On Tomorrow's Secret Meeting To Plot The End Of High Frequency Trading
The SEC's "definitive"(ly worthless) report on what happened on May 6th was a dud, and was nothing more than a distraction-based smear campaign against Waddell and Reed (an experiment in which we can only hope W&R participated involuntarily): a firm which did something that was completely in its right to do. But is this unexpected? After all had the SEC confirmed that it is indeed HFT who is responsible for a broken market structure, it would have effectively destroyed itself: if and when the SEC does indeed confirm that the entire market topology over the past 5 years has been hijacked by young and pustular math Ph.D.'s with fast computers, the implications to fair markets would be orders of magnitude worse than the fallout associated with the Madoff scandal, and could serve as grounds for the unwind of the SEC itself, which would have to explain why it has been avoiding calls against HFT impropriety for years. So in a sense Mary Schapiro's conclusion is nothing less than a lass desperate act of self preservation. Which however means nothing in the grand scheme of things. Tomorrow, as the WSJ reported a week ago, the Investment Company Institute, better known to Zero Hedge readers as the guys who track the now permanent weekly outflows from capital markets, is holding a secret meeting in which some of the participants "are determined to push for a plan to restrict high-frequency trading" (furthermore, the ICI was rather pissed about this particular leak, implying that things are really serious). While the SEC may have declared a market structure truce, and is peddling its usual worthless solution of circuit breakers (more on this below), actual market participants have had enough of seeing their profits plunge and HFTs extracting more capital out of the market than the much maligned ten years ago market makers and specialists ever did.
More from the WSJ on the dilemma facing the "whales" -traders who trade in blocks which are orders of magnitude larger than the typical (lately) trade size of 100 to 300 shares:
Using powerful computers and data feeds, high-frequency trading firms typically hold stocks a few minutes and sometimes only a few seconds at a time, churning roughly half of total stock-market volume. Whenever you—and your mutual fund or pension plan—buy or sell a stock, one of these fast-trading firms is likely to be on the other side of the trade.
The problem? While some fund leaders have praised high-frequency trading for making markets more efficient, others contend that the profits earned by fast traders may come partly at the expense of ordinary investors.
Mutual funds and other giant investors are often forced buyers and sellers. When money comes in they must buy stocks; when it goes out they must sell. Their typical buy or sell order is roughly 185,000 shares. Yet the average trade size on U.S. exchanges is only about 100 to 300 shares.
So institutions trade in dribs and drabs. A giant buy order would push up a stock; a huge sell order would knock it down. "Would you leave $100 in cash on the street corner and hope nobody takes it, or would you hide it in your pocket?" asks Andrew Brooks, head of U.S. equity trading at T. Rowe Price. "Information about our order flow is valuable, and we need to protect it."
Yet order flow leakage and anticipation is precisely the key issue behind the great battle currently raging between HFTs and slow money: algorithms constantly seek and find new ways to predict if a small block is isolated or if there are millions of shares lying in wait behind it, forced to transact at any price. Themis Trading's Joe Saluzzi has discussed this issue extensively before, for example in this article in Advanced Trading magazine, in which he notes that implicit costs associated with constant order frontrunning by HFTs, "have been rising and hurting pension and mutual fund performance. Even the most sophisticated buy-side quantitative funds are also experiencing higher trading costs, as their models are also being spotted and taken advantage of by their highspeed, high-frequency trading cousins."
The WSJ explains this phenomenon as follows:
Any institutional order for a couple hundred shares can have thousands
or even millions of shares behind it. A fast trader that can infer which
orders were placed by a big institution gains an insight into how stock
prices may be about to change. Whoever gets there first stands to make a
tiny profit on each of those trades.
Some of the larger exchanges are now fighting back with all they have - starting tomorrow the NYSE is now instituting a small way to throw a wrench in the HFT spokes: selective order ID elimination.
Direct data feeds supplied to fast traders by several major exchanges have customarily included an "order ID"—a kind of tag that, according to several traders, may assist a fast trader in deducing whether a large institution lurks behind a small order. Starting Oct. 4, the NYSE Arca exchange will give customers the option of having these IDs removed from its direct data feeds on orders they don't want displayed to the whole market.
Traders say the absence of the ID may itself alert rapid traders to the presence of a large, hidden order. "Without the order ID, we don't think anyone could map the order to any other information to divine that it is part of something larger," responds Ray Pellecchia, an NYSE Euronext spokesman.
Another exchange, recently known for doing pretty much anything to win order flow, and thus permitting all sorts of allegedly shady practices to take effect on it is BATS, which chimes in on another practice known as "partial post only at limit."
Or consider a type of trading order called a "partial post only at limit." Here, if a fast trader's small buy order is rejected instead of executed, the firm can deduce that a large block of shares may lie hidden in reserve, poised to sell at a given price. Thus a trader may be able to get information without executing the trade. Clever use of this order type can increase the trader's odds of being in the right place at the right time—capturing a splinter-thin, lightning-fast profit before the institution can move.
Chris Isaacson, chief operating officer at BATS Exchange, the third-largest U.S. stock market, downplays such concerns. "This order type is rarely used and would be very complex to implement for the purpose of detecting a large order on the other side," he says. Such a trader "would have to be willing to take considerable risk."
Which is why, since the SEC refuses to get involved, starting tomorrow, all readers should immediately notify their brokers to stop allowing their orders to be Flashed (a topic extensively discussed last summer and which is still continuing with the SEC's blessing) if they have not already done so (as this implicitly allows non-qualified orders to be front run), and to stop trading with any exchange with either has no idea what this is, or refuses to comply. More importantly, brokers should also be advised to drop all order IDs tagging each and every individual order. Since the HFTs front run stock blocks based on a statistical distribution of tagged versus untagged orders, the only hope of equalizing the playing field is if every single order is now untagged, thus fooling HFTs into believing that there is large money sitting in the bid behind the order. Granted, this may force prevailing prices higher for the time being, but the end result will be a faster divergence from market equilibrium, which eventually, when the balance inevitably reasserts itself, will force the vast majority of HFTs to be blown up once there is a market correction and the computers are caught with artificially dollar-cost average inflated prevailing prices. In other words: open war on HFTs has been declared, and since regulators refuse to stand on the side of the small and long-term investors, it is time to form a unified block against the HFT scourge.
And while we are once again (as always) discussing the SEC's terminal incompetence, we wish to present for one last time just what happened on Friday in the LQD flash crash, courtesy once again of Nanex. Note the dubious absence of Waddell and Reed:
Chart 1, which demonstrates what exchange(s) the rogue algo originated from: note the PACF repeater and BATS stubbing all the way down:
Chart 2, which shows the actual LQD trades:
Chart 3, which confirms that the PACF had the BBO all the way down. Once again, no W&R
Is it thus any wonder the industry itself is now ganging up against HFT? We believe that now that HFT is scapegoated by real money accounts for all P&L problems (and otherwise), its days are effectively numbered. Which is why as the HFT world enters its death rattle hours, expect markets to be even more unpredictable and volatile than ever... and of course, to ultimately break as usual.