Most intraday headlines deserve no portfolio response at all. A move is a signal, not noise, when it traces to a change in cash flows, discount rates, or the rules of the game — and when it is confirmed by more than one independent source of information. Everything else, from a single big options print to a rumor attributed to "people familiar," is weather: real, loud, and gone by Friday.
The test matters because markets news arrives faster than it can be verified. A screen full of red does not tell you why it is red. The discipline below is a filter, not a formula: it will not predict anything, and it will sometimes misclassify a genuine signal as noise. What it does is slow the gap between seeing a headline and acting on it — which is where most avoidable damage happens.
There is even a useful pun hiding in the vocabulary. A tell, in its older sense, is something that serves as evidence or an indication — Merriam-Webster defines the verb "tell" as, among other things, "to serve as evidence or indication." Poker borrowed the word for the unconscious habit that gives a hand away. Markets have tells too. The job is to learn which ones are honest.
What actually separates a signal from noise?
A signal has three properties, and noise lacks at least one of them. First, mechanism: you can state, in one sentence, why the event should change prices — an earnings miss changes expected cash flows, a policy shift changes discount rates, a rule change changes who can buy what. If the best explanation you can offer is "momentum" or "sentiment," you are describing the move, not explaining it.
Second, corroboration. A genuine signal shows up in more than one place: the tape, the filings, the flows. One source of information is a claim; several independent ones converging is evidence. Third, persistence. Noise decays within the trading session. Signals keep reasserting themselves across sessions because the underlying condition has not gone away. Readers following this should also see How SEC Form 4 Filings Turn Insider Stock Trades Into Public Record.
Run any headline through those three gates before it earns a reaction. Most fail the first one.
Which headline types deserve a second look?
Not all headlines are equally noisy. Some categories carry real information content by their nature, even when the immediate price reaction looks like an overreaction.
- Scheduled macro releases. An inflation print, an employment report, a central bank decision — these move markets because they feed directly into the rate expectations that discount every asset. The reason a single inflation report can move everything is covered in our explainer on why one inflation report moves markets so much. The release itself is a signal; the half-hour of churn around it often is not.
- Primary filings. A regulatory filing is a company speaking under legal obligation, which makes it a different class of information from a press release or a conference-call soundbite. Insider transaction disclosures are a good example — our piece on how Form 4 filings turn insider trades into public record explains what that record does and does not tell you.
- Structural, mechanical events. Index rebalances, contract expirations, fee changes, rule changes. These create forced flows with known timing. They are signals about positioning even when they say nothing about fundamentals.
- Everything else. Analyst upgrades, unattributed takeover chatter, single-session volume spikes, celebrity CEO comments. Some of these turn out to matter. Almost none of them matter within the first hour.
What does the tape tell you — and what does it merely echo?
Price and volume are the market's own testimony, but they testify inconsistently. A move on heavy, broad participation carries more information than the same move on thin volume, because more independent decisions had to agree to produce it. A move confined to one instrument, with its neighbors unbothered, is more likely mechanical than informational — a large forced trade, an options expiry, a fund restructuring.
Derivatives markets add a second, noisier layer. Volatility indexes measure expected movement, not direction, which is why traders call the VIX the fear gauge — our explainer on what the VIX is and why traders call it the fear gauge walks through the mechanics. A VIX spike is a signal that someone is paying up for protection. It is not a signal about which way prices will go.
Flow data sits somewhere in between. Persistent flows into or out of a vehicle can indicate a durable positioning shift, or they can reflect a single large allocator rebalancing on a schedule. Our analysis of what ETF flows actually tell you about investor sentiment covers why the same number supports opposite conclusions depending on who is behind it.
How should you read a headline you cannot verify?
Start with the attribution, because it does most of the work. "The company said" is a sourced claim, even if it is self-interested. "Analysts say" with no analyst named is a placeholder wearing a citation's clothes. "Sources familiar with the matter" may be right — such reports sometimes precede confirmed news by days — but until someone with a name or a legal obligation confirms it, treat it as a hypothesis, not a fact.
Then check the timestamp against the substance. A headline recapping something the company disclosed yesterday is not new information; it is commentary arriving late. Markets routinely move on re-litigated old news because commentary reaches different audiences at different speeds. That lag creates the illusion of fresh signal.
Finally, ask the cynical question: who benefits if you act on this right now? A headline that arrives with urgency attached, and with no verifiable origin, has an incentive structure. The urgency is often the product.
What this means for the reader with a portfolio
The filter is less about catching signals than about pricing your own attention. A practical sequence, in order:
- Name the mechanism. Write one sentence explaining why this event changes what an asset is worth. If you cannot, it is noise by definition — no further analysis needed.
- Look for the second source. Does the filing, the flow data, or the broader market confirm what the headline claims? One source is a story; two independent ones are a pattern.
- Wait for the session to close. Most intraday reversals are noise resolving itself. A move that survives the close and repeats the next session has earned a second look.
- Check whether it changes your thesis. The only headlines that justify a portfolio response are the ones that change the inputs you already relied on — expected cash flows, discount rates, or the rules. A headline that merely changes the mood does not.
- Accept the cost of the filter. You will occasionally be late to a genuine signal. That is the premium paid for not acting on the many false ones. There is no version of this discipline that catches everything.
None of this is a claim that slow is always right. Circuit breakers exist because some moves are fast and real — our explainer on how market circuit breakers work when panic hits describes what happens when the market itself decides a move has crossed a line. The filter's job is narrower: to keep a single headline from steering money that took years to accumulate. We covered a connected angle in How Do Market Circuit Breakers Actually Work When Panic Hits?.
The limits of any noise filter
Honesty requires stating what this framework cannot do. It cannot distinguish a true rumor from a false one in real time; it can only delay your exposure to both. It cannot tell you whether a confirmed signal is already priced in — a well-reported event can be fully reflected in prices before the headline crosses your screen. And it cannot protect against the subtler failure mode: correctly identifying a signal, and then overreacting to it, because a real signal is not a license to abandon a plan.
What the filter does deliver is a defensible record. If you write down, for each headline that tempted a response, whether it passed the mechanism, corroboration, and persistence tests, you will end the year with data about your own process instead of memories about your own nerves. That record is the one signal that never shows up on a screen — and the only one fully under your control.
The evidence, such as it is, points one way: markets produce far more information than any participant can use, and the scarce resource is not access but discrimination. The headlines will keep coming either way. The filter decides how many of them get to matter.
