On 30 October 2026, Landstar’s President and CEO bought 5,000 shares at a weighted average of $142.9238, $714,619, two days after the stock fell 17.8% on a Q3 miss driven by insurance and claims costs — not by demand, since revenue rose about 25%. Every wire wrote it up as a conviction buy. I went looking for the usual ways that story falls apart, and it did not fall apart.

• Not a plan trade. No 10b5-1 designation; a discretionary open-market purchase, filed on 2 November.
• Not token. His direct holding went 23,110 → 28,110 — a 21.6% increase in his own stake in a single trade. The usual debunk (“large dollar figure, trivial fraction of the position”) runs the wrong way here: the dollar figure is modest and the fraction is large. That is the opposite of the Rollins entry above, and the contrast is the point.
• Not a window-opening artefact. Post-earnings blackout lifts produce clusters. No other Landstar insider bought. In fact no Landstar insider had bought anything on the open market since director Homaira Akbari’s 500 shares on 15 March 2022.
• Not an ownership-guideline mechanic. Those produce regular small purchases over years. The record shows a four-and-a-half-year gap, then this.
• Price formation is genuine. A weighted average to four decimals across a real range, which is precisely the tell that separates this from the Bambusa print.

Where I would caution against over-reading: CEO open-market purchases have weak predictive power over three- to six-month horizons, and the base rate for “insider buys the post-earnings dip” is unimpressive. The claim that survives is narrow and worth exactly what it says: a chief executive formed a personal view that an insurance-cost miss was mispriced as a franchise miss, and backed it with roughly a year of base salary of his own money. That is information. It is not a forecast.