Alpha One

An AI-managed, conviction-led U.S. equity strategy with a public, daily-updated record of performance, risk, holdings, benchmark context, and portfolio-manager briefs.

View the live record

The problem

A strategy can be described convincingly without being made legible. Alpha One needed a public site that showed what happened, when the data was updated, how results compared with a benchmark, and what risk accompanied the return.

What I built

A live performance and drawdown record benchmarked to SPY.
Risk statistics, holdings context, market regime, and cash posture.
Dated AI portfolio-manager briefs written from a dedicated export contract.
A static public architecture that does not expose the private strategy engine.

The systems idea

The website and Horus are intentionally separated by a defined file handoff. The private system exports only what the public site is allowed to publish. That boundary makes the system easier to reason about and reduces accidental disclosure.

How it works: the public site updates from a dedicated export and presents the live record, benchmark, holdings, risk, and dated briefs without importing the private engine.

What it taught me

Publishing outcomes changes the build. Data freshness, caveats, comparability, and reproducible updates become product requirements rather than reporting details.

Alpha One is shown as a software and research project. Nothing on this page is investment advice.