THE METHOD

How the read is made.

Sources go in, one read comes out. This page covers what happens in between: how a headline becomes a score, how that score firms up, how it becomes the number next to your chart, and how the calendar flags a scheduled release before it lands.

Two rules govern all of it: every score is made once, on our servers, so everyone sees the same read at the same moment; and headlines are never held back — you can read a story the second it crosses the feed, before we've scored it.

The method is the product. All of it is on this page.EVERY FIGURE BELOW IS AN EXAMPLE · NEVER A MARKET PRICE

Every tier gets the same sources, the same speed, all 42 contracts. Compare tiers →

THE MAP

From live sources to one read

The eight steps below open up each stage of it.

The calendar (step 5) and the squawk (step 6) work beside this pipeline, not inside it.

STEP 1 · THE SOURCES

It starts with the sources

NewsBerri monitors live news feeds 24 hours a day, five days a week. The moment a headline crosses any of them, it shows up on your tape marked provisional. There's no queue and no hold for scoring; you can read the story before we have.

Provisional means just what it says: we see the story, we haven't read it yet. The row already carries its timestamp and its source, with a PROV chip where the score is about to land. Most headlines wear the label for well under a minute.

We use several sources instead of one big one for two reasons. First, coverage — stories break in different places. Second, confirmation: you can't confirm a story with a single source (that's step 3).

LIVE FEEDSIMULATED

—

Export corridor talks break off early — delegation heads home

08:31:04 ETSOURCE 3PROV · AWAITING SCORE

—

Finance ministers call an unscheduled meeting on bank rules

08:31:02 ETSOURCE 1PROV · AWAITING SCORE

STEP 2 · THE SCORING STEP

An AI reads the story

Next, an LLM reads the whole story, not just keywords, and scores it for each contract it touches. Three judgments, every time: direction, on a −100 to +100 sentiment scale; impact, how much this kind of story tends to matter; and confidence, how sure the read is. The same headline can be bad news for the Nasdaq and fine news for gold, so every contract gets its own line.

Each score is made once, on our servers, and then sent to everyone at the same time. It isn't recomputed per person, tilted toward your positions, or softened to keep you comfortable. Everyone looking at NQ sees the same read at the same moment.

A score only summarizes, so the reasoning behind it stays attached. Tap ✦ Explain on any scored story and you get the why: what the story says, why it scored the way it did, and what that means for the contract you're looking at.

A SCORED ROW, LABELEDSIMULATED

74

Export corridor talks break off early — delegation heads home

08:31:41 ET1ZW2SENT −523IMP 744CONF 715SCORED6

1 — WHEN. THE MOMENT THE SCORE LANDED ON THE ROW.
2 — THE CONTRACT. SCORED PER CONTRACT, NOT PER STORY.
3 — DIRECTION. SENTIMENT, −100 TO +100. NEVER A PRICE.
4 — IMPACT. HOW MUCH THIS KIND OF STORY TENDS TO MATTER.
5 — CONFIDENCE. HOW SURE THE READ IS.
6 — STAGE. PROVISIONAL → SCORED → CORROBORATED.

THE SAME WHEAT STORY, 37 SECONDS LATEREvery part of the read is on the row, not hidden in a tooltip.

STEP 3 · A SECOND SOURCE

A second source makes it solid

A story from a single source is just one source's report. When a second source carries the same story, that's about as confirmed as a fast tape gets. So when a second source confirms a headline, we don't print it again — the row you already have firms up in place, and its label changes to corroborated.

That means every headline on the tape sits at one of three stages, and the label always tells you which: provisional (“we see it, we haven't read it yet”), scored (“we've read it — here's what we think it means”), and corroborated (“a second source agrees it happened”). The stage label never claims more certainty than the story has.

When a story never picks up a second source, it stays visibly single-sourced and its confidence stays modest. It's scored, but never corroborated.

ONE STORY, THREE STAGESSIMULATED

—

Export corridor talks break off early — delegation heads home

08:31:04 ETSOURCE 3PROV · AWAITING SCORE

74

Export corridor talks break off early — delegation heads home

08:31:41 ETSCOREDZW −52

74

Export corridor talks break off early — delegation heads home

08:33:12 ETCORROB ×2 · SOURCE 1 CONFIRMSZW −57

SHOWN AS THREE ROWS SO YOU CAN SEE THE LABELSOn your tape this is one row that firms up in place.

STEP 4 · ONE NUMBER PER CONTRACT

Everything rolls up into one number

Individual stories come and go; the index is the running mood. For every contract, the recent scored stories roll up into a single number, weighted by freshness — a story from five minutes ago counts a lot, one from fifty minutes ago barely registers. The result answers a simple question: what does the news look like for this contract, right now?

That happens for 42 contracts across 7 asset classes — equity index, rates, energy, metals, agriculture, FX and crypto — with sector and asset-class rollups above them, and every plan gets all of it. When a contract's number turns decisive, it gets a plain-words label like STRONG BEAR — and from that moment the call sits on the public track record (step 7), where it'll be resolved against whatever actually happens next.

Every number on this page is sentiment, on a −100 to +100 scale. None of it is ever a market price, and none of it ever tells you what to do.

STEP 5 · THE CALENDAR & GUARDRAIL

The calendar tracks releases in advance

Everything so far reacts to news that already happened. The calendar works ahead of it. Every scheduled release that touches your contracts is on it — CPI, auctions, jobless claims — and for each one, the what-ifs are written in advance: what a stronger-than-expected number would likely mean for your contract, what a weaker one would. When the release lands, the read for it is already written.

Then the Guardrail gives you a heads-up three times on the way in. ARMED means the release is 30 minutes out and the countdown has started. HOT at ten means it's getting close. LIVE at two means it's about to happen. Each step brings a message on screen, a notification, and a spoken callout, so you get the heads-up even if you're looking at another contract. And if your prop firm has rules about trading through releases, Topstep's and Apex's are on screen next to the countdown, so following them on time is straightforward.

The chip carries one small piece of history too, always in the same plain format — HIST ±0.1% TYP 15M (N=15): how this release has typically behaved, for how long, across how many past occurrences. It's the only backward-looking number we allow ourselves, and it's there so you can decide how much to care before the countdown runs out.

CPI · 08:30 ETHOT · T−09:42

HIST ±0.1% TYP 15M (N=15) — HOW THIS RELEASE HAS TYPICALLY BEHAVED

TOPSTEP & APEX EVENT RULES · ON SCREEN, NEXT TO THE COUNTDOWN

STEP 6 · THE SQUAWK

The squawk says it out loud

Trading floors have kept a squawk box on for decades — a voice reading the news out loud so nobody has to stare at the tape. NewsBerri builds that voice into the terminal: headlines above your chosen importance level are read aloud as they land and as they firm up, in one calm sentence that includes the read — what happened, and what it means for the contract.

The bar is yours to set. Keep it high and the squawk only speaks when something matters; open it up and it quietly narrates the working day. Either way, you'll hear the headline even when your eyes are on your trading platform.

The Guardrail uses the same voice for its two-minute callout. Between them, the calendar and the tape both reach you out loud, not only on screen.

“Gold — central-bank reserve buying picks up. Read on GC: sentiment plus forty-seven, corroborated, two sources.”
— THE SQUAWK, READING A SCORED ROW ALOUD (EXAMPLE)

STEP 7 · THE TRACK RECORD

We keep score in public

A private track record is easy to flatter, so ours is public. Whenever the read turns decisive enough to earn a label, that call is logged publicly the moment it's made — contract, label, timestamp — and later resolved against what actually happened next.

Misses stay on the record next to the hits. Nothing is quietly deleted or reworded, and no miss gets a kinder result after the fact. The record stays qualitative on purpose: we don't publish an accuracy percentage about ourselves, because a self-graded number isn't proof of anything. Read the ledger and make up your own mind.

The ledger is public from the very first call. You can read all of it and check any call against what actually happened.

FROM THE PUBLIC LEDGER — RECENT CALLS
Logged	Call	How it resolved
JUL 11 · 09:12	6E · STRONG BULL	Matched what followed
JUL 10 · 14:47	ZN · STRONG BEAR	Missed — kept on the record anyway
JUL 10 · 08:34	CL · STRONG BEAR	Matched what followed
JUL 09 · 10:03	GC · BULL	Matched what followed

STEP 8 · WHAT WE DON'T DO

What we deliberately don't do

The features we've chosen not to build, and how the system behaves on a bad day.

You won't find here

No market prices, no charts, no candlesticks. Your trading platform already handles price. We sit beside it and handle the meaning; we don't repeat its job.
No buy or sell calls. We tell you what a story means for a contract. What you do about it is yours.
No promised outcomes. The Guardrail gives you a heads-up before every scheduled release and puts your firm's rules on screen. It helps you act on time; it can't tell you how that turns out.
No fake social proof. No invented testimonials, no decorative percentages. The public track record above is the only scoreboard we keep.

If something breaks

Headlines keep flowing. If one source stalls, the others keep going. If the scoring itself has a bad moment, stories still arrive — marked provisional, exactly like step 1.
You'll see the problem. A stretch with more PROV chips than usual is the system showing you what it knows, not hiding a fault.
Nothing is back-filled. When scoring catches up, labels advance in place — provisional to scored to corroborated. They're never quietly rewritten to look like we knew all along.

THE METHOD, RUNNING

See it run

The demo is the full terminal on simulated data — click around, open the explanations, hear the squawk speak. Or take the real thing for seven days and judge the method by what it shows you.

START YOUR 7-DAY TRIAL TRY THE DEMO