How it works

From public data
to a plain-language call.

We don't hold your money and we don't place trades. We read publicly available news and price data on companies that have IPO'd or are about to, and turn that into a call with a confidence score attached — methodology shown, not hidden.

The pipeline

Every prediction goes through
the same four steps.

01
Gather signals
We pull recent news coverage for every company we track, and price history where the market data exists.
02
Score each layer independently
News tone and price momentum are each scored on a bearish-to-bullish scale, with the reasoning kept attached to the number — not thrown away.
03
Combine, don't average
Agreement between layers raises confidence. Disagreement lowers it. We never just split the difference between two conflicting signals.
04
Publish with a band
The result becomes a direction, a confidence score, and a plain-language thesis you can actually read — filed under Accumulate, Reduce, or Watch.
example run
News sentiment layer
Positive earnings coverage
Favorable analyst mentions
Regulatory mention (neutral)
Price momentum layer
5-day trend: up
1-month trend: up
Consistent across windows
Result
Confluence
CONFIDENCE RAISED
Signal layers

Two live today.
More on the roadmap.

We'd rather ship two layers that work honestly than five that don't.

📰
News sentiment LIVE
We scan recent news coverage of each company and read the overall tone. Works for every market we cover, since it doesn't depend on price history existing.
📈
Price momentum LIVE
Where recent price history is available, we look at how the stock has moved recently, weighting newer movement more heavily than older movement.
🏦
Institutional flow ROADMAP
Tracking how large holders adjust positions. Not live — needs a reliable filings data source we haven't wired up yet.
📋
Insider filing activity ROADMAP
Reading insider buy/sell activity from regulatory filings properly means parsing individual transaction codes, not just counting filings. Future work.
Confluence

Why agreement
matters more than volume.

🤝 Layers agree — confidence gets a boost. Independent signals pointing the same way is a stronger case than either alone.
⚖️ Layers disagree — confidence is reduced, not resolved by picking a side. Genuine disagreement is itself useful information.
1️⃣ Only one available — confidence is capped below our top tier. A single signal shouldn't produce a high-conviction call alone.
Confidence bands

Every call lands
in one of three bands.

Accumulate 75–100%
Strong directional signal with multi-layer agreement.
Reduce 75–100%
Multiple bearish behavioral indicators in confluence.
Watch 60–74%
Signal forming — insufficient confluence yet to act.
See it in action

Every active call,
with the reasoning shown.

Filterable by country, sector, direction, and confidence — with the full thesis behind each one.