" (ECN), like the ones operated by
the New York Stock Exchange/Euronext conglomeration, one, OMX, operated by
NASDAQ, and, equally or perhaps more important than the previous two, an
independent ECN run by a company called BATS.
You, the average investor, may be involved in the stock market because you hope
that, over 25 years or so, the shares you own of Google will appreciate so much
as to be able to fund a comfortable retirement. That's not the case at the ECNs
- they're breathing, eating, living and dying, every single hour of every
single day, the bid/ask spread.
If you hear a media report telling you the price of a stock, say, "GE up three
quarters" or "ExxonMobil up an eighth", what you are most likely hearing is
news of the most previous actual trade the stock was exchanged at. This does
not take cognizance of the bid/ask spread. When you do understand it, you will
have taken a giant step forward in your understanding of financial markets, for
playing the bid/ask spread is at the core of what real financial market
professionals call trading.
All during the American trading day, you can see on the BATS web site
(http://www.batstrading.com/home) bid/ask spreads in action in near real time A
recent example had Wells Fargo bank, WFC, being quoted at a bid ask spread of
24.60 bid, 24.61 offer; the standard jargon here would be to then quote WFC as
What this means is that some trader using BATS thinks he can make money buying
WFC at 24.60 and immediately selling 24.61. Granted, it's only a penny, but if
it's on a million shares, it's $10,000, for what would probably be only a few
seconds work, the time the trader held the stock until he could dump it.
Looked at it this way, the bid/ask spread appears in a new light; what it
actually is is the business profit margin of the trader making the two-way
market in WFC. I'm sure that if he had his way, the trader would rather be
making a lot more than a profit margin that comes out to about .04% (1 cent
divided by $25) per trade; if he could quote a spread of, say, 25.00-50.00; his
profit margin would be 100%.
But in that case, the trader would be faced with the same age-old dilemma that
all businesses with wide profit margins face - the fact that competitors can
come in and, in exchange for a less-bountiful profit margin, try to quote a
narrower spread in order to steal some of the lucrative business from the
trader quoting the big spread.
That's the big difference between trading under the old NYSE specialist regime
and trading with modern day ECNs. Specialists were granted an essential
monopoly in the business of making bid/ask spreads in their individual stocks.
With the ECNs, any qualified trader can put his own individual bid/ask spread
on the board; this competition acts to narrow the spread through arbitrage.
Then again, unlike the standard practice with the specialists, on the ECNs, the
traders putting up spreads they're willing to buy and sell stock at are under
no obligation to keep them there should prices start to move so violently that
operating a profitable (for the trader) two-way market can't be maintained.
This was the case on the NASDAQ during the crash of 1987, with prices falling
so far and so fast that traders simply stopped answering the phones of those
requesting them to make a two-way price.
This is the real essence of the trader's day-to-day life, intently watching and
altering bid/ask spreads to steal business from other traders but not being so
aggressive with a buy price that they wind up buying stock at a price above
that which they can sell it out for, what is known as a "crossed spread".
What is it that moves the price of the stock, the bid/ask? Here, it's not too
much different for a five-minute time interval than for a five-year interval -
volume. If, at any one moment, the combined market makers of an ECN have, say,
five million shares of Yahoo that they're willing to sell at 15, and if
somebody comes in and buys a million of those shares, the other market makers
are going to know that Yahoo is hot, that there's buzz about it, that the train
is leaving the station. The other 4 million shares will be marked up
accordingly, maybe to 15.20-25.
Now, things are getting interesting. Let's transfer the above example from
today's brave new world of the ECNs to the old days of the specialist system.
Let's say that you're the retail trader that wants that million shares of
First, you do it the old-fashioned way. You call your broker, read the trade
off, he reads it back to you a few times, and then, only then, the order is
sent by the brokerage to its trader on the floor, who would walk it to the
For the sake of the example, let's say that, at the same time your order is
making it through the bowels of the brokerage, you have a buddy with a
cellphone standing next to the Yahoo specialist. Quite illegally, you call him,
tell him to execute the order.
He does, instantly, at the original ask price of 15. Then, your brokerage's
floor trader comes lumbering into view. He still has your trade in his hand
that he is delivering the old fashioned way, but now, the specialist, having
sold one million shares to your buddy, has raised the price to sell the next
million shares to 15.25.
Okay, the broker delivers an order to buy a million shares of Yahoo at 15.25.
Who wants to sell? Wouldn't your buddy, he just bought at 15 - selling a
million shares at 15.25 nets a quick $250,000 profit, which, he, presumably,
will share with you.
In financial lingo, speed of order delivery is called "latency", in that a
system that executes orders slowly is said to possess "high latency", while a
fast system, like your friend at the other end of your cellphone, possesses
"low latency". Didn't Serge Aleynikov's Linkedin online resume note that he
specialized "in architecture and implementation of massively concurrent
low-latency, highly available distributed systems in the areas of