Early Access · live since April 2026

What works today, and what doesn't yet.

Most products hide this page. Publishing it is the cheapest honest signal available to a young company: if the limitations are written down before you pay, the claims that remain are worth more.

150+

tickers live under continuous read across three sectors

4 / 4

engines in production — News, Technical, Crowd, Fusion

1

engine mid-recalibration — Technical Analysis

Not yet

published accuracy figure — see the calibration note below

Component status

Every part of the system, with its real state.

News Impact engine

Reads every article that crosses our feeds for a covered ticker, ranks them, feeds the top 50 into the score. Spillover routing active.

LIVE

Technical Analysis engine

Running in production and setting base direction. Thresholds are being recalibrated, which is why no hit-rate is published yet.

RECALIBRATING

Crowd Focus engine

X attention and sentiment, measured and displayed. Zero weight in the score by design.

LIVE

Fusion layer

Reconciles all three, writes the synthesis, risk note and levels. Rule set fixed and documented.

LIVE

Dashboard, ticker detail, News, Intelligence Feed

All four screens in production, refreshed on the published schedule with freshness badges throughout.

LIVE

Coverage — 150+ tickers

AI & Tech, Defense and Materials. Expansion to 200+ is the next coverage milestone.

LIVE

Published accuracy metrics

Logging and forward-window tracking run today — the public figure waits for the Technical recalibration to close.

IN BUILD

ML confidence calibration

A learned layer to tune conviction by regime, trained on the logged assessment history.

IN BUILD

Options flow layer

Implied volatility, Greeks and positioning as a new evidence layer. Data foundation in build — scoring ships only after validation.

IN BUILD

Copilot — conversational why

Ask why a Fusion call shifted between two dates and get the layer-by-layer answer in conversation.

IN BUILD

Why there is no accuracy number here

Every assessment is logged, timestamped, versioned against the engine build that made it, and tracked through forward windows against what the market actually did. The measurement infrastructure exists and has been running since launch.

What is missing is a figure worth publishing. The Technical engine — which sets the base direction for every Fusion call — is being recalibrated, so a hit-rate measured against the outgoing build would describe a product that is being replaced.

Reserved for the published figure

—

directional hit rate

—

sample size

—

forward window

Known limitations, stated up front

—

No published hit-rate until Technical recalibration closes.

—

Coverage is 150+ curated tickers — anything outside that universe is only reached through sector spillover.

—

Crowd Focus reads X only — Reddit and StockTwits are not yet ingested.

—

Paywalled publications are not read, so some primary reporting is missed.

—

Technicals refresh once daily after the close, not intraday.

—

No broker connection, no portfolio sync, no execution — and none planned.

Roadmap

Published, dated, and allowed to slip in public.

One engineer ships one thing at a time. These are directional commitments, not contractual dates — and when one moves, this page moves with it.

NOW

IN PROGRESS

Technical engine recalibration

Threshold and regime tuning across the full logged history. Closing this unlocks the published accuracy figure.

H2 2026

PLANNED

Published accuracy, live

The reserved slots on this page get filled: directional hit rate, sample size and forward window — updated continuously on the site, not quoted once as a marketing number.

H2 2026

IN BUILD

Options flow layer

Implied volatility, Greeks and positioning enter the evidence set as a new layer. Data foundation first — scoring ships only after validation.

H1 2027

PLANNED

SEC filings layer

10-K, 10-Q and 8-K filings read and validated into Filing Briefs that ground the News and Technical context for every covered name.

H1 2027

PLANNED

Coverage 200+

Deeper Materials and Defense benches, with subscriber requests weighted first in the queue.

H2 2027

PLANNED

ML confidence calibration

Regime-aware conviction tuning trained on the logged assessment history — on top of the deterministic scoring, never replacing it.

H2 2027

PLANNED

Copilot

Ask why a call moved between two dates and get the layer-by-layer answer in conversation, sourced from the same logged data.

Roadmap items are directional and may change. Nothing here is a guarantee of delivery on a specific date.

You now know the weak spots. Try it anyway.

Seven days, full product, nothing withheld. Read the engines on tickers you already have a view on — that is the only test that settles it.

Start 7-day free trialSee pricing

No charge during the trial · cancel in one click · informational only, not financial advice

Early Access · live since April 2026

What works today, and what doesn't yet.

Most products hide this page. Publishing it is the cheapest honest signal available to a young company: if the limitations are written down before you pay, the claims that remain are worth more.

150+

tickers live under continuous read across three sectors

4 / 4

engines in production — News, Technical, Crowd, Fusion

1

engine mid-recalibration — Technical Analysis

Not yet

published accuracy figure — see the calibration note below

Component status

Every part of the system, with its real state.

News Impact engine

LIVE

Reads every article that crosses our feeds for a covered ticker, ranks them, feeds the top 50 into the score. Spillover routing active.

Technical Analysis engine

RECALIBRATING

Running in production and setting base direction. Thresholds are being recalibrated, which is why no hit-rate is published yet.

Crowd Focus engine

LIVE

X attention and sentiment, measured and displayed. Zero weight in the score by design.

Fusion layer

LIVE

Reconciles all three, writes the synthesis, risk note and levels. Rule set fixed and documented.

Dashboard, ticker detail, News, Intelligence Feed

LIVE

All four screens in production, refreshed on the published schedule with freshness badges throughout.

Coverage — 150+ tickers

LIVE

AI & Tech, Defense and Materials. Expansion to 200+ is the next coverage milestone.

Published accuracy metrics

IN BUILD

Logging and forward-window tracking run today — the public figure waits for the Technical recalibration to close.

ML confidence calibration

IN BUILD

A learned layer to tune conviction by regime, trained on the logged assessment history.

Options flow layer

IN BUILD

Implied volatility, Greeks and positioning as a new evidence layer. Data foundation in build — scoring ships only after validation.

Copilot — conversational why

IN BUILD

Ask why a Fusion call shifted between two dates and get the layer-by-layer answer in conversation.

Why there is no accuracy number here

Every assessment is logged, timestamped, versioned against the engine build that made it, and tracked through forward windows against what the market actually did. The measurement infrastructure exists and has been running since launch.

What is missing is a figure worth publishing. The Technical engine — which sets the base direction for every Fusion call — is being recalibrated, so a hit-rate measured against the outgoing build would describe a product that is being replaced.

Reserved for the published figure

—

directional hit rate

—

sample size

—

forward window

Known limitations, stated up front

  • ·No published hit-rate until Technical recalibration closes.
  • ·Coverage is 150+ curated tickers — anything outside that universe is only reached through sector spillover.
  • ·Crowd Focus reads X only — Reddit and StockTwits are not yet ingested.
  • ·Paywalled publications are not read, so some primary reporting is missed.
  • ·Technicals refresh once daily after the close, not intraday.
  • ·No broker connection, no portfolio sync, no execution — and none planned.

Roadmap

Published, dated, and allowed to slip in public.

One engineer ships one thing at a time. These are directional commitments, not contractual dates — and when one moves, this page moves with it.

NOWIN PROGRESS

Technical engine recalibration

Threshold and regime tuning across the full logged history. Closing this unlocks the published accuracy figure.

H2 2026PLANNED

Published accuracy, live

The reserved slots on this page get filled: directional hit rate, sample size and forward window — updated continuously on the site, not quoted once as a marketing number.

H2 2026IN BUILD

Options flow layer

Implied volatility, Greeks and positioning enter the evidence set as a new layer. Data foundation first — scoring ships only after validation.

H1 2027PLANNED

SEC filings layer

10-K, 10-Q and 8-K filings read and validated into Filing Briefs that ground the News and Technical context for every covered name.

H1 2027PLANNED

Coverage 200+

Deeper Materials and Defense benches, with subscriber requests weighted first in the queue.

H2 2027PLANNED

ML confidence calibration

Regime-aware conviction tuning trained on the logged assessment history — on top of the deterministic scoring, never replacing it.

H2 2027PLANNED

Copilot

Ask why a call moved between two dates and get the layer-by-layer answer in conversation, sourced from the same logged data.

Roadmap items are directional and may change. Nothing here is a guarantee of delivery on a specific date.

You now know the weak spots. Try it anyway.

Seven days, full product, nothing withheld. Read the engines on tickers you already have a view on — that is the only test that settles it.

Start 7-day free trialSee pricing

No charge during the trial · cancel in one click

Informational only · not financial advice