杏吧原创

IT鈥橲 mysterious, intangible and invisible, and exists in far larger quantities than we can actually see. No, it isn鈥檛 dark matter, and it doesn鈥檛 involve galaxies. Dark liquidity is the latest way for investors to get the best out of financial markets 鈥� and the cause of a technological arms race between stock traders.

When an investor wants to buy or sell a large block of stock, their biggest concern is that other traders will notice their intention and take positions against them 鈥� changing the market price to their detriment. 鈥淭he fundamental problem is trying not to show people what you鈥檙e doing, while trying to read where everyone else is going,鈥� says Richard Balarkas, CEO of Instinet Europe, a London-based brokerage firm.

This 鈥渕arket impact鈥� is a real problem for investors such as pension fundholders, which tend to trade stock in very large quantities and often at times specified rigidly by investment policies 鈥� for example, selling stocks that have fallen out of equity indexes such as the FTSE-100. Traders can spend months slicing up large blocks of stock and discreetly selling off the pieces, just to minimise market impact, says Doyne Farmer, an expert in complex systems at the Santa Fe Institute, New Mexico, and founder of Prediction Company, a quantitative trading firm.

So big investors are always looking for liquidity 鈥� the ability to place orders without moving the market. Markets that offer 鈥渄ark鈥� liquidity cater to this need by the simple expedient of not publicly displaying the order. That means no one can exploit that information for their own benefit, but creates a bizarre situation: vendors may put large blocks of stock up for sale, but would-be buyers have no concrete way of knowing that it鈥檚 there. Both sides rely on the market operator to make sure a deal gets done (see table).

鈥淒ark liquidity is designed to be found, but not to let you know that it鈥檚 there鈥�

That might seem perversely complicated, but dark liquidity may have a critical role to play in financial markets. On an average day, $16 billion worth of stock is traded on the London Stock Exchange (LSE), less than 0.5 per cent of the total capitalisation of stocks listed on the exchange. Dark liquidity will facilitate much greater volumes of trade, argues Balarkas, provided that algorithms can be developed that allow buyers and sellers to find each other efficiently in the dark.

Dark liquidity is made available in a number of ways. Some brokerages offer 鈥渋ceberg鈥� orders, in which they break up a big order into smaller orders that are each sent to market in turn. Others have set up 鈥渄ark pools鈥�, which resemble any other stock-trading venue 鈥� except that they make a secret of the price being asked and the volume of stock being offered.

Many of these dark pools don鈥檛 advertise their existence, but it鈥檚 estimated that up to 40 have been set up. Most are based in the US, but they are starting to appear in Europe. The highest profile is Turquoise, a London-based dark pool established by nine big securities dealers as a venue for European share trading that is due to start trials next month. Its advent prompted the London Stock Exchange to announce in June that it would also be developing a pan-European dark pool whose name 鈥� Baikal 鈥� refers to the world鈥檚 deepest lake.

Developing a dark pool is far from straightforward: Turquoise has been unkindly nicknamed 鈥淭ortoise鈥� in the City 鈥� London鈥檚 financial district 鈥� because of its slow rate of development. On the one hand, dark pools have to ensure that buyers can find sellers; on the other, they must avoid giving away information about the quantities and prices of the stocks lurking in their depths. 鈥淭he name of the game is controlling information leakage,鈥� says Fred Federspiel, president of Pipeline, a New York-based company that runs a dark pool.

That鈥檚 far from easy, because there are third parties actively trying to extract that information. For buyers, trading in the dark is a bit like seeing a product for sale in a shop window, but not being able to read the price tag or guess how many others there are in the stockroom. Does the shop have lots of stock to get rid of at bargain rates, or is the product as scarce and overpriced as a popular toy at Christmas? But at least when you buy something in a shop you can be fairly confident that there aren鈥檛 any third parties actively plotting to make your purchase more expensive. That鈥檚 not the case for big investors making massive stock trades.

What鈥檚 in the water?

In an effort to find out what鈥檚 floating in the dark pool, so-called 鈥渟tatistical arbitrage鈥� firms place small orders to probe how much stock a pool conceals, and its going price. If the orders are fulfilled, they yield information that can be used to bet on likely price movements.

Statistical arbitrage firms specialise in this activity, but they鈥檙e not the only ones taking an interest: 鈥淚f the large pension funds know you are trying to get rid of a large block of stock they would probably try and do the same thing. It鈥檚 a herd mentality,鈥� says Balarkas.

To counter this sniping, investors use their own algorithmic trading systems to read the market. The idea is to work out the most effective way to allocate chunks of a block of stock to different dark pools. For example, Pipeline offers its customers a system that looks at 50 different variables across the market, such as trends in the pricing and availability of shares being traded in the public market, as well as trying to infer what鈥檚 available in other dark pools.

鈥淲e are also using techniques developed by statistical arbitrage firms, looking at what shares are sold and what price they are sold at, to infer prices under hidden orders,鈥� says Federspiel. 鈥淭here are thousands of transactions a second. It鈥檚 a significant computational effort.鈥�

This information is used to decide which of many different algorithms should be used to place orders in different pools. 鈥淭his requires switching between dramatically different tactics between bids,鈥� he says.

Of course, legitimate buyers still need to be able to find this liquidity. So a plethora of companies have started to offer 鈥渄ark aggregation鈥� services, which use their own algorithms to find out where these large blocks are hidden. This is technically challenging, says Balarkas. 鈥淚n a perverse sort of way, it鈥檚 designed to be found, but not to let you know that it鈥檚 there.鈥� After all, if you can find it, then in theory so can anyone else. The other side of the coin is that buyers need to ensure they don鈥檛 reveal what they鈥檙e looking for, for fear of pushing the price up.

Instinet鈥檚 NightHawk system uses probabilistic analysis to create what Balarkas calls 鈥渢hermal maps鈥� of where the dark stuff is. For any firm to reveal precisely how these algorithms work would amount to commercial suicide, he says. But what works in the favour of these dark aggregators is that their orders are legitimate, so they can afford to probe for larger amounts of stock; statistical arbitrage firms, who may not actually be interested in bulk buying, need to minimise their expenditures.

Another approach is to integrate visible and dark pools. This is what Turquoise will offer, allowing buyers to see small stock orders and their prices, but letting them ask for larger quantities than those publicly offered. They may just end up with the small amount of stock that鈥檚 visible, but if there are larger blocks of stock floating in the dark pool, they will walk away with a much larger chunk, potentially at a better price than they originally offered.

This integration offers customers the best possible prices, according to Yann L鈥橦uillier, Turquoise鈥檚 chief technology officer. 鈥淵ou get the best of everything in a one-stop shop,鈥� he says. But such a radical approach is bound to unsettle market supervisors. About four years ago, L鈥橦uillier was involved in an effort to create a similarly integrated pool at the Boston Stock Exchange, but the move was vetoed by the US Securities and Exchange Commission. 鈥淏ack then the SEC was afraid of the dark,鈥� he says.

There may still be cause for concern. 鈥淒ark pools are in a certain way parasitic,鈥� says Farmer, because their participants rely on public markets to provide reference prices, but their transactions don鈥檛 contribute directly to the public prices that most investors see. So public prices may not accurately represent the true value of stocks. 鈥淚f things get too far out of hand, governments would have to start to create rules prohibiting dark pool activity,鈥� says Farmer.