Barclays Isn't Out of Its Dark Pool Trouble Yet

If you ever want to tell a client something that isn't strictly true, maybe start it with the word "sample."

Eric Schneiderman, preying on the predators.

Photographer: Louis Lanzano/Bloomberg.

Last year I proposed that the word "sample," in financial presentations, means "not true." This was in the context of discussing a chart that Barclays had made, showing the "sample liquidity landscape" in its dark pool, which New York Attorney General Eric Schneiderman felt was misleading insofar as it did not represent the actual liquidity landscape in the dark pool. 1  My view was that Schneiderman had simply made a mistake about the meaning of words: "Sample" means "not true," so it's fine that Barclays's landscape picture was inaccurate.

I was, like, 50 percent kidding. But then Barclays took the same line, responding to Schneiderman's fraud lawsuit by arguing that the chart "plainly" was not "purporting to represent the 'liquidity landscape' of Barclays' dark pool." And last week a New York state judge endorsed that proposition as a matter of law, saying that "charts labeled 'sample' are not actionable" and explaining in a footnote (citation omitted):

The graph labeled "Sample liquidity landscape by category" clearly does not represent what the NYAG says it does for the reasons explained in Barclays' briefs. That chart was not representing the actual composition of the Dark Pool's traders, and, in any event, that composition was obviously subject to change. Sophisticated investors know this. Hence, that chart is not materially misleading.

Justice Shirley Kornreich denied Barclays's motion to dismiss Schneiderman's lawsuit, so in some narrow sense that's a loss for Barclays, but let's stay on the big picture for a minute. The big picture is that -- and I remind you that I'm not your lawyer and this is not legal advice and in fact it is explicitly sarcastic and you should ignore it -- but the big picture is that now you can say whatever you like as long as you precede it with the word "sample." 

This is a matter of no little import. If, for instance, you are pitching to lead an initial public offering, and you construct an accurate ranking of the biggest IPO banks, and it shows that you're ranked 17th, that's not great. If you construct a totally fake ranking that puts you first, someone might find out, and you might get in trouble. If you construct a totally fake ranking that puts you first, and label it "Fake League Table," the client will not be impressed. But "Sample League Table"? "Sample" works. Print it!

That's not the only positive takeaway for Barclays in last week's decision, which was a bit odd procedurally. Barclays had moved to dismiss Schneiderman's case against it, and then Schneiderman filed an amended complaint a few weeks ago, and then Barclays moved to dismiss that complaint too, and then Justice Kornreich decided not to dismiss the last complaint. Her decision turned on whether or not the Martin Act, New York's weirdly broad anti-fraud law, applies to dark-pool fraud. Barclays cleverly argued that the Martin Act applies only to fraud in buying and selling securities, not fraud in marketing a venue on which to buy and sell securities, which is a pretty good argument, but which suffers from too-cuteness and was rebuffed. 2

But Barclays's other argument for dismissing the complaint, on which it placed rather more emphasis, is that it didn't commit fraud in marketing its dark pool. Sure it may have pushed its use of terminology in rather hypertechnical directions, but no one was misled by that. Its customers, after all, were sophisticated investors and were used to hypertechnical parsing of terms.

And Justice Kornreich did not exactly decide on this question, but I guess she still might, and she showed some sympathy for Barclays:

It should go without saying that a Martin Act claim cannot be based on representations on which no reasonable sophisticated investor (the only investors who trade in dark pools 3 ) would rely. Drafts of power point marketing decks and charts labeled "sample" are not actionable. Moreover, though Barclays may have colloquially referred to certain high frequency traders with seemingly derogatory words such as "toxic", that is a meaningless term. A valid fraud claim, for instance, must be based on representations about types of trading counterparties that include enough of a degree of specificity about the characteristics of the traders for the representations to be actionable, as opposed to impermissibly vague descriptions and sheer puffery. The focus is on what a reasonable, sophisticated trader, trading millions of dollars, actually would consider material. The court, however, will not assess the pleading sufficiency of the AC until the parties have fully briefed the forthcoming motion to dismiss it.

I don't know, that sounds pretty promising for Barclays, no? There is a strangely popular and successful method of argument that consists entirely of calling high-frequency trading mean names, but Justice Kornreich will have none of it: She was unimpressed by "the NYAG's rhetoric about the harms of HFT," and said: "The investors in the Dark Pool are highly sophisticated and, hence, no liability will be found simply on the basis of meaningless words, such as 'aggressive', 'predatory', and 'toxic'." 4   

One problem for Barclays -- a marketing problem if not a legal one -- is that those were its own words. It was going around calling its own high-frequency-trading clients "toxic" in internal materials, at the same time that it was at least implying to other clients that they could avoid "toxic" or "predatory" or "aggressive" high-frequency trading by using Barclays's dark pool. 

The problem is partially resolved if you remember that, just as "sample" means "not true," "toxic" means nothing. "Toxic" is a way of conveying general world-weariness about high-frequency trading; it's a code word that lets you convey sympathy with those who are uncomfortable with the modern world of speedy trading. When you say "toxic," it sounds like the high-frequency traders are doing something wrong. But it doesn't mean that. It just sounds like it does. In fact, the disputed measure that Barclays used to define toxicity turns mostly on whether the high-frequency traders made money on their trading. 5 If you define "toxic" high-frequency trading to mean profitable high-frequency trading, you can hint that the way high-frequency traders make money is illegitimate without actually saying it. 

And, obviously, if those words don't mean anything, then saying them can't be fraud. They can't be material to sophisticated investors if they don't convey any factual claims: Sophisticated investors aren't misled by mere sympathetic noises, but only by cold hard (true or false) factual statements. Or so says the law. Obviously, as a matter of marketing you might want to make sympathetic noises even to sophisticated investors, and sometimes those sophisticated investors might respond.

That was Barclays' strategy, marketing its dark pool as a way to avoid "toxic" high-frequency trading without meaning anything too definite by it, but it's Schneiderman's strategy too. He uses words like "toxic" and "predatory" because Barclays does, but he doesn't define them. 6 His complaint seeks damages but doesn't explain how anyone was harmed by Barclays' misrepresentations. 7 There are a lot of claims that Barclays exposed investors to "toxic" and "predatory" traders, but no claims that those "toxic" and "predatory" traders did any harm to anyone. "Toxic" and "predatory" and "aggressive" are just marketing terms. You shouldn't expect them to mean anything.

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.

  1. Ugh have we talked about the metaphor in "liquidity landscape"? Seascape, surely.

  2. A little lamely, to be honest:

    Finally, to the extent the applicability of the Martin Act to representations about trading venues can be considered a close call -- and, for what it's worth, Barclays' arguments are not entirely unreasonable -- this court views the Court of Appeals' guidance on the Martin Act to be that doubts in favor of the Martin Act's applicability should be resolved in the NYAG's favor. In other words, if it is a close call, the Martin Act should be held to apply. New York, the center of the financial universe, benefits greatly from having powerful blue sky laws.

    So ties go to the Attorney General, basically.

  3. By which one assume she means something like, "The only investors who make the affirmative choice to trade in dark pools (and are thus susceptible to fraud in dark-pool marketing)." Obviously lots of investors trade in dark pools without even knowing about it. Ponder UBS's retail-heavy dark pool, for instance.

  4. That's from footnote 8 of her opinion, which, disclosure, contains a citation to this post by me. Neither Justice Kornreich nor Barclays, however, cited me for the proposition that "sample" means "not true."

  5. See item 12 on this FAQ for Barclays' dark pool, about "Liquidity Profiling," the metric that got Barclays into much of its trouble. The two worst categories for liquidity profiling -- the categories that Schneiderman and Barclays thought were "toxic" -- are the ones with the highest one-second take alpha, that is, the ones that make money quickly.

  6. I mean, there's a sort of pseudo-definition in paragraphs 21 and 22 of the amended complaint, saying that Barclays used "toxic," "predatory" and "aggressive" to describe strategies like latency arbitrage to "exploit the small, temporary pricing dislocations in a security that occur because of differential and/or delayed access to market data." And latency arbitrage is mentioned a few more times, by way of example, e.g. in paragraph 148. But that doesn't seem to be a definition; Schneiderman doesn't argue that there was a lot of latency arbitrage, and mostly relies on Barclays' own measurements of "toxicity," i.e., profitability.

  7. From Justice Kornreich's decision:

    One wonders that, if not for the existence of HFT in the Dark Pool, whether these investors could possibly have processed their desired order volume in the Dark Pool, or without HFT, would the resulting diminished liquidity of the Dark Pool necessarily have required the investors to execute in other venues, where they might have ended up trading with those same HFT counterparties anyway.

    Barclays's failure to protect clients from HFT maybe looks less bad if no one could protect them from HFT.

To contact the author on this story:
Matt Levine at

To contact the editor on this story:
Zara Kessler at

Before it's here, it's on the Bloomberg Terminal.