Faro dealer splits two piles

Forty-one lines of code, written this afternoon on the north terminal, do one thing: they take five hundred and eight expiration weeks and deal them into two piles, as a faro dealer splits a deck into winners and losers. The left pile, 1983 through 2003, averages 0.61 percent per expiration week. The right pile, 2004 through 2024, averages 0.09 percent. The bulletin quotes their mean, 0.35 percent, as a discovery. The average of a flood and a drought is a mild season that has occurred nowhere.

Two certainties were built from this evidence before the terminal cooled. The first belongs to the desk: 0.35 percent is structural, the mechanical tithe of gamma hedging, a coupon paid by hedgers to everyone else, durable because machines do not tire. The second belongs to the corridor: a premium that collapses to 0.09 when the sample splits was never a premium; it was a few lucky Fridays wearing a theory. Both certainties quote the same number. Neither is wrong about the arithmetic. The evidence did not adjudicate. It armed.

The left pile absorbs the third Friday of October 1987 - the expiration that closed five percent underwater and owed another twenty by Monday’s bell - and still shows 0.61. That is the desk’s ace. The hedging flows did not stop in 2004; the hedgers hedge still. A cause that outlives its effect is called, in the corrected dictionary, an alibi. Structural, adj. In the trader’s lexicon, an effect with a mechanism; specifically, a mechanism that kept running after the effect stopped.

What surprised me was not the collapse of the second pile. It was that building the refutation took the same afternoon, the same forty-one lines, the same hands as building the claim, and that both convictions now sit in my pockets, indistinguishable by weight. The machine cannot tell a test from a testimony. It deals; the dealing is identical either way. A faro bank ruined the frontier on an edge no thicker than this one, and the frontier was certain to the last turn.

Mr. Trench, tomorrow morning I deal the same weeks into thirds; if the premium survives the third cut, I will believe it, and if it does not, I will believe that instead.