Retail brokerages, charged with identifying the best trading venues to route their orders, are turning to new execution analysis technology for guidance. The trend could spell an end to the practice of basing routing decisions on the much-maligned "dash-5" reports.
Penson Financial Services and JP Morgan Invest/BrownCo. are among a number of order-sending shops using the market monitoring and trade quality analysis technology of Austin, Texas-based vendor Quantum 5.
"If your order was filled on Archipelago," says Ben Brown, Quantum 5's chief scientist, "we will tell you if, on average, for like orders, INET would've been a better venue."
Quantum 5's technology, originally designed for the telecommunications industry, performs two tasks. First, it judges the quality of the order-senders' fills in light of the liquidity conditions existing when the orders were filled. Second, it assigns a score to an execution or group of executions that can be compared against an average as well against other individual scores. Those scores may represent orders sent to competing execution venues.
Quantum 5's offering is a response to the Securities and Exchange Commission's four-year old effort to bring transparency to and promote competition in the retail order execution business.
In 2001, the SEC promulgated two rules designed to shed light on the murky world of retail order executions: Rule 11Ac1-5 and Rule 11Ac1-6.
Best Execution Clarification
Rule 11Ac1-5, the more important of the two, requires market centers market makers, exchanges and ECNs to publish monthly certain execution quality statistics. These metrics would result, the SEC hoped, in some sort of mathematical definition of "best execution," a heretofore nebulous concept enshrined in 1930s-era SEC laws.
Rule 11Ac1-6 requires order-sending firms to publish quarterly the names of the market centers executing their orders and the percentages of their flow each venue received. Both rules contend strictly with market and marketable limit orders of fewer than 10,000 shares.
The Rule 11Ac1-5 stats, the dash-5 stats, are not universally liked, but they have impacted the business of retail order execution. Those market centers with better stats have seen an uptick in their business. Those market centers with inferior numbers have suffered.
The stats have their critics. "Dash-5 statistics don't measure best execution," Matt Lavicka, a senior trading official at Goldman Sachs, said at an industry gathering earlier this year.
The dash-5 stats include such metrics as effective spread', which compares the execution price to the market's best quote at the time of the fill; speed; and amount of price improvement an order received.
Indeed, the numbers provided by the trading venues may not even be accurate. The NASD recently fined Instinet and its INET ECN division $700,000 for "publication of inaccurate [dash-5] reports" and "related supervisory violations." Only weeks earlier, at the stock trading industry's annual conference in Boca Raton, NASD market regulation chief compliance official Richard Wallace promised an investigation into firms' compliance with Rule 11Ac1-5.
Brown, the architect of Quantum 5's market surveillance engine, claims the dash-5 methodology has two major defects. First, it doesn't take into account the complete array of order types. It includes only market and marketable limit orders.
"You're not monitoring stop orders," Brown says. "You're not monitoring stop-limit orders. You're not monitoring good-til-cancel orders. You're not monitoring orders over 10,000 shares. You're excluding half of your order flow." Quantum 5's technology takes into account all of a firm's orders, the exec claims.
Second, the dash-5 methodology only takes into account a market's best bid or offer when determining the quality of a fill. It does not consider those quotes away from the inside market. Those quotes may be crucial, Brown says, if the size of the order exceeds that displayed at the NBBO.
In the Nasdaq marketplace, Quantum 5's system takes into account every publicly displayed quote. The vendor's methodology got a boost recently when it concluded a deal with Nasdaq for its TotalView data feed. TotalView gives subscribers a look at every quote in the Nasdaq MarketCenter. Quantum 5 combines those quotes with those of the ECNs to produce a comprehensive view of the market for Nasdaq names.
In the listed marketplace, however, Quantum 5 has less data. Its technology takes in all the exchanges' best quotes, but not every quote. That is still better than just the single NBBO though, Brown contends.
Waiting for a Bargain
The vendor is in negotiations with the New York Stock Exchange to take its OpenBook data feed. That would give it every quote on the Big Board.
"We've had our eye on that for quite awhile," said Brown. "We are waiting for them to come down on price."
Those snapshots of market conditions, plus a customer's execution data, allow Quantum 5 to produce an execution quality score, or Trade Performance Index (TPI), for every execution or group of executions. Those scores can then be compared against a baseline average score as well as against other market center-specific scores.
To illustrate, assume, on a given day, a Quantum 5 customer sent 50 5,000-share market buy orders in Apple Computer to Knight Capital Group. Quantum 5's technology would assign an execution quality score to those 50 fills.
That score could then be compared against an overall average score (normalized to 500) for like orders as well as scores for like orders sent to other trading venues. If the customer's score was 700 with Knight, he did better than average. If it was 300 with Knight, he did worse. If the score for like orders sent to Goldman Sachs, for instance, was 600, the customer did well by routing to Knight. If the score for like orders sent to Goldman Sachs was 800, the customer may want to reconsider his routing arrangement.
"The SEC says you must comply with best execution," Brown points out, "but they don't define it for you. Our Trade Performance Index defines best execution." Not every order-sending firm is a Quantum 5 customer, of course. Therefore, the Quantum 5 database does not contain every execution that occurred in the marketplace on any given day.
That does not mean the TPI is ineffective, Brown maintains. As long as the number of fills in the database represents a credible statistical sample, the TPI is valid, according to the exec.
"To get a satisfactory baseline," Brown explains, "you only need 5 percent to 8 percent of the actual market. We are around there to be statistically accurate." The baseline TPI does incorporate a margin of error, Brown adds.
In deriving the TPI scores, Quantum 5 takes into account market conditions from the time the order was received by the market center to the time it was filled. And while it takes into account the actual share amounts available at the various price points, it allows users to configure the software to reflect business realities.
Users, for example, may want to factor in minimum share amounts available at the NBBO that exceed the number of shares that are actually available.
Many wholesalers will guarantee fills at the NBBO that are larger than the actual amount of posted liquidity. The baseline TPI is derived from all trades done over the course of a full day's trading. That means that the score for a trade done at 10:00 a.m. is consolidated with one for a trade done at 3:00 p.m. Scores are made available to Quantum 5's customers the next day.
One-day delivery is another advantage Quantum 5 has over the status quo, according to Rod Taylor, Quantum 5's vice president of business development. Dash-5 data is typically delivered only monthly by market centers and third-party providers such as the Transaction Auditing Group and Market Systems.
Quantum 5's stated goal is to present customers with their scores every five minutes during the trading day. Data would be based on a rolling 24-hour period.
Traders and compliance execs would then be able to see how the market centers were performing on a near real-time basis.
Order-sending shops may be embracing Quantum 5's technology, but market makers are not, according to Brown. They have resisted Quantum 5's overtures, preferring to view them as enemies or cops. That's despite the vendor's entreaties that its TPI scores would make worthwhile advertising fodder.
"If a market maker is at 900," Brown says, "he can say he is beating most of his competitors. I am the best.' That is what market makers need."
