Home / Recipes / Follow earnings coverage for a listed company
Worked queryFollow earnings coverage for a listed company
/stories in keyword mode. The parameters matter less
than the reasoning behind them, so that is most of what is below.
The request
GET /stories
?q="<TICKER>"
&mode=keyword
&from=2026-07-22
&to=2026-07-29Why these parameters
One of the few cases where keyword mode is the right answer. Tickers are among the very small set of identifiers publishers render consistently across languages and registers, which is exactly what keyword matching needs and almost never gets. Company names are not: the same firm appears as its registered name, a short form, a local-language rendering or a ticker, sometimes several within one article.
The obvious approach, and why it loses
The obvious approach is the company name, and it is worse than it looks in both directions. Common names collide with unrelated entities, and distinctive ones still appear in four or five forms — registered name, short form, local-language rendering, former name after a rebrand — with different outlets choosing differently. Adding the exchange or the sector helps precision and costs recall. The ticker sidesteps all of it for the narrow set of companies that have one, which is exactly why this is the rare recipe where keyword mode wins: you are matching an identifier rather than a description, and identifiers are the one thing keyword search is genuinely good at.
When it returns too much, or too little
Too much usually means the ticker collides with an ordinary word or with another exchange's symbol, and the fix is to add the exchange or the company's short name as a second term rather than to change mode. Too little means the outlets covering this company do not use the ticker in their copy, which is common outside the financial press and almost universal in non-English coverage — fall back to hybrid mode on the registered name and expect to review more. Around results days, tighten the window rather than widening it: the interesting coverage is concentrated in hours, and a fortnight buries it under preview and recap pieces.
How you would know it is working
Check it against the company's own disclosure calendar. For four consecutive quarters, compare when each result was published against when the coverage appeared here, and note anything the query missed entirely. That is your recall on the one event type this recipe exists for, and it is measurable without any judgement calls. Separately, count how many results were about a different company with the same symbol; if it is more than a handful, the ticker is not safe on its own for this listing. Both checks take an hour a quarter and they are the difference between a monitor you can quote and one you assume is working.
When this is the wrong tool
Ticker matching fails for companies with dual listings, recent symbol changes or listings on exchanges whose codes collide with others. It also fails for anything not publicly listed, which excludes most of the interesting private-market activity. Where the ticker is not usable, hybrid mode on the registered name is the fallback and recall will be materially lower. This recipe is also a poor basis for anything resembling a trading signal: coverage follows announcements rather than preceding them, so what you are measuring is the reporting of a fact that the market already has. Treat it as a monitoring and summarisation tool, not as an information edge.
Running it for real
Response fields, failure modes, adaptations and scheduling cost: Follow earnings coverage for a listed company, in the docs.