Who raised guidance this week? A screening habit that works

Why guidance changes are the signal

A raise is a company telling you, under legal scrutiny and with its credibility at stake, that business is running ahead of its own plan. A cut is the opposite confession. Both are costly signals: management pays a price for being wrong in either direction, which is exactly what separates signal from talk.

Clusters amplify the meaning. One raise is a company story. Five raises in the same industry inside two weeks is an industry story, and it often fronts the data you will read in macro statistics a month later. The screening question exists to catch the cluster while it forms.

The weekly screen, as questions

• "Who raised guidance this week, and in what words?" The base screen, with the framing attached.
• "Who cut, and what reason did they give?" Cuts cluster too, and the stated reasons compare revealingly.
• "Which raises were hedged?" A raise with three conditions is a different object than a clean one.
• "Any sector where raises and cuts both appeared?" Divergence inside a sector marks winners and losers separating.
• "Show the guidance history of the three most interesting names." The follow-up that turns the screen into candidates.

Reading the wording, not just the direction

Two companies raise full-year revenue guidance. One says "we now expect at least 4.2 billion, driven by demand we can already see in orders." The other says "we are raising the lower end of our range, assuming current conditions hold." Both hit the raise screen; only one deserves your attention. The words carry the conviction, which is why the screen must return quotes, not checkmarks.

Watch the specific tells: "at least" versus "approximately", named drivers versus general strength, whether the raise covers revenue only or flows through to margin and earnings. Flow-through raises are the strong form; revenue-only raises with flat earnings guidance often mean the upside is being spent.

From screen to position, carefully

A guidance screen produces candidates, not trades. The follow-up chain for each candidate: what does the guidance history look like over four quarters, does this management habitually sandbag or habitually stretch, and does the rest of the sector confirm the demand picture? Three questions, one conversation per name, and the screen's raw list turns into two or three names you actually understand.

The failure mode to avoid is treating the raise itself as the buy signal. By the time you read it, the raise is public and priced. Its value is as a filter for where to spend attention, and attention, applied through the follow-ups, is where any edge actually comes from.

Making it a ritual

The screen works because it repeats. Same question, every Friday of earnings season, same follow-ups on whatever surfaces. In earnings.chat the conversation persists, so week six can reference week two, and the season ends with a written record of who promised what and who delivered. That record is the input for the next season's first question: "who did what they said they would?"