Weekly market forecasts, sealed and scored
TROY · Siege Desk
Built on proof
A weekly forecasting desk for any market: a seven-day outlook with its fan of outcomes, every call sealed before the week begins and scored when it ends.
Evidence
Watch the demo
Three-minute demo film with English subtitles. Music by LNV.
How it works
Forecasts are built walk-forward with a purge, so no training row can see the future, and intervals come from the empirical distribution of past out-of-sample errors rather than an assumed curve. Every number stands next to its baselines (random walk, climatology and uniform); when the model does not beat the random walk, the desk says so. Calls are pre-registered in a SHA-256 hash chain and anchored in the Bitcoin blockchain with OpenTimestamps, then scored automatically when their window closes. Nothing is ever edited. The desk is honest about itself: over 303 weeks its model has not beaten the random walk, and it says so on screen. It rests on five published papers and the original research code, run unchanged. News and the implied-volatility gauge follow the market chosen.
Inside
- This Week's Forecast
- Compare Two Markets
- Market Scan
- Support & Resistance
- Seal a Call
- Track record and PDF reports (EN, EL)
In the same family
Ready for pilotsNavigator
Global markets and the official voices that move them, on one screen: 1,247 price series beside what central banks, ministries and regulators have just said.
Learn more →Greek Stock Exchange
A market console for the Athens exchange that shows what is measurable and nothing else: the six headline indices, every sector and every instrument against its own history and against the Composite.
Learn more →Synedrion
The Investment Council: a council of statistical methods that debates your own data. It forecasts, builds portfolios and then tries to prove itself wrong before it lets you trust a result.
Learn more →
Ready for pilotsChresmos
Explainable forecasting for series whose cycles drift, built on Locally-Adaptive Karhunen–Loève forecasting (LA-KLF), with its reasoning shown at every step.
Learn more →See it on your own data.
We run pilots with banks, funds, energy and shipping firms, public bodies and universities.