Signals Versus Noise: What Actually Carries Information

If people-signals were easy to read, everyone would read them and the edge would vanish. The reason an edge survives is that most of what looks like signal is noise, and separating the two takes discipline. So before we start hunting, let me be precise about what carries information and what only looks like it does.

Start with a simple test I apply to any candidate signal: what did it cost the person to send it, and could they have sent it just as easily if nothing were happening? A costly, specific, hard-to-fake move carries information. A cheap, vague, easily-faked one usually does not. A company hiring a named head of integration from a competitor is costly and specific. A CEO saying on an earnings call that they are 'always looking at opportunities' is cheap and vague — every CEO says it, in every quarter, whether or not anything is brewing. The first is worth logging; the second is filler.

This cost-and-specificity test does a lot of work. It tells you why executive departures can matter: a senior operator leaving a healthy company for a quiet advisory role at a smaller one is an odd, costly move that a pending transaction would explain neatly. It tells you why a single job posting is weak but a coordinated wave is strong: one posting is noise, but twenty postings in the same function and same geography, opened in the same fortnight, are a coordinated act that took planning and budget. And it tells you to distrust anything a company can produce for free to shape perception — vague strategic language, a splashy but non-binding 'partnership,' an anonymous rumor that conveniently moves a price.

Now, the harder half of the discipline: base rates. A signal is only useful in proportion to how much it changes your estimate, and that depends on how common it is. Executives change jobs constantly. Companies post jobs constantly. Bankers pitch constantly. If you treat every hire as a pre-deal tell, you will 'predict' hundreds of deals for every one that happens, and you will be right often enough by chance to fool yourself. The way to stay honest is to always ask: out of all the times I have seen this exact signal, how often was a real deal actually behind it? If the honest answer is 'rarely,' then the signal on its own is close to worthless, no matter how compelling the story around it feels.

This is where triangulation earns its place. One weak signal times another weak signal can be strong, because the coincidence of two unlikely-together things is far rarer than either alone. A strategy hire is common. A specialist banker mandate is common. A hiring wave in a specific city is common. But all three, pointed at the same target, in the same quarter, is genuinely uncommon — and uncommon is exactly what makes it informative. The goal is never to find one magic tell. It is to find a convergence that innocent explanations cannot cover.

Let me add a caution about timing, because it bites people. Signals decay and they also mislead on speed. A hiring wave might precede a deal by two quarters, or the deal it hinted at might collapse and never appear at all. Reading the signal correctly — 'this company is preparing to transact' — is not the same as knowing when, or whether it closes. Many well-founded reads are early, and being early can look identical to being wrong for an uncomfortably long time. A disciplined reader separates the two claims: 'preparation is underway' is one statement; 'a deal will close by a certain date' is a much stronger one that the people-signals rarely support on their own.

Finally, guard against the story you want. The most dangerous noise is not random; it is the noise that flatters a thesis you already hold. If you have decided two companies should merge, you will find hires, moves, and mandates that seem to confirm it, because those things are always happening somewhere. The antidote is to write down, in advance, what specific signals would count and what timing would count, and then to hold yourself to that list. Deciding the evidence rules before you see the evidence is the single best protection against reading tea leaves.

So: cost and specificity tell you what to log. Base rates tell you how much to move. Triangulation turns weak signals into a strong read. And pre-commitment protects you from your own narrative. Keep those four in mind, and the rest of the course is just applying them to specific kinds of moves.