Glossary

The principal terms used across the Drusus platform and the Gravenos documentation, defined in plain analytical language.

Markets

ADR: American Depositary Receipt. A negotiable certificate issued by a US bank representing a specified number of shares in a foreign company, traded on US exchanges. Many large Chinese companies maintain both a Hong Kong listing in their underlying shares and a New York listing through an ADR.
Cross-listing: The arrangement under which an instrument is listed on more than one exchange. The principal analytical question raised by cross-listings is the relationship between the prices on each venue, after adjustment for foreign exchange and any structural premium or discount.
Forward rate: In foreign exchange, the rate at which a transaction will be settled at a specified future date, agreed today. The forward rate embeds the interest-rate differential between the two currencies.
HKEX: The Stock Exchange of Hong Kong, operated by Hong Kong Exchanges and Clearing Limited.

Beta: A measure of the volatility of an instrument relative to the market. A beta of 1.0 indicates that the instrument moves in line with the market; greater than 1.0 indicates higher volatility; less than 1.0 indicates lower volatility.
Expected Shortfall: Also known as Conditional Value-at-Risk. The expected loss given that the loss exceeds the Value-at-Risk threshold. Reported alongside VaR because it characterises the loss in the tail rather than only the threshold of the tail.
Monte Carlo simulation: A computational technique that draws repeated random samples from a probability distribution to estimate the distribution of an outcome. In Drusus, the principal application is multi-year portfolio projection under various withdrawal and return scenarios.
Stress test: A scenario analysis in which a portfolio is exposed to a hypothetical or historical adverse market event to estimate the impact. Distinct from Value-at-Risk in that the scenario is specified rather than derived from a probability distribution.
Value-at-Risk: Often abbreviated VaR. A measure of the loss that, with a stated probability, will not be exceeded over a specified horizon. A 95% one-day VaR of £100,000 means that, with 95% probability, the loss over one day will not exceed £100,000. See also Expected Shortfall.

Hallucination: In the language-model literature, the production by a model of a confidently-stated assertion not grounded in its training data or its provided context. The Drusus architecture is designed to constrain the model away from hallucination, principally through the verification layer that rejects unverified content.
Large language model: A class of machine-learning model trained on a large corpus of text to predict the next token in a sequence. The substrate of the modern AI analyst, including the analyst in the Drusus chat interface.
Prompt: The instruction provided to a language model that elicits a response. In the Drusus AI analyst chat, the prompt is the user's query, augmented by a system prompt that constrains the model's behaviour.
System prompt: The instruction provided to a language model that defines its persona, constraints, and operating principles for a given application. In Drusus, the system prompt constrains the AI analyst to verified data and to the analytical register.

Drusus Daily: The twice-daily financial briefing published under the Gravenos masthead, dispatched at 9am and 9pm in each subscriber's local time zone.
Drusus Insights: The long-form editorial archive of the platform, where the founding thesis, methodology notes, and market commentary are published.
Institution tier: The highest paid tier of the Drusus subscription, priced at £299 per month per user. Includes the highest-grade AI model, priority support, and the full analytical suite.
Sample query: An illustrative interaction with the Drusus AI analyst published on each use-case page, showing the form a typical interaction takes for the audience in question.
Strategy Builder: A feature of the Drusus platform, available at the Strategist tier and above, that supports the structural work of repositioning a portfolio. Models the projected outcome of structural changes before they are executed.
Verification layer: The component of the Drusus architecture that rejects any data line lacking complete provenance metadata. The principal defence against the publication of unverified figures.
Watchlist: A user-defined collection of instruments tracked together. Watchlists in Drusus carry full analytical context, not merely price.

FCA: The Financial Conduct Authority. The conduct regulator for financial services firms in the United Kingdom.
MiFID II: The Markets in Financial Instruments Directive (Recast). The principal European regulatory framework governing the provision of investment services. The 2018 implementation introduced the unbundling of research from execution that materially reshaped the independent research market.
Regulatory perimeter: The set of activities for which a financial services firm must be authorised by the relevant regulator. Drusus is positioned outside the FCA regulatory perimeter; it produces analytical commentary, not investment advice within the regulatory meaning.