An ATS loses ground to an upstart
Liquidnet is making a run at becoming the largest electronic crossing network.
The block trading system is closing in on archrival POSIT as the largest crossing network for the buyside. At the same time, it is steadily moving up the ranks of institutional brokerages.
"We've reached critical mass," says Seth Merrin, Liquidnet's founder and chief executive. "People are feeling the pain of not being in the system."
For the quarter ended June 30, Liquidnet averaged 19.8 million shares per day. That's more than double its volume of a year ago. The surge puts the upstart within spitting distance of POSIT, a service of brokerage Investment Technology Group. POSIT, said to be losing business to Liquidnet, traded an average of 22 million shares per day in the second quarter.
The growth at Liquidnet has outpaced that of the overall market in each of the past five quarters. In this year's second quarter, in fact, Liquidnet's share volume jumped 16 percent from the first quarter. That's while trading in the broader market fell.
Liquidnet's growth is making it a bigger threat to the block desks of the major brokers. In the fourth quarter of last year, the broker dealer ranked 15th in a survey as measured by the dollar value of all trades in listed securities by institutional brokerages. The survey, conducted by the Plexus Group, a unit of J.P. Morgan Chase, placed Liquidnet 20th in Nasdaq trading. About 60 percent of the shares traded by Liquidnet are listed; the balance are Nasdaq.
The survey is relevant because Liquidnet, legally an alternative trading system, positions itself as an institutional brokerage. It does not consider itself an exchange-like utility, in contrast to most ATSs.
Although Liquidnet describes itself as a community' and its users as members,' it does not permit brokers to join. Liquidnet is only open to buyside accounts. Liquidnet sees itself merely as a block desk, without the human traders. That's even though it has operational staff.
At least one bulge-bracket shop has come out with a similar anonymous trading model. Banc of America Securities launched its service this spring. Customers submit the name and quantity of the security they wish to trade. BofA then responds with a two-sided quote. If the buyside trader likes the pricing, he clicks and the trade is done.
With two key exceptions, BofA's system works pretty much like Liquidnet's. A buyside trader is notified of a potential contra to an order sitting on his order management system in the Liquidnet system. If he wants to trade, he goes active.' If the contra also wants to trade, he submits a quote. The two parties then negotiate anonymously via text chat.
The differences between the two methodologies, however, are significant. First, BofA is committing capital, thus the price will likely fall outside the spread. With Liquidnet, most trades occur within the spread, many at the mid-point. In other words, the pricing on Liquidnet is better.
Second, BofA is a broker. One of the reasons buyside traders use systems such as Liquidnet is to avoid human intermediaries and risk leaking information. Although, in BofA's case, the trader's identity and side of the market are secret, information is still imparted. With Liquidnet, naturals' – or other buyside desks – trade with naturals.
"Traders tell us a natural is the Holy Grail," says Steve Greenblatt, in charge of product management at Liquidnet.
Greenblatt does have some kind words for the BofA product. "Good for them," he says. "They are heeding what the market is screaming for."
So, what is driving the surge in trading on Liquidnet? And why now, given all the ballyhoo surrounding algorithmic trading, the antithesis of Liquidnet? Liquidnet execs offer two explanations: familiarity with the system and the "network effect."
"We're adding a record number of institutions every month," explains Merrin. "People who had been holding out are now signing up. They see the value." Liquidnet counts 271 money managers with $6.6 trillion in assets under management as members. That's up from 152 members a year ago. As more orders become available, there are more opportunities to trade. That's the network effect.
Early Adopters
Also, its early adopters are using the system more, Liquidnet says. "It takes some time for people to feel comfortable trading this way," says Merrin. "Because it is a different way of trading. You can't expect they will trade as much in the first month as they will after they've been on the system for six months."
To be fair, Liquidnet was launched just as the market started to tank. Buyside traders had fewer orders to spread around, but plenty of obligations to their traditional brokers. They had to pay for research, directed brokerage, soft dollars and other services. Last year, as the market picked up and buyside assets grew, traders had more leeway to use alternatives.
Indeed, the recent scrutiny of money managers by regulators and law enforcement has also given traders more freedom to trade where they want. The cutbacks in directed brokerage and the negative climate for soft dollars has put more control into the hands of traders, according to industry observers.
"We could not have written a better script," says Merrin of the changes whipsawing money managers on both sides of the Atlantic. "They say it's better to be lucky than smart and we've been very lucky."
One trader believes Liquidnet comes by its good fortune by default. "Order flow has been parceled down and parceled down," explains Tom Hearden, head trader at Strong Capital Management. "One of the last places where orders are getting successfully blocked is at Liquidnet."
The use of algorithmic and program trades, which can spew hundreds and thousands of small orders into the market, has exploded this year. That has resulted in fewer blocks available for brokers' block desks to cross. "The old formula is not as successful," says Hearden. "Because the other side is looking like an algorithmic order."
Orders traded by algorithms and programs tend to be small, typically in the hundreds of shares. The average execution on Liquidnet is 43,000 shares, according to the company.
One satisfied Liquidnet customer is Dennis Fox, head trader for Munder Capital Management. "It's going very well," Fox says of his Liquidnet experience. "We trade with it often. We get big prints without much effort. And certainly without any market impact or information leakage."
Fill Rate
Things could be better, Liquidnet execs admit. Traders get matches on only about one of five orders sitting in their OMSs. And of those matches, only about 20 percent result in trades. That gives Liquidnet a low four percent fill rate, not much different from those of competitor systems.
"Traders don't take advantage of every opportunity we provide them," says Merrin. "There are lots of reasons for it."
Structural inhibitors such as directed brokerage, third-party soft dollars and bundled executions have all played a part, says Merrin. Buyside desks typically divvy up their commission dollars among their traditional brokers every three, six or 12 months. The amount already earmarked can range as high as 75 percent at some shops. That leaves the buyside trader with discretion over only 25 percent of his blotter.
At the same time, some traders say they have no plans to use Liquidnet. "I have an issue with them tying into my order management system," says Jeff Albright, head trader at Waddell & Reed. "I'm not comfortable sharing every open order with them."
Cultural and organizational factors also account for traders' reluctance to use Liquidnet even if they find a match. "Three of five traders on a desk may be avid Liquidnet users," says Greenblatt. "The fourth and fifth aren't. We are still focusing on getting more people to use the system."
Traders often feel constrained by their portfolio managers, Merrin notes. Buying a large block of stock through Liquidnet may mean trading at a price far away from the day's average. So, it is possible the fill could be worse than the average.
"There is a lot of Monday morning quarterbacking going on by portfolio managers," says Merrin. "When, in fact, the trader did exactly what he should've done. There are limits being placed on traders."
Portfolio managers will also warn their traders to buy more slowly, according to Greenblatt, precluding their use of Liquidnet.
For their part, traders are also more inclined to only trade their small- and mid-cap names in Liquidnet. The Microsofts on their blotters are doled out to their primary brokers. "It's frustrating," says Greenblatt. "Very often, they'll say: I'll give that to anybody because no one is going to hurt me in trading Microsoft.'" The small- and mid-cap bias also exists in POSIT.
In addition, with algorithmic trades and programs buyside traders now have other electronic ways to deal with their large blocks.
Most bulge-bracket shops are either offering or investigating sliced-and-diced computer trades for clients. The methodology is pitched -as is Liquidnet – as a way to reduce market impact.
Market Impact
Merrin argues, though, it is a mistake to assume algorithmic trades do not cause market impact. Much algorithmic trading is done to achieve the volume-weighted average price (VWAP) of the day. But "most institutional orders are greater than a full day's trading volume," says Merrin. "To get VWAP, you don't want to be more than 10 percent of a day's volume."
Given those restrictions, it takes more than a day to get a large block done using an algorithm. That gives the market's sharks – daytraders, hedge funds and floor traders, for instance – time to profit off of a large order.
"They know they have lots of time to go in and buy ahead of [the large order]," Merrin adds, "because they know they have someone large to sell back to."
Nevertheless, the current popularity of slice-and-dice trading presents a huge challenge to Liquidnet. "Their hit rate is pretty lousy," says a trader. "And I'm not sure how high it can go. There's a lot of macro factors keeping people from trading in block fashion."
Traders at many shops – Fidelity Investment Management is perhaps the most prominent – are compensated by how close their trades match the VWAP. Therefore, they have less incentive to trade in blocks, according to the trader. "You have no interest in trading on Liquidnet, unless it is a VWAP-beating trade," he says.
Blind Crosses
Crossing systems in general get poor marks for their hit rates, a recent study of institutional trading practices found. The Tabb Group surveyed a cross-section of buyside desks and found their number one complaint about blind crosses was "not enough liquidity."
Merrin makes no bones about Liquidnet's low fill rate. "Our job is to get our members to trade more," he says. The exec argues Liquidnet's members have an "opportunity" to trade 100 million shares per day. Now, the members are only taking advantage of 20 percent of their matches. Merrin would like to see that number rise to 50 percent in the near term. "As time goes by, our expectation of everyone is to take advantage of a minimum of half of all opportunities," Merrin says.
Whether Liquidnet can boost its fill rate remains to be seen, but working in its favor is the current decline of POSIT. The industry's oldest blind crossing network is feeling its age. Shares traded per day on POSIT have dropped steadily from an average of 37 million in 2001 to 22.6 million so far this year.
Traders say the explanation is simple. Before Liquidnet came along, they would drop their orders in POSIT before turning them over to their brokers. Now, Liquidnet grabs them first. If no match is found in Liquidnet, then they go into POSIT.
"Now you've got Liquidnet standing in between," says Fox. "POSIT is used if there is no active contra in Liquidnet. After it goes through the Liquidnet filter, then it goes to POSIT."
For its part, ITG blames the drop in POSIT volume on a downturn in overall market activity. Competition from Liquidnet is not a factor, according to an ITG spokesperson, who added that POSIT is used for different reasons.
While POSIT has watched Liquidnet catching up fast, Liquidnet itself has an upstart nibbling at its heels. This rival is Harborside+, a system similar to Liquidnet, jointly owned by Thomson Financial and Jefferies Group. Harborside+ does not publish its volume data, but an executive at the firm told Traders Magazine that the system had done 4.5 million shares by noon one day in July. (Thomson is an affiliate of Thomson Media, the publisher of Traders Magazine.)
The system, which actually failed in its initial launching in 1999 and has since been re-tooled, has 230 members. That's 140 from the buyside and 90 broker dealers. The average trade size is 72,000 shares.
The Tabb Group survey lists four blind crossing systems with the most use. Half of the desks surveyed use POSIT; 40 percent use Liquidnet; four percent each use Harborside+ and Instinet's cross.
Is 20 million shares per day Liquidnet's limit? Despite its standing in the Plexus survey, 20 million is tiny when compared to volume reported by the bulge brackets to the AutEx/BlockData service.
Liquidnet boasts that its members represent $6.6 trillion, in mostly U.S. assets. But, in 2001, there were a total of $14 trillion in assets under management at U.S. mutual and pension funds, according to a study published last year by the U.K.'s International Financial Services group.
Volume Growth
That suggests Liquidnet is almost halfway there. Does that mean it has a ceiling of a relatively small 40 million shares per day? Merrin says no. Due to the network effect, volume growth in Liquidnet is not linear, Liquidnet believes. Match rates should increase as do the number of assets under management represented.
Merrin prefers to discuss the "opportunity" within Liquidnet. "I believe there is a ceiling, but I don't agree that if we double our assets, we will have 200 million shares [per day] of opportunity," he says. "I believe that opportunity will be quite large. More like 400 million or 500 million shares."