How this works
geomon is a personal research tool. Every night it collects upcoming
elections from IFES ElectionGuide,
who holds which office (heads of state and government, cabinets) from
Wikidata,
dated role histories of politically exposed persons (national posts) from
OpenSanctions,
US congressional committee chairs from
congress-legislators,
and a hand-curated list of events such as
Federal Reserve meetings.
Simple hand-written rules then link each event to the markets it might move
(for example "a US election moves S&P 500 volatility").
The Energy & reserves table and the map's Reserves layer show how long
each economy could ride out a supply shock: crude oil stocks against refinery demand
(JODI), the US
Strategic Petroleum Reserve (EIA),
EU emergency oil stocks (Eurostat),
gas storage (GIE AGSI+, EIA) and foreign-exchange reserves
(World Bank).
Where a country doesn't publish a figure, it says so rather than guessing.
It also reads headlines from public news feeds (BBC, Guardian, CNA, Straits Times,
Malay Mail, SCMP, DW, France 24, NPR, the Fed) and GDELT. Articles that name a known politician are sent to an
AI model through OpenRouter,
which lists who did what to whom. A link is kept only if the article says it word for word
and both people are already known; these are marked llm.
It refreshes once a night. It runs on Fly.io in Singapore with its database on
Neon. Visitors are not
tracked and nothing about them is stored; the only thing sent to the AI model is the
text of public news articles.
Spending on the model is capped each month. It does not trade, give
investment advice, or update in real time. Links marked rule are
inferred from those rules rather than stated by a source, and the percentage shows how
confident the rule is.
Data freshness
PEP data © OpenSanctions, licensed
CC BY-NC 4.0
— non-commercial use only.
Version
