SEC’s Plan for Real-Time Dark Pool Attribution Spooks Industry

Sellside and buyside firms are worried that the Securities and Exchange Commission’s proposed plan for dark pool reporting could hurt institutional investors by enabling others to game their flow. The SEC last month proposed that each dark pool identify its prints in real time.

"There’s significant reservation from the buyside about symbol-specific real-time reporting," said Dave Johnsen, head of equities business development at Goldman Sachs Electronic Trading. "It’s not welcome by long-term investors." Goldman operates the Sigma X dark pool, which is the industry’s second-largest.

The SEC in mid-November proposed several rule changes that would affect dark pools. One proposal would require dark pool trades to be attributed in real time to the dark pool where they occurred. All dark pool trades already print to the consolidated tape in real time, but they are identified simply as over-the-counter trades.

Under this proposal, all dark pools would have to identify their trades, unless those trades have a market value of at least $200,000. The SEC recommended this exception to its proposed post-trade reporting rule to avoid hurting institutions that are working big orders.

"Our clients have said that it could introduce potential inconvenience without material benefit," Johnsen noted. "These investors aren’t saying, ‘I’d prefer symbol-specific volume from each dark pool rather than pool-level information." 

In his view, certain high-frequency trading shops could benefit more from that information than traditional buyside firms, since the former tend to be more reliant on market data to fuel their strategies. "It could empower that segment of the market more than longer-term investors," Johnsen said.

Morgan Stanley also thinks real-time attribution would affect the ability of institutions to trade quietly in dark pools. "Real-time attribution to a dark pool would impact volume in dark pools," said Andrew Silverman, global co-head of electronic trading at the broker-dealer. "Long-onlies would be less likely to put orders into dark pools," because of information leakage concerns. Morgan Stanley operates two dark pools.

Silverman said he would not mind if dark pool trades are identified as TRF.D dark pool prints in real time. In his view, that would preserve intraday anonymity and give customers some knowledge about whether a trade took place in an ATS or at an upstairs desk.

A recent TABB Group report on U.S. institutional trading discussed the buyside’s view of real-time dark pool attribution of trades. The report was based on interviews with 66 institutions. Institutions, the report said, use dark pools to "to protect institutional orders from adverse price movement caused by overexposure in lit markets." Interviews with these institutions were conducted before the SEC proposed rule changes for dark pools, but the issues were already being discussed in the industry.

The report noted that 18 percent of firms thought some form of dark pool reporting would improve market transparency, although the nature and timing of that reporting would be critical.

According to the report, buyside traders think that "any trade reporting that would identify the specific dark pool should be late enough in time so as to be of little value to the fast-money players." The report noted that "post-market close reporting could offer institutional traders insight as to how liquidity is shifting in certain market names or market centers, while offering little opportunity for gamesmanship."

Kevin Cronin, global head of equity trading at mutual fund giant Invesco, thinks real-time information about where dark pool prints take place would harm institutions. Identifying the source of dark pool prints will benefit high-frequency traders, he said in a webinar sponsored by the Investment Company Institute last month. "That absolutely is not good from an institutional perspective," he said.

Whit Conary, CEO of Level ATS, a dark pool operated by a consortium of five broker-dealers, noted that the SEC’s current post-trade reporting proposal, if adopted, "might accomplish the opposite of what the regulator hopes."

"As more blocks are traded in algorithms, there is the potential for more information leakage with real-time, venue-specific reporting of smaller trades," he said. "People can see the small trades and try to figure out whether there’s an aggressive buyer or seller in a particular venue." Conary thinks the SEC should allow buyside firms to opt out of having their orders attributed to the dark pool that executed the trade.

But doing that could raise other complications. Goldman’s Johnsen noted that, since all trades have two sides, the other side might also have to opt out of having the dark pool identified in its prints.

Goldman’s Johnsen suggested another potential outcome from real-time attribution. Buyside firms, he said, might change their order routing logic and spread their liquidity among more destinations to avoid leaving footprints in particular markets that could be viewed by high-frequency firms and others. That could increase market fragmentation, he said.

An alternative to real-time attribution of dark pool prints that appears to be gathering industry support is end-of-day reporting of symbol-specific information. Morgan Stanley’s Silverman advocates this position. 

Silverman suggested that "fuller granularity" at the end of the day about where trades occurred would be preferable to real-time granularity. "Reporting by each dark pool in the aggregate or by symbol at the end of the day would be okay," he said. "But identifying the actual dark pool in real time could harm investors by signaling predatory traders who are reading the tape."

Goldman’s Johnsen said that, for symbol-specific reporting from dark pools, end-of-day or end-of-week data would be more palatable than real-time data.

Another benefit of transparency about dark pool prints is likely to bee more confidence about volumes in dark pools. This information is currently self-reported by these pools and is published at will. Dark pools would be able to build more credibility by publishing more information in a "regulatorily mandated fashion," Silverman said. Morgan Stanley has long argued for uniformity in how dark pools calculate their volumes.

Goldman has also endorsed a single reporting standard for all dark pools as well as greater transparency about volume, Johnsen said. It was the first dark pool to advocate single-side counting with normalized data. Credit Suisse has also recommended that volume be counted the same way by all dark pools.

Research firm TABB group and broker Rosenblatt Securities have published reports for the last several years tallying the volume occurring in specific dark pools. However, their counting methodologies have differed and both must rely on information given to them by dark pools about their executions, rather than getting that data from an independent source. SEC executives in the Division of Trading and Markets have acknowledged that discrepancies in how volume is counted are not good for investors.

Both NYSE Euronext and Nasdaq OMX Group announced post-trade transparency initiatives for dark pools ahead of the SEC’s proposed changes. NYSE Euronext was first out the gate, in October, with a plan to publish end-of-day data for a half-dozen dark pools, including Goldman Sachs, Barclays Capital and GETCO Execution Services. Nasdaq followed with its own plan a week later. In both cases, the trade reporting facilities the exchange operators run in conjunction with the Financial Industry Regulatory Authority would publish information about dark pool executions at the end of the day.

The SEC’s dark pool attribution proposal includes an exception for the disclosure of a dark pool’s identity for trades with a market value of at least $200,000. Goldman’s Johnsen suggested expanding that exception for blocks to trades generated by a parent-level order of $200,000. This would allow what Goldman and Credit Suisse call "virtual blocks" to be included in the block exception for prints.

Providing an exception at the parent-order level would be "initially challenging from a reporting and infrastructure perspective, but it could be a better solution for investors than the current proposal," Goldman’s Johnsen said.

Silverman of Morgan Stanley doesn’t agree. In his view, it would be too easy for firms to include volume-weighted-average-price orders and other orders that were meant to be spread out over the course of a day in a block exception. "They weren’t intended to be blocks, so why should they get a block exception?" he said.

Silverman, however, said the SEC should reconsider its definition of a block. In the current environment where displayed markets have an average trade size of around 200 to 300 shares, he said, an argument could easily be made that a block might be 1,000 shares. He suggested that the SEC consider whether its definition of a block should be adjusted downward in terms of size.

In most cases, the SEC defines a block as an order or trade of 10,000 shares or valued at $200,000. Silverman noted that this definition was put in place "more than three decades ago, when the average trade size on exchanges was much, much larger."