If you only knew the power of the Dark Side. - Darth Vader
The Efficient Market Hypothesis rests on the notion that all information is fully priced into the marketplace at all times. Like the human brain, the market does not need to be consciously aware of the information for this process to have its effect. This means there is no advantage to having inside information, or to running the billions of dollars worth of software and algorithms on which the world of finance revolves.
Not everyone agrees with this theory, and securities laws take a conservative approach to this academic debate and strictly prohibit trading on the basis of information not fully disseminated throughout the marketplace.
Except when they don't.
This week the Nasdaq marketplace implemented new "Flash" orders, with the SEC's approval. This was met by an immediate challenge from NYSE Euronext, pressing for legislation to impose exchange-type oversight on private order matching networks, or "Dark Pools" (TradersMagazine.com, 10 June, NYSE Euronext Asks Congress To Press The SEC On Dark Pools").
There are two kinds of flash orders. Routable Flash orders will be opaquely (read: "secretly") matched against the Nasdaq order book for half a second, then routed over Nasdaq's proprietary ITCH feed. The second type of order, the INET-Only Flash Order, appears to Instinet subscribers for half a second, with any unmatched portion of the orders being canceled back to its origin. The average investor might be forgiven for thinking this a tempest in an exceedingly small teapot.
Far from it. Orders executed in the marketplace in fractions of a second are increasingly the norm, and technology keeps up a frantic pace in an effort to continue to shorten the time of trade executions.
Algorithm-driven "high-frequency" traders run programs that route orders into the market, sometimes thousands of times a minute. Taking advantage of pricing inconsistencies, these statistical arbitrage programs buy stocks that their model identifies as underpriced, and sell them when they are overpriced.
Even highly liquid issues can become starkly illiquid in this fast-moving world. As more players, managing more billions of dollars, engage in the same trades - and with the inside market based on only a hundred shares on either side of the quote - it is not uncommon for a stat arb program to identify a stock as underpriced and, by its own automatically generated buy orders, drive the price up to the point where it triggers a sell signal.
It is a violation of market rules to enter transactions into the marketplace which create market action, but with no beneficial change of ownership. This goes back to the days when stock manipulators would create a false impression of interest in a stock by placing simultaneous buy and sell orders at different brokers. Trading volume in the stock surged noticeably. Meanwhile, the manipulator bore no price risk, as his orders would meet at the inside price, and the only cost was the commissions paid.
And hedge fund operators used to enter matched market-on-open buy and sell orders to reward brokers who had secured allocations on profitable IPOs, paying out commissions with no market exposure.
In the new electronic world, a program may buy back its own trades as the algorithm identifies the stock as underpriced, then triggers fail-safe sell orders to prevent a loss from widening. Sometimes program glitches cause this process to run in a ceaseless loop and among the tens of thousands of trades sent through in a day, it can be difficult to identify.
The exchanges have surveillance programs to identify and review these instances. Because of the functional lack of liquidity in many trades, transactions of 100 or 200 shares often lie at the heart of these inquiries. Regardless of the small trades at issue, failure on the part of a firm to have in place a compliance system "reasonably designed to detect and prevent" self-trading is grounds for regulatory action - fines, and perhaps censure if it is deemed a repeat violation.
One way to avoid these cross trades is to arrange with other market participants to meet in private to arrange trades. My algorithm says BUY, yours says SELL - let's make a deal. The new Flash Orders may become the secret meeting place of choice for high-frequency executions.
Meantime, the attraction of the established dark pools is the ability to execute large blocks without displaying orders in the marketplace. This cloaks the orders from stat arb traders whose "pinging" and "probing" algorithms identify large blocks and trade alongside them, "picking off" the larger and less liquid orders. Nimble-footed traders can short 600 shares in a fraction of the time it takes to sell a block of 400,000 shares. While these operators capture pennies by trading in front of the larger order, the rush of small sell orders disadvantages the market for the block, resulting in slower execution, worse pricing, and occasional inability to complete an order.
As the name "dark pool" indicates, this process is not intended to be transparent. The regulatory logic of the tradeoff is that cloaking these orders protects legitimate large market participants - mutual funds, pension managers and other entities managing large pools of investor capital - against predatory interference by professionals and speculators, ultimately benefitting the small investors whose interests the mutual and pension fund managers represent. Think of the large merchantmen of yore, and the swift-moving corsairs that harried them on the high seas. Wouldn't you want your ship to travel in a cloud of invisibility?
The NYSE has had its feathers ruffled by competitors for some time. Private firms also operate dark pools. Large brokerage firms trade for their own book and execute for customers alongside. Most of these operate continuous pools, giving them a perceived edge over the fixed hourly matching sessions at the NYSE.
Firms route customer orders to their own internal dark pool, while sending proprietary orders out to other firm's pools. Unmatched portions of a customer order are routinely sent, not to the exchanges or Nasdaq, but to other dark pools where the balance of the order will be matched and not displayed.
When a firm accepts an order from another firm's dark pool, it has no way of determining whether it is actually taking back the unexecuted portion of its own order. The practical outcome is that firms can never determine whether they are executing against their own orders, or for an arm's length customer.
How often does this happen? There is currently no way to determine, and firms do not want to know - because if they did, they would have to report trade violations.
The NYSE appears to understand this. One takeaway from the debate is that Flash orders are a next step in creating a multi-tiered market in which participants look down from the top to all underlying liquidity, but investors at the bottom have a very low visibility ceiling.
It should surprise no one that there is also a compelling commercial aspect to this dispute, as all these trading facilities compete with the NYSE's MatchPoint matching sessions, advertised by the NYSE as "the best environment for finding natural opaque block liquidity. Complete control of order information and execution remains in the hands of users."
The concept behind market transparency is exactly the opposite: failure to make full information available about every order unfairly disadvantages certain participants and encourages manipulation. One of the political realities is that, with so much private money being handled through large mutual funds and pension funds, it becomes a question of trade-offs to the investor. Which is worse: loss of transparency on individual trades, or being forced to show too much on large blocks where the individual's pension money or mutual fund investment is compromised?
Wherever you stand on this, the NYSE is not taking this lying down. NYSE Euronext EVP Thomas F. Callahan, testifying before the House Financial Services Capital markets Subcommittee, said "Through these so-called dark pools, alternative trading systems operators have been allowed to create private markets for securities transactions."
A "private market" that exists for half a second at a time. How big a problem might this pose?
There appears to be no control on the number of times the same order can be sent into the flash facility, and as the unmatched part is automatically canceled, orders will never be required to be sent to the open market until they are actually reported post-execution. Those of us who remember SOES are invited to speculate on what new forms of manipulation will arise here. Sellers - "customers", which was the way the SOES bandits flew beneath the radar - will be able to send out the same order until it is filled, without ever showing the broad market what they are doing. Large numbers of sellers can gang up on a stock, and we predict that someone will come up with a non-traceable way to be on both sides of a trade simultaneously.
In its rush to embrace every new wave of technology, and to promote the business of the marketplace, the SEC also must come up with new working definitions of Frontrunning and Inside Information.
"We ask the subcommittee to encourage the SEC to revisit its regulatory regime for alternative trading systems and assure that ATS are held to the same standards as organized exchanges," said NYSE Euronext's Callahan.
We suggest the SEC start by figuring out what the standards of today's marketplace are. On the hop.