Metric Risk — FX engine that scores itself - Sell Project

What it is A finished .NET 10 platform that reads the daily state of FX and metals markets and publishes a structured report: regime, liquidity, volatility, risk, cross-asset correlation, statistical context. Engine, REST API, subscriber dashboard, public site, Telegram publishing, and a human approval gate before anything goes out. What makes it unusual It scores itself. Every published reading declares a 95% risk boundary, and every one is graded afterwards against what the price actually did. That record is public, not behind a login: EUR/USD — boundary breached 5.3% of the time against a declared 5%, over 4,975 scored readings Gold — 5.6% over 3,922 Both pass Kupiec and Christoffersen, the tests a risk desk would apply. Check it yourself before you read anything else here: https://metricrisks.github.io/metric-risk/calibration-en.html The number is slightly worse than target, and that is the point. A rate that landed exactly on 5.0% would be the suspicious one. Nobody tuned it afterwards. Chart-pattern detection is table stakes in this sector — every provider ships it. None of them publishes a scored track record against a declared target. No trade signals anywhere in the output. It describes state and publishes uncertainty, which is a design decision enforced in code — and which is why the output is built to reduce signal-related compliance exposure for regulated distributors. What you get 565 files (531 C# + 34 Razor), 70,791 lines — of which 48,263 are code, 10,833 are comments and documentation, and 11,695 are blank — .NET 10, layered architecture with no domain dependency on infrastructure 1,248 tests green, zero warnings — 20,470 lines of test against 50,321 of source 107 markdown files of documentation, including a written justification for each of 166 engine constants. 38 are marked as having none, and that count is published rather than hidden 5,042 daily EUR/USD bars back to 2007, gold and silver back to 2011, plus cross-asset and US macro series Blazor web app: landing, pricing, docs, FAQ, status, changelog, sign-up, login, password reset, and privacy / terms / refund pages Subscriber dashboard: reports, on-demand generation, review queue, alerts, API keys, billing, subscription REST API: documented endpoints, API-key auth, per-plan quotas, OpenAPI spec Three plan tiers built and enforced — an Observer account is served a delayed report and refused the institutional endpoint with a 403, verified end to end Stripe is built, not stubbed: hosted Checkout, hosted Billing Portal, and a signature-verified webhook covering activation, status changes and cancellation, behind a provider interface so another PSP can be swapped in. Card data never touches the platform Logo and icon set 395 commits, single author, no third-party contributors What is honestly missing No customers. No revenue. This is an asset sale, not a business with cash flow. The billing code has never processed a real transaction. It is complete and covered by tests, and activates the moment you supply your own Stripe keys and price IDs — until then the dashboard detects that and routes upgrades to manual activation, and says so to the user. The Blazor app is not deployed. It runs locally; the public link is the evidence — the calibration record, the archive and the documentation. Silver runs on the engine but is not scored yet — its feed carries no intraday range, so several readings are withheld for it rather than estimated. The probability cone is the weakest surface and the report flags it. EUR/USD's 20-day lower edge runs 6.3% against 5%; gold's upper edges run worse. The daily VaR — the headline — passes on both. Three sections ship disabled: fundamental analysis, the event-driven trigger, and parts of the economic calendar. All three are configuration values rather than missing code.