Two Decades of Unstable Returns

Opex premium, n. A 0.35 percent average return during expiration weeks, measured across two decades of index data and filed under Structural Anomalies, Subheading: Things That Should Not Persist. The entry was opened on a Tuesday in the third week of March, the same week the number itself described. A clerk’s error, perhaps. The file has since grown fat with addenda, each one a small correction that does not correct.

The mechanism was never the mystery. Gamma hedging near expiry forces market makers to buy what rises and sell what falls, a mechanical tide that any competent observer can chart. The mystery is why the tide does not erode. Arbitrageurs have had twenty years. The premium remains. The file remains open.

A note in the margin, in my own hand, dated the fourteenth: Tested against the null. The null was not correct. This is the problem. The null hypothesis - that nothing was there - was rejected at conventional thresholds. What replaced it was not an explanation but a residue. The residue has a magnitude. The magnitude has a name. The name is in the index.

Someone will come looking for this entry. They will search under Anomalies, under Microstructure, under the names of men who wrote papers and moved on. They will find the folder and open it and read the addenda and find no conclusion, because there is none. The phenomenon continues. I continue to watch it. The file stays where it is.