Skip to main content

Broker Fined For Helping Some Robots Rip Off Other Robots

  • Author:
  • Updated:

What is your model of the SEC's recent crackdown on the naughty forms of high-frequency trading? I don't really like mine, and my model of high-frequency trading in general is pretty weak, so I hesitate to speculate. Here is me hesitating. Okay! Now let's speculate.

Go read this this SEC press release, or don't really, just search for "algorithm," spoiler: you won't find it. And why would you? Basically the SEC fined the auspiciously named Hold Brothers On-Line Investment Services $5.9 million1 for allowing some customers to do ... things ... that ... distorted some forces:

Daniel M. Hawke, Chief of the SEC Enforcement Division's Market Abuse Unit, added, “The fairness principle that underlies the foundation of our markets demands that prices of securities accurately reflect a genuine supply of and demand for those securities. The SEC will not tolerate any abusive practice that is designed to distort these natural forces.”

Which natural forces got distorted? Well, you could read the SEC's order, if you wanted. Here's a sample:

Certain overseas traders trading for Demostrate [one of the baddie customers - also, apparently, sic] engaged in extensive manipulative activity. Such traders induced algorithms to trade in a particular security by placing and then cancelling layers of orders in that security, creating fluctuations in the national best bid or offer of that security, increasing order book depth, and using the non-bona fide orders to send false signals regarding the demand for such security, which the algorithms misinterpreted as reflecting sincere demand. These overseas traders’ orders were intended to deceive and did deceive certain algorithms into buying (or selling) stocks from (or to) the Demostrate traders at prices that had been artificially raised (or lowered) by the Demostrate traders.

So the natural forces of supply and demand were "certain algorithms," which I suppose from one perspective is true. These algorithms looked at order book depth2 to adjust the prices at which they wanted to buy or sell, hoping to pick off Demostrate; instead Demostrate picked them off.

A sample scheme is set out in paragraphs 24-26 of the order; you can read it if you like3 but here's a summary:

  • at 11:08 and 55 seconds, W.W. Grainger stock (NYSE: GWW) was quoted at 101.27/101.37;
  • these dudes sold 1,000 shares at 101.34;
  • then they bought 1,000 shares at 101.30;
  • then they were done, at 11:09 and 2 seconds, and GWW was quoted at 101.24/101.37;
  • during the intervening seven seconds, they put in some phantom orders and generally kerfuffled around, with the inside bid getting as high as 101.34, and the insider offer as low as 101.31, at different times.

So, I mean, okay, I'll grant you that that's manipulative and distorting, and I suppose it picked off Good Algorithms, the kinds used by ... um ... retail investors. Wait, no, the kinds of algorithms used by Fidelity and stuff to invest your retirement money, and they shouldn't be picked off, they should invest your retirement money subject only to the whims of supply and demand and also each others' algorithms, which I suppose are sniffing around trying to find price moves and pick them off.

I was sort of amused a while back when the SEC got all "oh my God won't someone think of the markets" in fining NYSE for providing some services that gave a timing advantage to people who paid it money, while not worrying about other services - colocation being a big one - that did the exact same thing. Not because there's anything wrong with that! Some services are okay, some are not okay, far be it from me to judge which ones, I guess the SEC should. But the give-everyone-fair-access rhetoric was weird because everyone gets different access, and which access you get depends on how much you pay, and that's true both of the illegal thing and the legal thing, and, y'know, put that in your rhetoric and smoke it.

Same here right? These Hold and Demostrate characters DISTORTED THE VERY FORCES OF SUPPLY AND DEMAND, I guess, but by a couple of cents over the course of seven seconds. If you were all "I would like to buy some shares of W.W. Grainger today," and you logged on to your E*Trade account, and you saw it was at 101.27/101.37, and you were like "I'd lift that," and you went and pressed the button to buy 100 shares, you'd already missed the whole drama, because this far into this sentence you are slow at pressing buttons, okay?4 Seven seconds! No fundamental forces of supply and demand are changing in that seven seconds. This is fundamental forces of some algorithms versus other algorithms, and each of those algorithms represents in some imperfect and abstracted way "supply and demand," but not in a way a human could formulate in his mind or feel in his heart. And some of the algorithms are crafty, in that they step in to take the other side of the market when they see strength in a stock, and others of the algorithms are craftier, in that they put in fake orders to psych out the regular crafty ones.

There's this thing in the Journal about how the Germans are going to crack down on high speed trading by some combination of (1) telling you you can't be "excessive" and (2) telling you you can't just put in lots of orders that you don't trade on:

If the bill becomes law, "excessive use" of trading systems would come with added fees. Traders also would have to maintain a balance between orders and executed transactions. Both proposed rules are meant to to deter firms from a practice known as "quote stuffing," where traders quickly enter and withdraw orders, flooding the market with quotes and burdening systems.

If you're worried about lots of computers putting in fake orders that don't trade and distort natural forces, this seems like a path you might wander down.5 I'm less sure what the goal is of talking a big game about natural forces of supply and demand while evening out the landscaping on the playing field between high-speed algorithms.

SEC Charges N.Y.-Based Brokerage Firm with Layering & Order [SEC]
SEC Says New York Firm Allowed High-Speed Stock Manipulation [Bloomberg]
Day-Trading Firm Fined $5.9 Million Over Manipulative Trading [WSJ]
Germany Moves to Brake High-Speed Trading [WSJ]

1.Or something. It's parceled out weirdly; $5.9 million, from the Journal, is the largest number I found and therefore presumptively correct, but Bloomberg says $4 million.

2.Incidentally, if you wanted to look at order book depth, NYSE would suggest you look into their premium-priced Open Book Ultra product. The consolidated feed, you'll recall, doesn't carry depth-of-book information.

3.But you have to come down here to do so:

24. ... On June 4, 2010, the trader layered the stock of W.W. Grainger (NYSE: “GWW”) on NASDAQ and the Boston Stock Exchange.

25. That day, at 11:08:55.152 a.m., the trader placed an order to sell 1,000 GWW shares at $101.34 per share. Prior to the trader placing the order, the inside bid was $101.27 and the inside ask was $101.37. The trader’s sell order moved the inside ask to $101.34. From 11:08:55.164 a.m. to 11:08:55.323 a.m., the trader placed eleven orders offering to buy a total of 2,600 GWW shares at successively increasing prices from $101.29 to $101.33. During this time, the inside bid rose from $101.27 to $101.33, and the trader sold all 1,000 shares she offered to sell for $101.34 per share, completing the execution at 11:08:55.333. At 11:08:55.932, less than a second after the trader placed the initial buy order, the trader cancelled all open buy orders. At 11:08:55.991, once the trader had cancelled all of her open buy orders, the inside bid reverted to $101.27 and the inside ask reverted to $101.37.

26. Because the trader was now short 1,000 GWW shares, at 11:09:00.881, the trader placed an order to buy 1,000 GWW shares at $101.30, thereby changing the inside bid to $101.30. From 11:09:00.929 a.m. to 11:09:01.060 a.m., the trader placed eleven orders offering to sell a total of 2,600 GWW shares at successively decreasing prices from $101.35 to $101.31. During this time, the inside ask declined from $101.37 to $101.31, and the trader bought all 1,000 GWW shares she offered to buy for $101.30 per share, completing the execution at 11:09:00.977. At 11:09:01.662, less than a second after the trader placed the initial sell order, the trader cancelled all open sell orders. At 11:09:01.792, once the trader had cancelled all of her open sell orders, the inside bid reverted to $101.24 and the inside ask reverted to $101.37. This round trip transaction, which took less than seven seconds to complete, yielded the trader approximately $40.

4.Also you (1) use E*Trade but (2) talk like my stereotype of a bond trader. Your hypothetical self is a bit fractured, in that sentence.

5.But I'm sure it's a terrible idea for a million other reasons, that's not the point, etc.


Now Here Are Some Guys Who Knew How To Rip Off A Client

One aspect of good salesmanship is that you have to offer an attractive proposition not merely to the abstract entity that is your nominal client - El Paso, Italy, Greece - but also to the specific human being who is your contact at that client. Telling a corporate treasurer who is five years from retirement that a trade will have a significantly positive NPV due to huge cash flows in years 11-15 is not always as effective a sales technique as buying him a nice steak and an evening of unclothed entertainment. I suspect, though, that the latter strategy is more highly correlated with whatever you're selling ending up on the front page/op-ed page/ Anyway, I definitely admire these guys for this particular con*: The SEC alleges that Argyll Investments LLC’s purported stock-collateralized loan business is merely a fraud perpetrated by James T. Miceli and Douglas A. McClain, Jr. to acquire publicly traded stock from corporate officers and directors at a discounted price from market value, separately sell the shares for full market value in order to fund the loan, and use the remaining proceeds from the sale of the collateral for their own personal benefit. Miceli, McClain, and Argyll typically lied to borrowers by explicitly telling them that their collateral would not be sold unless a default occurred. However, since Argyll had no independent source of funds other than the borrowers’ collateral, Argyll often sold the collateral prior to closing the loan and then used the proceeds to fund it. Got it? Argyll gave corporate executives margin loans at 50-70% loan-to-value based on the market price of their stock (based on the volume weighted average price over five days leading up to the closing of the loan). They took the stock as "collateral." They then trousered the stock and sold it for, y'know, 100% of the market value, with 50-70% of that funding the loan and the remaining 30-50% funding miscellaneous expenses that presumably included unclothed entertainment for themselves. The loans had three-year terms and were not prepayable for 12-18 months, so the expected life of the scam was at least 12 months (but see below).