Verify us

Level 3, step by step

Three downloads and one command. You need Node.js 18 or later; the script has zero dependencies and touches nothing but the files you hand it — read it before running it, it fits on one page.

curl -O https://geo-pulse.app/api/proof/dataset.csv
curl -o manifest.json https://geo-pulse.app/api/proof/dataset
curl -O https://geo-pulse.app/api/proof/verify-kit.mjs

sha256sum dataset.csv

node verify-kit.mjs dataset.csv --expect manifest.json

Why prices are pinned (not re-fetched)

The CSV holds the entry and exit prices exactly as the engine saw them, not quotes re-fetched from an external source. A kit that went looking for them would surface differences that have nothing to do with our honesty — revised bars, time zones, a different vendor for the same underlying — and you'd spend your time chasing data mismatches instead of checking arithmetic. The entry price is itself one of the sealed fields: if you doubt one, it sits in the seal's canonical payload, whose hash has been public since emission.

The script rewrites our formulas on purpose

Everywhere else, GeoPulse shares its measurement functions so that no definition drifts between two pages. The kit does the opposite: its formulas are rewritten by hand from the public protocol. A verifier running our own functions would verify nothing. An automated test compares both implementations on every release — but don't take our word for that either: run the script.

Anchoring: proof of precedence

Every night, the day's seals are condensed into a digest, itself anchored on OpenTimestamps (timestamping in the Bitcoin blockchain) and archived on archive.org. Neither we, nor a host, nor a court can rewrite those two anchors after the fact. That's what turns "here are our numbers" into "here are our numbers, and they were already written". Every closed day has its digest, at its date's address:

curl https://geo-pulse.app/api/proof/digest/YYYY-MM-DD

What these checks do NOT prove

• —The kit proves our aggregates follow from our rows. On its own it does not prove the rows predate the outcome: that's what the seals and the anchoring are for. They complement each other; neither suffices alone.
• —Some seals were applied retroactively to older signals (their sealing date is clearly later than their emission, and it shows in the data). Those only prove precedence from their sealing onwards, not from their creation — we don't present them otherwise.
• —None of these checks says the engine will be profitable tomorrow. They say the past numbers are real, complete and untouched. That's a necessary condition for forming an opinion, never a promise of performance.
• —The public track record was reset on 4 July 2026, after an internal audit whose post-mortem is published. Earlier decisions are excluded from public metrics: the counter you're verifying is young, and we'd rather say so than pad it.
• —The engine evolves, and the dataset shows it rather than smoothing it over: you'll find decisions whose confidence sits below today's emission floor (57), simply because they predate its introduction — there was no floor at all before 7 July 2026, then 55, then 57 from 10 July, and historically losing families left the traded flow on the 11th. Some decisions published here predate those tightenings and would therefore not be emitted by today's engine — on 21 July 2026, the day this page went up, that was true of the entire resolved corpus, then just two decisions. We are not removing them and not adjusting any figure: rewriting the corpus to today's threshold would make it prettier and entirely false. The CSV carries each row's date and confidence — you can redo this check yourself, at any later date.

Regulatory disclaimer

GeoPulse is an information and analysis tool. GeoPulse is not a registered financial adviser, is not regulated by the French Autorité des Marchés Financiers (AMF), and provides no personalised investment advice. Nothing on this site is tailored to your situation, objectives or risk tolerance — we don't know them.

The signals, position sizes and example orders shown are educational and generic. The founder's trading journal documents HIS own use of the tool, as a public demonstration; it is neither a recommendation nor an invitation to replicate those positions.

All investing carries a risk of capital loss, up to the entire amount committed. Past performance, real or simulated, does not predict future performance. Before any decision, consult a licensed professional.