SEC database hoax sent Avon stock soaring

Did people with dollar signs in their eyes jump on an obviously fake bid for Avon? Or is the story that what was thought to be a secure way of gaining information has been manipulated?

An S.E.C. filing that appears to have been a fake bid for the cosmetics company raises questions about the potential for market manipulation.

Source: A Phantom Offer Sends Avon’s Shares Surging – NYTimes.com

The federal government’s system for filing securities documents may not be as secure as many on Wall Street assume, considering what appears to have been a fake bid on Thursday for Avon Products.

The bid, which was reported on the Securities and Exchange Commission’s online database, points to a hole in the plumbing of the United States’ securities market.

Around 11:30 a.m., a firm calling itself PTG Capital Partners disclosed in a regulatory filing that it had offered to buy Avon for $18.75 a share.

The filing, from what was supposedly a British investment firm, caused an immediate sensation after news wire services flashed headlines of the offer to their trading customers. Shares of Avon, which had opened the day’s trading at $6.71, jumped by more than $1.

In about an hour, some had figured out that PTG Capital was not real and the law firm listed in Texas, Trose & Cox LLC, did not exist at the locations given in the regulatory filing. Avon issued a statement saying that it had not received an offer. The filing also contained grammatical and spelling errors which appeared to be the first sign of trouble. (You will see similar in many advertisements for dietary supplements and alt med treatments, it’s at least a sign that the company does not have any English-speaking overseers. If they don’t care to proof their ads, would you trust buying from them?)

Fortune notes:

[I]t is unclear how the perpetrator would have gotten the document into the SEC’s system. All corporate filers, including Avon Products, file electronically via a platform called EDGAR. In order to do so, however, they must use both privately-held confirmation codes and passwords.

EDGAR is seen as secure and trusted. Now, it appears vulnerable to manipulation.
capture2

A Bloomberg piece is harsh, saying that this was a transparent hoax, in part, “attributable to dumb people scanning the press release, seeing the offer and buying”. It should have been clear that there was a problem with this offer. (Easy to see this in hindsight, though.) However, it’s probably a bit difficult to be the first one to raise the red flag when a money grab is at hand and seconds count. But it does reveal that too many didn’t engage any skeptical thinking.

Forbes comments that this bid was “designed to fool word-scanning, dumb computer trading systems.”

“This was a fraud designed for algorithmic traders,” said John Fahy, a former SEC enforcement attorney and member of Whitaker Chalk in Fort Worth, Texas, where “PTG’s” lawyer was supposedly located and where TPG is based. “It was not designed to fool anybody who’d actually read it. It was designed to fool some system that scans SEC filings for certain words but doesn’t actually read them.”

That raises a second question, Fahy said: Since securities fraud laws target information that is “material” to investors and this shouldn’t have fooled any human investors, does it still meet that standard?

“I would not be surprised to see someone argue that no reasonable investor could have possibly relied on the filing due to its absurdity and automatic trades by algorithmic trading programs cannot cause a stupid filing to meet the materiality standard,” he said.

So, the world of finance is greatly in need of some Practical Skepticism.

  8 comments for “SEC database hoax sent Avon stock soaring

  1. J
    May 15, 2015 at 9:41 AM

    A very interesting and unique story for this site. Although, the “no reasonable investor” smacks of the “no true Scotsman” fallacy, so it’s not surprising that the algorithms designed to analyze bids also rely on the language investors use to do the same, even if the intent was not to fool the people. Can biased mathematicians make biased algorithms? I think so.

  2. May 15, 2015 at 10:02 AM

    Point taken.

  3. Steve
    May 15, 2015 at 10:22 AM

    Does the Turing test test for lies and include skepticism?

  4. May 15, 2015 at 11:16 AM

    * soaring

  5. May 15, 2015 at 2:55 PM

    I can’t believe how long it took me to register that…

    Thanks. Fixed.

  6. Rook
    May 16, 2015 at 10:56 AM

    Stock price spikes (either up or down) due to automatic trading isn’t really new; these systems specifically look for key words and phrases in the news and react accordingly. And it’s certainly not that first time that a bogus news story or misinterpreted news story has caused such reaction. Usually the stock price returns to normal fairly quickly, due to activity by actual human investors.

  7. Bill T.
    May 18, 2015 at 2:05 PM

    No, it is a proposed protocol to determine intelligence of a machine (algorithm) only.

  8. Kevin Fries
    August 3, 2015 at 6:16 PM

    I disagree with the use of the Scotsman fallacy categorization. The “reasonable person” test is a standard piece of judicial process. Negligent homicide, for example, is often judged based on what a reasonable person would be expected to know and do. As well, it is not really being used as an ad hoc justification for some other untenable position. However, the Scotsman fallacy could come from somebody arguing about this characterization. No reasonable person would think of this example as an example of a Scotsman fallacy.

    I don’t understand why a biased analyst would want to construct a biased model, as bias would result in incorrect results. DOn’t you think that the whole reason to build these models, and to included data outside of the trading numbers, is done to reduce the level of bias in the model? Perhaps I misunderstand what you are saying, if you instead mean that individuals bias can leak into his models, this is of course possible, especially as most bias is unconscious. Presumably the models that are actually used have a demonstrated records of utility, and thus at least balanced bias in analyzing data. I think the issue has more to do with computers using these human directed communications and using these directly as data points so they can make decisions on trading in thousands of a second. Having a human vetting or abstracting the data could avoid things like this, but at the cost of time, which defeats the whole point of trading algorithms.

    I think hyper frequency trading is the real problem in this situation, where human judgement is removed.

Comments are closed.