Before a company reports, option buyers pay up for the move nobody can forecast, and implied volatility climbs into the release. Once the number is out, that uncertainty is gone, and implied volatility usually collapses within hours. The bet is to sell into the tail end of that premium: after the announcement, while options are still priced as if the news were pending.
The seller collects the premium and carries the risk that the stock keeps moving the next day. That risk has no ceiling on a short straddle.
How it was tested
Pre-registered on 2026-09-02. The company reports after the close; 90 minutes later, implied volatility rank must be above 70 and the term structure flat or inverted. Sell an at-the-money straddle with 0-2 days to expiry, on names above $500M in market cap with options implied volatility above 40. Close at 50% of max profit, at the next day's close, or at a −5% stop.
There was no Jensen alpha gate. In its place stood a home-made "alpha vs theta baseline" that compared returns with an assumed 1.0% theta, never with a benchmark.
Where it died
The backtest reported a Sharpe of 7.15 and the strategy was marked Tier 1 for the demo. On 2026-10-17 a lab-wide review found the lab held no historical option chains and no earnings implied volatility, so nothing could have priced those trades. The verdict became void.
The twist came on 2026-10-23. Unlike the other option sleeves, whose executors only simulated, this one had really been sending orders to the paper account, selling straddles. It was removed from the demo launcher that day. No real money was involved.
The record
Thesis
Implied volatility is bid up into an earnings release and collapses within hours after it. Sell an at-the-money straddle shortly after the announcement while implied volatility is still elevated, and close it by the next day's close.
Specification
signal
earnings released at the close; IV rank > 70 and a flat or inverted IV term structure 90 minutes later
structure
short ATM straddle, 0-2 days to expiry
exit
50% of max profit, next day's close, or −5% stop
universe
market cap > $500M, options IV > 40
Gates
Pre-registrationpass
Was the hypothesis, its parameters and its pass/fail rule written down before any data was looked at?
Gate 7 — Jensen alphanot evaluated
After regressing out market exposure, is there alpha left?
Replaced by a home-made "alpha vs theta baseline" gate that compared returns to an assumed 1.0% theta. Never measured against a benchmark.
Why it died
The "Tier 1" backtest (67 trades, 64.2% win rate, 11.68% average P&L, Sharpe 7.15, t 3.69) cannot be verified: the lab has no historical option chains and no pre- or post-earnings implied volatility, so there is no source for the option prices those trades would need. Unlike the other option strategies, its connector did send orders to the paper-trading account — selling straddles, an unvalidated tail risk — until it was taken out of the demo. No real money was involved.
What it taught
A Sharpe of 7.15 on short straddles through earnings is not a discovery, it is a symptom. And an executor that really places orders is the most dangerous place for an unverified strategy to sit: unlike the two other option sleeves in the demo, whose executors only simulated, this one would have carried its mistakes straight to a funded account.
Verdict history
2026-09-02Pre-registered; reported 67 trades, Sharpe 7.15, t 3.69; marked Tier 1 for the demo
2026-10-17Declared unverifiable: no historical option chains or earnings IV Void
2026-10-23Removed from the demo launcher after confirming it sent orders to the paper account