Yesterday, we looked at whether GS was front-running, well, everyone, via a sniffer program that saw all trades on the NYSE prior to their execution. Theoretically, this would allow GS to buy (or sell) stocks, selling (or covering) them back to the now compromised trader towards the end of their purchase (sale). Or, they could take a position, assuming there was more flow behind the initial order. Or, they could arbitrage a few fractional cents each trade.

The idea is fascinating. If the allegations are true — and I have no idea if they are — there are all sorts of fascinating repercussions.

Consider whether any of these are realistic, or even possible:

•  The Goldman Sachs vig — a tax really — accompanies any trade large enough to catch their software’s attention: Meet the algo parameters, pay a GS tax;

• Why hasn’t the NYSE caught this? Is this another nail in the coffin of SRO (self-regulating organizations) ?

• Alternatively, if they saw it, why haven’t they done anything about it? Its not like the NYSE is run by Goldman alums (oh, wait . . . )

• What corporate or regulator can challenge GS? Congress? SEC? Federal Reserve?  How powerful is Goldman?  Is anyone, aside from Matt Taibbi, willing to take them on ?

• Perhaps Goldman Sachs is less smart as a group than previously believed — their returns are a function not of genius, but of cheating.

Now for a does of reality.

How might this work? The GS sniffer sees an order for 1 million shares. The computers pick up a 100,000 shares and based upon conditions, put them out for sale at some price higher. At 10 cents, its $10,000. In order for this to amount to any real amount of money, it would have to happen 1,000s of times a day. If GS’ program had 10% of daily volume of a billion shares, picking up a dime, its only $10 million per day.

One the other, hand, in what has been described by the pros as one of the most challenging trading environments in history, Goldman Sachs Trading Revenue is set to make all time records.

Any thoughts on this? How likely is a Goldman Sachs cheater program?


Is Goldman Stealing $100 Million per Trading Day? (July 9, 2009)

Goldman Sachs Trading Revenue May Beat Record
Christine Harper
Bloomberg, July 9 2009

Goldman Sachs’s $100 Million Trading Days Hit Record
Christine Harper
Bloomber, May 6

Category: Markets, Trading

Please use the comments to demonstrate your own ignorance, unfamiliarity with empirical data and lack of respect for scientific knowledge. Be sure to create straw men and argue against things I have neither said nor implied. If you could repeat previously discredited memes or steer the conversation into irrelevant, off topic discussions, it would be appreciated. Lastly, kindly forgo all civility in your discourse . . . you are, after all, anonymous.

113 Responses to “The Goldman Sachs Tax”

  1. Christopher says:

    A snippet on GS….

    “”In the financial realm, they are smartest guys around,” said Bill Hackney, chief investment officer of Atlanta Capital Management Co, which owns Goldman shares. ”

    Didn’t Bernie Madoff use to be called the “smartest guy around”??

  2. Maneki-neko says:

    From airtrader: “A similar tactic was used by a programmer long ago to put the interest payment rounding errors on savings accounts normally kept by the bank into his personal bank account.”

    You might’ve seen it in a movie, but it did actually happen at a British bank many moons ago….

  3. Peter Peter says:

    I am a high frequency self employed alg. trader with several computer science degrees and a long career in the IT space.

    GS eats my lunch every day, and I am left to pickup the bread crumbs that they leave behind. I also have been called a tinfoil hat guy more times than I can recall…. and I totally dismiss the notion that GS is doing anything wrong.

    Anyone with money can get co-location at an exchange. If you want it bad enough, you just pay to play. If you want access to a broker dealer with very fast servers co-located with the exchanges, all you really need is an account of $100K. So… the notion that GS is in a special class from a network access standpoint is sadly not the case.

    As for the theory that GS was putting their ethernet into promiscuous mode to listen for other orders being sent to exchanges/ECNs before they posted in that exchanges book, that seems highly highly unlikely.

    In the 80s, ethernet was a cable, much like the coax running to your TV. Any data sent on the cable was readable by anyone else – if they set their interface into promiscuous mode (i.e. sniff the data on the wire regardless of who it is intended for). In the 90s, everyone ditched ethernet coax for UTP (unshielded twisted pair) – which are the ethernet cables with those RJ45 connectors (looks like a phone cable but wider). With the UTP cables, you connect a computer to either a hub or switch. A hub mimics the old coax, and does allow everyone plugged into the same hub to see all of the data on that hub – however a switch learns by reading the packets who the source and destination are, figures out which port in the back of the device each packet is destined for, and it sends it on its way – only to the destination device.

    So, if any exchange is using a hub for the servers and plugs in anyone else to the hub – then there is a problem in that GS and anyone else with that level of access could sniff data. But… it is kind of crazy to think that anyone is using a hub. In a hub, the bandwidth is all shared, so that if you are running at 1Gb/sec, that is the total that each and every connection on the hub can theoretically send (and in practice, much less). With a switch, there is much more bandwidth available. And, I would bet the farm that every exchange and ECN uses switches (crap – you can barely even find hubs anymore, I just looked on Amazon for fun).

    With the administrative passwords to some switches, it is possible to put an interface into a mode where it could listen to all traffic on that switch – but:

    1) You first need the password
    2) Stolen code in no way gets you #1

    They are not front running at the exchanges – they are just better than anyone else in responding to the same data any of us could get if we were willing to pay up for the direct connection.

  4. Craig says:

    There is no question that the market suffers from some manipulation. For kicks you could look at the trading on the S&P mini (ES contract) that occurs on options expiration week. At the end of the week, no matter the volume, the market is “pinned” to a certain point, narrowing as the day progresses. If orders come in pushing the price up, zillions of orders then come into counter act that and drive it down. vice versa, if it goes too low, immediately orders show up to bring the price back up.

    I suspect what GS(or someone) does is pick a point where they think they can drive it to, and purchase enough puts and calls that are out of the money, and then drive the market there gradually, in the last week or so. Certainly it’s something a computer could control.

  5. dss says:

    Given the interchangeable roles in the investment banks, the Fed, the government, where the banksters go from the private to the public sector back to the private sector, and there is no real oversight (SEC has been gutted) the corruption is real and just now seeing more light of day since the financial collapse.

    It isn’t enough that the privileged few get to survive, they also get to steal all that they can through any method they can. The coverage on ZeroHedge has been outstanding and in trying to connect the dots they have exposed just a fraction of the shenanigans of the banksters. And the chosen few also get bonuses as well.

    Do I think that GS, who is the annointed bank of the government, uses programs to defraud investors? Does a bear shit in the woods? They have been picking our pockets for generations only now they are more sophisticated about doing so, and rarely have they had the cooperation and blessing of the government to do so.

  6. alfred e says:

    @BR: Congratulations. You are smoking out lots of contributors that just would not show up in the MSM.

    True or false there’s lots of info, dialog and sharing.

    Seems pretty obvious this is not the market most thought/think it is/was.

    Transparency. Yeah right. In your dreams and up your ….. Just re-election words.

    My suspicion is there’s more to learn. But at least we’re someplace we would not have been without blogs like this one and others.

    MSM is more like TASS every day.

  7. danm says:

    They are not front running at the exchanges – they are just better than anyone else in responding to the same data any of us could get if we were willing to pay up for the direct connection.
    Well first of all they have a pretty good idea of what institutional accounts have in their portfolios and they also have a pretty good idea of what these portfolio managers can decide to liquidate or are thinking or buying. You think they don’t arbitrage this info?

    They get a much better market picture than the typical portfolio manager does. And if you try to convince me that they don’t use that info, I won’t buy it. It’s so totally against human nature.

  8. How do you beat these computers? well, when you want to buy IBM at $85, put in a limit order and go play golf. Don’t watch the stock move – don’t watch the computers try to out-pysche you, causing you to panic and pay up. If you don’t change your limit price, they can’t beat you.

    Right price plus and adequate amount of time factored in will eventually reduce all scalping profits to zero.

    Now I will tell you the law of the jungle:

    There are all different types of beasts in the trading jungle. You have elephants(pension, mutual funds), you have water buffalo(retail traders) and you have cheetahs(high speed computer firms like GS). When the food sources grow thin and the big eaters(elephants) get hungry, they migrate. The cheetahs, being quick, can dart in and out from the herd stealing scraps of food. If they try to stop an elephant on his way to his next feeding ground through battle, they will get stepped on over the long run. The cheetah can also beat the water buffalo over the short run but if the water buffalo understands the cheetah and itself, the cheetah cannot take anything from it over the long run.

    So we have defined how the cheetah and the elephant feed. How does the water buffalo feed? By ‘knowing thyself’. Many water buffaloes go into the jungle thinking they are cheetahs and they try to live like cheetahs and steal lunch like cheetahs. Unfortunately, if they are dumb, water buffaloes eventually starve due to all the real and quicker cheetahs in the jungle. That is the first stage of a water buffaloes maturity. They learn they are not a cheetah

    The next stage is learning they are not an elephant. Many try to be elephants. They try to do the work of forging ahead through the jungle in a search for greener pastures(forcing prices) and on the short run it works…until a real elephant comes along and sees a free meal of unnatural prices. Those are quickly taken away. Even if there is a group of water buffaloes forging ahead the elephant can just call out to its mates and by their sheer brute strength they can push any number of water buffaloes any which way they choose. The water buffaloes can win for a long time. There is the old saying that a herd of water buffalo can remain foolish longer than an individual water buffalo can keep from starving. Refer back to the internet bubble for that.

    Eventually though the law if the jungle will reexert its natural order and the water buffaloes will lose their strength pushing through to greener pastures.

    That is the second level of maturity that water buffaloes have to pass through to be able to live in the jungle. They then become full grown mature water buffaloes when they finally learn the rules of the jungle for the benefit and feeding of the water buffalo. Here they are:

    1. You never try to fight a cheetah for his lunch. You’ll only make him laugh

    2. You eat the same type of food as an elephant. Not a cheetah

    3. When elephants are out of food don’t try to stop them from finding more food. They will step on you and eat your lunch

    4. The best and most cost effective way for a water buffalo to eat is follow closely behind an elephant. They make pathways through the jungle a whole herd of water buffaloes can pass with ease. So they will ‘pave the way’ to greener pastures. Plus they will be the ones ‘feeding and doing battle with the cheetahs’.

    5. Know that you are more agile and quieter than an elephant and you can avoid being stepped on by them in their hunger when they want to go a certain direction. So you will more likely avoid detection by all the cheetahs

    So the moral of the story is:

    As a water buffalo it is your job to watch the elephants and learn their desires and herd movements in order to get your free meal. Don’t not try to be or fight a cheetah because the real cheetahs will eat your lunch

  9. [...] we know that Goldman-Sachs hasn’t been using this code all along to manipulate the markets?  The theory goes something like this: Goldman-Sachs gets to put a unix server with the code on a network cable somewhere and uses a [...]

  10. bergsten says:

    @Peter squared

    Not that I have any particular credibility around here (let me put it this way, I have INFINITELY more claim to “inventing the Internet” than Al Gore does), but for whatever it’s worth, Peter2 is absolutely right. If anything, he’s understating the technical situation.

    The exchanges will provide you rack space, power, and one or more (for redundency) ports of up to 10 Gbs (think of your best office network speed and multiply by 100) to very high end, expensive and sophisticated network switches (probably from Cisco), switches way bigger than the things they are attached to.

    The switch routing is completely programmable, right down to the particular message origin/destination, type…, so “following the physical wires” doesn’t prove anything. Aside from normal routing, messages can be broadcast to multiple locations, logged, disgarded, and so forth. These capibilities are used for network debugging, performance monitoring, error detection, and to support (hopefully) legal wiretapping.

    Switches can be hacked, and network administrators can be bribed. If I were writing a Tom Clancy novel (does HE even do this anymore?) I’d have GS power Cisco into putting backdoors right into the switch internal software. Heck, you could put the trading algorithms right into the switch software and none would be the wiser.

    One would fervently hope that the exchanges and those watching the exchanges and those watching the watchers prevent this sort of thing.

    If not, it’s time to get out some folding tables, and start trading stocks amongst ourselves down in Battery Park. No algos, no computers, no brokers, no CNBC, no sales people, no market makers.

    Hmmmm…. I kinda like it!

  11. [...] – “Incredibly Shrinking Liquidity” as Goldman Flushed Quant Trading (Daily Kos) – The Goldman Sachs Tax (The Big [...]