In a series of fascinating articles all weekend, covering the NYTimes article about low-latency/high-frequency electronic trading tactics, there has been a healthy debate about whether such trading tactics were the force behind Goldman Sachs astronomical trading profits. To brush up on the issue, check out these links:
This story has been further complicated by the recent news story surrounding Sergey Aleynikov, an ex-employee of Goldman accused of stealing code for use in Goldman's extremely fast trading system. Some of the articles above even build the correlation between Sergey's theft and Goldman's trading profits - link
And in case you were concerned that this was not entirely a full three ring circus, Senator Charles Schumer has recently decided to throw his hat in the ring - link. Now we're ready for a party...
Understanding High-Speed Trading
The articles each try to articulate the act of 'high speed' or 'low-latency' trading, but we'll try to simplify it. In essence, the strategy involves a combination of a series of passive and active trades, used to guage and profit from investor limit order strike prices.
On the passive side, investors who send trades through a broker, such as Goldman Sachs, along with a specific strike price, will likely find several micro-trades occuring before their trade is executed, to bring the PPS closer to the strike price. The result is profit for Goldman's trading system, since it has front-loaded the trades it already has in it's backlog, with micro trades that will generate guaranteed profits. Now, in highly liquid trading volume, the margins are razor sharp (recall that on the NYSE, the specialist firms used to do this, and thus fought decimilization with tooth and nail, to try to avoid getting "penny pinched"). Today's bots would certainly run circles around those Spear Leads specialists, leaving a tiny fraction of the margins they used to collect on the table.
On the active side, the claims of these articles, and of Mr. Schumer, is that several firms actually actively seek out market demand strike prices by issuing micro dummy trades at various price points. The result is a poor price for the trader, since the bot makes the market that much more efficient for the seller (at least in terms of pure economic theory). In essence, the trader buys or sells at the predefined strike price, since the system finds it through quick micro-orders that are executed or cancelled immediately after the key information about the other side's willingness to pay is revealed.
Implications of High Speed Trading
There is no doubt that high speed trading is a questionable practice, despite being in line with traditional broker and specialist businesses. In today's market, the value of high speed trading is the manipulation of the anonomous nature of the trading platform. In essence, the system collects strike prices, and moves markets to buy or sell at a price as close to that as possible. This manipulation is a perfection of the historical specialist/market maker margins, which were never optimized or automated to this extent.
So what do we do now? We can attempt to cut out this kind of trading, making it illegal to send quick buy-sell orders for price discovery. We can require that traded shares be held for a minimum period of time. However, this will inherently cause harm of unknown proportion, and likely hurt firms who have strategies that do not cause these damages to begin with. Sometimes there is a value in letting an inefficiency such as this be countered by equal and opposite market forces, such as more sophisticated trading anonymity controls (maybe the market order will become a reasonable alternative for 'best price' execution).
Additionally, as several authors have insinuated, this market is probably not that large, and likely not the bulk of Goldman Sach's trading profits. As nearly everyone pointed out in their counter points, no one truly knows what Goldman is trading and how its having such success, particularly in these crummy markets. Every indication, for at least the 5 years that I have followed the company, leads me to believe that they actually have decent risk management, and understanding of the 'insurance products' offered on their exposures. It's quite possible that Goldman has achieved what we all thought Wall Street had achieved over the last 5-10 years - virtually indestructable risk management. Ultimately, we just don't really know.
If you've got a free moment, browse the stories above in greater detail, and post some comments. Would love to open up the discussion deeper to others.