The Death of Money Read Online Free PDF Page B

trade and
     was sent to prison.
    When it comes to betting on a sure thing, greed trumps common sense and makes the
     bet irresistible. The record of insider trading is replete with such cases. A terrorist
     associate is not likely to show better judgment than a superrich celebrity when the
     opportunity arises.
    Given the weight of the social network analysis, statistical methods, signal amplification,
     and expert opinion, why did the 9/11 Commission fail to conclude that terrorists traded
     in AMR and UAL in advance of the attack? The answer lies in the 9/11 Commission Report
     itself, in footnote 130 of chapter 5.
    Footnote 130 admits that activity in AMR and UAL before 9/11 was “highly suspicious.”
     It also says, “Some unusual trading did in fact occur, but each such trade proved
     to have an innocuous explanation.” A closer look at these “innocuous” explanations
     reveals the flaws in the commission’s reasoning.
    For example, the report finds “a single U.S.-based institutional investor with no
     conceivable ties to al Qaeda purchased 95 percent of the UAL puts on September 6 as
     part of a trading strategy that also included
buying
115,000 shares of American.” This explanation falls down in two ways. First, the fact
     that a high percentage of the trades were found to be innocent is completely consistent
     with signal amplification. Only the small initial trade is done by terrorists. The
     9/11 Commission Report presented no evidence that it had made any effort to drill
     down to the small initial signal. Instead, the staff were beguiled by the innocent
     noise.
    Second, the 9/11 Commission relies on the fact that the investor it interviewed said
     he bought UAL puts as part of a strategy involving the purchase of AMR shares, a kind
     of long-short trade. This shows naïveté on the part of the commission staff. Large
     institutional investors have numerous positions that have nothing to do with one another
     but that can be selected post facto to show innocent motives to investigators. On
     its face, this investor’s AMR position says nothing about why it so heavily shorted
     UAL.
    The report goes on to say that “much of the seemingly suspicious trading in American
     on September 10 was traced to a specific U.S.-based options trading newsletter, faxed
     to its subscribers on Sunday, September 9, which recommended these trades.” This analysis
     shows that the commission staff had a limited understanding of how Wall Street research
     works.
    There are thousands of trading tip sheets in circulation. On any given day, it is
     possible to find at least one recommending the purchase
or
sale of most major companies listed on the New York Stock Exchange. Going back after
     the fact to find a newsletter that recommended buying puts on American Airlines is
     a trivial exercise. No doubt there were other newsletters in circulation recommending
     the opposite. Selecting evidence that fits a theory while ignoring other evidence
     is an example of confirmation bias, a leading cause of erroneous intelligence analysis.
    Another problem with the newsletter rationale is the belief that the recommendation
     arose independently of the insider trading already going on in AMR. Why treat the
     newsletter as a signal when it was actually part of the noise? For example, on September
     7, trading volume in AMR doubled from the previous day and reached a near three-month
     high with a declining stock price. This pattern is consistent with insider trading
     ahead of an attack on September 11. It is more likely that the September 7 put volume
     caused the September 9 newsletter recommendation than it is that the newsletter caused
     the September 10 put buying.
    The more likely explanation is that the entire sequence from September 6 through 10
     was a signal amplification caused by a small initial insider trade. To isolate a single
     event like the newsletter and give it explanatory power without reference to prior
     events