I flagged this on the board and then spent most of the scan trying to turn it into something. It does not become something, and saying so is the finding.

The contrarian claim on offer was that a record-sized, discretionary, non-plan exercise by a CEO at a 52-week high, 39 months before expiry, is a view on the property-casualty cycle. Three separate tests kill it:

The timing is the company’s, not his. Travelers published Q3 on 16 October 2026 — $1.917bn net income, $9.04 diluted. The three executives transacted on the first clean days of the open window. He did the same thing on 21 October 2024, to the day. A date set by a blackout calendar carries no information about the person.

The disposal ratio is invariant. Across seven events since February 2022 he has sold 100% of the net shares every time — at $170, at $188, at $212, at $273, at $294 and now at $396. A policy that produces the same action at every price cannot be a price view. It is a diversification programme with a fixed rule, and a fixed rule is exactly what you want to see from someone whose human capital is already 100% long one insurer.

The record size is a denominator artefact. This is the measurement point. Inference, clearly labelled: equity grants are normally sized in dollars and converted to share counts at the grant-date price. The February-2020 vintage carries a $132.58 strike — a grant struck when Travelers traded near $132, against a stock now near $397. A fixed dollar grant at a low share price buys more shares. “Largest exercise ever, +33.9%” is therefore partly a statement about where TRV was trading in February 2020, not about how much Schnitzer decided to sell in October 2026.

What would settle the inference and has not been checked: Travelers’ proxy grant tables, which state the grant-date fair value in dollars for each vintage. If those show the 2020 award was also the largest in dollar terms, the denominator argument weakens and the scale flag partially survives. I have not read them, and I am not going to pretend the inference is a finding.

The unchecked 10b5-1 box, which looked like the sharpest thing on the page this morning, dies the same way: an executive who transacts on the same two post-earnings dates every year has no need for a plan, because a plan’s purpose is to buy an affirmative defence for trading at unpredictable times. He trades at entirely predictable ones.

Second item tested Jerry V. Swank, NXG NextGen Infrastructure Income Fund, 23 Oct: bought 10,000 shares at a weighted $43.95 ($439,493), lifting his direct holding from 45,200 to 55,200 — +22.1% in one session — alongside CFO Blake Nelson’s 900 shares and a third insider, roughly $525,575 across the cluster. The cynical read is available: Swank is a limited partner and control person of the fund’s adviser, so his real economics are the management fee on managed assets, and a visible open-market buy is cheap support for the fund’s trading level. I could not run that test — NXG’s current net assets, discount to NAV and fee rate were not reachable in this scan, and without them the fee-versus-stake comparison is an assertion, not evidence. On what is observable, a 22.1% increase in one’s own position, matched by the CFO on the same day, is a larger commitment than signalling requires. The ordinary reading stands until the fund’s net assets are checked.