Nasdaq to Offer Algorithms to Members, Competing With Brokers
Nasdaq Stock Market (NDAQ) plans to compete with brokers by offering tactics for executing larger stock orders as a way to boost its share of U.S. equities trading.
The exchange will provide three algorithms, or computer-driven strategies that chop larger orders into smaller pieces based on a stock’s volume and trading characteristics, in a bid to increase its market share by between 3 percent and 7 percent, said Eric Noll, executive vice president for transaction services at Nasdaq OMX Group Inc. The venture requires approval from the Securities and Exchange Commission.
The provision of algorithms, a service most brokers offer to institutional clients, is part of a plan to bring more equity orders to the New York-based company’s three exchanges. Nasdaq OMX will also try to boost business by changing the way orders are executed on one of its smaller venues and will begin a program to entice retail customers to its market, Noll said at a May 10 conference. The algorithm effort will target Nasdaq’s brokerage members, many of which may no longer need to build the “commoditized” strategies the exchange will supply, he said.
“It’s a very interesting example of how technology brings exchanges and brokerage firms into competition,” Bruce Weber, dean of the Alfred Lerner College of Business and Economics at the University of Delaware, said in a telephone interview. “Before electronic trading really took off, it was clear where the exchange function ended and the brokerage function began. That line is getting blurred.”
Nasdaq OMX’s share of U.S. equities was 21.2 percent in the first quarter while NYSE Euronext’s was 24 percent, data compiled by Barclays Plc showed. The pursuit of new trading business is occurring against declining average daily volume, with this year’s level at 6.75 billion shares on all U.S. exchanges, compared with 7.69 billion in the same period last year, according to data compiled by Bloomberg.
Brokers provide clients with algorithms that send smaller pieces of a bigger buy or sell request to exchanges or alternative venues such as dark pools, private markets that don’t display bids and offers. Traders at securities firms and asset managers use the automated strategies to mute price impact and mask activity as they try to scoop up visible and hidden orders spread across exchanges and about 40 other markets.
Nasdaq will furnish what it calls benchmark orders to member firms seeking to achieve an average price for their stock sales or purchases in line with one of several common measures of trading performance. One algorithm will employ a volume-weighted average price, or VWAP, strategy that seeks to buy or sell stock over a certain period, weighted for the number of shares traded at different levels, Nasdaq said in an SEC filing.
Nasdaq also plans to offer a time-weighted average price algorithm and another that tries to trade a specified percentage of share volume over a predetermined period. Such algorithms are standard offerings at brokers such as Credit Suisse Group AG, Goldman Sachs Group Inc., Citigroup Inc. and Morgan Stanley.
The company won’t market the services to mutual funds and other asset managers, Noll said. The offering also won’t displace the more complex strategies brokers develop for themselves and their customers, he said.
“What this is intended to do is provide order functionality to our broker-dealer base that allows us to provide a cheaper, better, faster product than having them all develop their own algo offerings and go out into the marketplace and compete for the low-end commodity business,” Noll said at the conference. “This is not intended to compete with the high-end, high-touch, highly sophisticated algo providers.”
U.S. asset managers used algorithms to trade 19 percent of the dollar value of their U.S. equity orders in the year ending mid-February 2011, according to Greenwich Associates. Institutions paid their brokers an average commission rate for algorithms of 1.3 cents per share, the Stamford, Connecticut-based research firm said in a June 2011 report.
Nasdaq hasn’t decided how much to charge, it said in its filing to the SEC.
The benchmark orders will generate smaller buy or sell requests to compile executions over a period of time such as an hour or day. These so-called child orders will be traded on Nasdaq or sent to other markets, the company told the SEC in its filing. The orders will be processed the same way other buy or sell requests that enter the exchange’s systems are handled.
The exchange will license technology for the benchmark orders from a company it didn’t name in the filing. While no information about the identity of member firms or their customers employing the algorithms will be shared with the vendor, that company will furnish a scorecard with analytics about the performance of the orders when the trade is completed, Nasdaq told the SEC.
The algorithms will offer what Nasdaq called an “exchange-based alternative” to tactics that seek transactions in public markets as well as dark pools, the exchange told the SEC. Noll said the automated strategies, which may augment the exchange’s share of trading, will be introduced by the end of June, pending SEC approval.
“The benchmark order will permit members to achieve on an exchange via a single order type what previously has required access to multiple venues using multiple order types,” Nasdaq wrote in the filing. The orders will provide cost savings to members and allow them “to manage more order flow at a single trading platform,” the company said.
Nasdaq is going too far afield with its proposal, according to Robert Almgren, former head of quantitative strategies at Bank of America Corp. Almgren, who has written academic papers about equity algorithms, is co-founder and head of research at Quantitative Brokers LLC in New York, which builds execution strategies for interest rate futures.
“Exchanges should focus on efficient executions and shouldn’t act like a broker-dealer,” Almgren said in a phone interview. “A broker-dealer applies intelligence to getting the order done beyond the simple order matching, which is what you get from an exchange. There are a lot of subtle decisions and I would hate to see it favor one exchange.”
The computer models that drive algorithms make decisions based on so-called anticipated order fill rates, or the percentage of trading submissions that are completed on different exchanges or dark pools, Almgren said. They must also select the price level at which they’re willing to supply bids or offers to a particular venue and when to cross the bid-ask spread to grab liquidity, he said. Almgren added that these decisions should take into account all markets and not focus on one to get the best overall execution.
Nasdaq’s initiative may appeal to customers that aren’t comfortable entering all their trading interest into brokers’ algorithmic engines “because they’re concerned there’s some sniffing going on,” the University of Delaware’s Weber said. The worry is that securities firms could use data about customer decisions to their own advantage, he said.
The proposal may also benefit users since the algorithms could be run close to the exchange’s computers that match buy and sell requests and “could probably deliver good latency performance,” Weber said. Brokers and traders seek fast executions to capture orders at the best available bid or offer price before they disappear or are traded against by others. Requests that don’t execute on one venue are usually forwarded to other markets based on the available orders and fill rates.
Weber said algorithms should seek the best execution and be “agnostic” about the venue to which they route orders.
“If they truly create a level playing field for non-Nasdaq venues, then it’s a pretty compelling proposition for the customer,” Weber said. “If they don’t, customers will ask a lot of questions about when they do outbound routing and when they will keep orders hoping for an internal cross.”
To contact the reporter on this story: Nina Mehta in New York at email@example.com.
To contact the editor responsible for this story: Nick Baker at firstname.lastname@example.org.