The Manifesto

Intelligence, not execution.

The gap in the market

Global crypto exchange volume runs into hundreds of billions of dollars a day. Anyone can send an order. Exchanges have solved execution.

No one has solved, at institutional standards, what should be executed and why. The upstream question — the intelligence question — is where the market breaks down. Fragmented data. Opaque signals. Narrative dressed up as analysis. Backtests that do not survive contact with real markets.

That is the gap. We are here to close it.

What CRYPTYX is

CRYPTYX is a multi-factor, multi-horizon intelligence engine for digital assets.

Two hundred assets today, headed to five hundred. Six hundred and seventy metrics across eight factor classes — trend, volatility, flow, futures, order book, options, correlation, efficiency. One hundred and sixty walk-forward-validated signals in five geometry types. Nine composite conviction scores, six thematic and three regime-aware. Five years of backtested history.

It is architected like a quantitative research platform, not a dashboard tool. Every metric is defined. Every signal is parameterised. Every composite is versioned. Every claim is traceable.

What we believe

Evidence, not narrative.

Every claim on this platform traces to a signal. Every signal traces to a factor. Every factor traces to a metric. Every metric traces to a source. If we cannot show the chain, we do not surface the claim.

Explainability is not optional.

Black-box intelligence is worthless to anyone who has to answer for the trade. We publish the methodology. We publish the walk-forward performance. We publish the regime-aware attribution.

Walk-forward or nothing.

A signal that cannot survive an out-of-sample split does not ship. A composite that overfits the training window does not ship. If it is not real, it does not surface.

Agents are the users.

Autonomous agents are already reading, deciding, and trading. We built for them from day one — MCP tools, x402-priced endpoints, structured JSON, trader-friendly display names. Intelligence has to arrive in a shape agents can act on.

Who we serve

Anyone who wants intelligence directly — self-serve, agent-first, transparent pricing, no sales calls. Bring your workflow, hit our endpoints, compose your conviction.

Autonomous trading agents — MCP tools, x402-priced endpoints, structured JSON, trader-friendly display names. The intelligence layer that makes agentic execution worth the electricity.

Treasuries, funds, and macro PMs — the same catalog, the same backtested profiles, the same institutional-grade sanitisation. Different lens, same source of truth.

The standard

Institutional-grade. Auditable. Every schema versioned. Every proposal reviewable. Every change reversible. Every backtest walk-forward validated. Every metric grounded in five years of continuous history. The platform is the argument.

That is the whole point.