high-frequency trading vs. computer-driven trading: the Trillium fine

In the aftermath of the “flash crash” of a few months ago, regulatory authorities have been, almost frantically, looking for a cause.  My guess is they won’t be successful.  If there is a single culprit, I think he’s highly unlikely to come forward himself.  (Would you, if you thought you would be made a scapegoat and driven from the business?  Neither would I.)  And the complexity of Wall Street–almost any business, for that matter–comes from taking relatively simple ideas and repeating them over and over again, sometimes with subtle variations.  The result is an end product that’s difficult to unravel by anyone not immersed every day in that line of work.

But the search for an answer has apparently turned up non-related market abuses.  One of the is is the case of Trillium, a small brokerage house and some of its proprietary traders just disciplined by FINRA (the Financial INdustry Regulatory Authority).

to step back a minute

High-frequency trading (admittedly not an area I’m very familiar with, so leave comments if you want) seems to me to come in two flavors:

arbitrage This is the computer-driven search for pricing discrepancies that provide risk-free or low-risk trading opportunities.  These can be as simple as quirks in the bid-asked spreads of different market makers in the same security.  Or they can be differences in the prices of derivatives linked to the same underlying security, or closely-related securities.  Or they can be differences in the prices of a given theoretical financial attribute as contained in different securities or derivatives.

This activity isn’t particularly new.  Harry Markowitz started the ball rolling in the early 1950s.

order execution A large money management institution deals in very large position sizes.  It has two basic choices in controlling its buying and selling.  It can deal in the traditional way, by developing a staff of highly-skilled (and highly compensated) professional human traders, or it can use computers.  I’ve only worked in firms that took the former strategy, so I can’t say from experience how the latter works.

There are two (maybe three) problems with using human traders:  the buying and selling process can be slow, and the firm’s intentions can easily leak into the market through the counterparty before the transaction is completed.

Firms using computers break large orders down into many small ones which they hope to execute quickly with a number of different counterparties and on various trading platforms.  So they hope to get the trading done before information about their intentions has spread, and therefore with limited market impact.  As I mentioned above, I have no idea whether this works well or not.

The third issue, which I haven’t seen discussed anywhere, is who pays for the trading.  A traditional buy-side trading staff can easily cost several million dollars a year, money that is paid by the management company out of its management fees.  My guess is that high speed automated trading systems get poorer executions but are much less expensive to run.  If so, the management company saves the traders’ salaries and the clients pay the economic cost of trading through higher buy prices and lower sell prices.   On the other hand, if the blitzkrieg approach gets better executions, everybody (except the displaced traders) wins.

returning to Trillium

According to the FINRA documents linked above, what Trillium did was this:

Let’s say the best (highest) bid for stock ABC was $50 and the best (lowest) offer for ABC was $50.25.

Trillium traders would decide they wanted to sell ABC.  They would place a limit order to sell just inside the best offer, say, at $50.24.  They would then rapidly place a bunch of orders to buy ABC at different prices just below the best bid.  These orders were “often in substantial size relative to a stock’s over all legitimate pending order volume.”

The Trillium traders never intended to buy ABC.  What they wanted to do was use the false impression of rapidly building buy volume to trick day traders into thinking a powerful upward movement was about to start.  They wanted short-term traders to rush in to buy the stock to ride the impending uptrend.  Of course, the first shares to be bought would be the ones Trillium had pre-placed in the market through their limit order.

As soon as the limit order was executed, the Trillium traders cancelled their phony buy orders.

This is a boiler room operator’s ploy that’s as old as the hills.  It’s the kind of market manipulation the syndicates of the 1920s-1930s did, though on a far larger scale than Trillium.  And it’s one of the abuses that Depression-era reform of the securities markets was intended to stamp out.

is this high-frequency trading?

What does this have to do with high-frequency trading, though?  Nothing that I can see.

Yes, the Trillium traders probably used computers to enter their orders, both real and fake.  And they worked this scam over 46,000 times during the three months FINRA cited in its disciplinary action.  That’s about once every 30 seconds while Wall Street was open for business.  Otherwise, what’s the connection?

More interesting, what would cause the Huffington Post, or the Financial Times, or Fortune to say there’s one?  I don’t know.  The Fortune Street Sweep blog does have a clue, though.   In its post on Trillium, it quotes FINRA’s Thomas Gira, who’s identified on the FINRA website as executive vp of the Market Regulation Department as saying (as I read it) that Trillium’s is a high-frequency trading abuse.

Maybe there is a dark side to high-frequency trading, but I don’t see Trillium as a case in point.  I don’t get why FINRA would say it is, other than the agency must be under pressure to do something.

Leave a Reply

%d bloggers like this: