Asking Questions
Tips for context-aware questions, portfolio questions, and market research queries.
On this page (6)
The quality of AI answers improves significantly with better-structured questions. Here are patterns that work well with Worldview's RAG system.
Be specific about the instrument
Mention the ticker when asking about a specific company:
Less effective:
"What's happening with the chip company that makes GPUs?"
More effective:
"What's driving NVDA's recent move? /:NVDA"
The /:NVDA slash command (see Slash Commands) loads NVDA's current price, fundamentals, and recent news directly into the context window — making the answer far more accurate and grounded.
Ask for synthesis, not just facts
The AI is best at synthesizing multiple sources. Ask for synthesis explicitly:
"Synthesize the last week of AAPL news — what are the main themes and what do analysts seem to be watching?"
"Compare the earnings quality of MSFT and GOOGL this quarter — which has better FCF conversion?"
Portfolio-scoped questions
When you want the AI to reference your actual holdings, include that context:
"Given my current portfolio, which of my holdings are most exposed to rising interest rates?"
"What's the biggest macro risk to my tech-heavy portfolio right now?"
The assistant queries your Portfolio service for current holdings when you phrase questions this way. It uses live prices and recent news to give you portfolio-relevant answers.
Portfolio privacy
Portfolio data is only accessed when you ask a portfolio-scoped question in your own authenticated session. Portfolio data is never used to train models or shared with other users.
Time-bounded questions
Specify a time frame when recency matters:
"What has changed for TSLA in the last 48 hours?"
"What were the key macro events in Q1 2026 that affected equity markets?"
The retrieval pipeline filters the document corpus by the time range you specify.
Structured output requests
Ask for specific formats when you want scannable output:
"Give me a bull/bear table for NVDA based on the last month's news and fundamentals."
"List the top 5 risks for AAPL as bullet points with evidence."
What the AI cannot do
- Real-time data — prices in answers are sourced from the last data refresh, not live tick data. Use the instrument chart for live prices.
- Future predictions — the model generates probabilistic synthesis, not forecasts. Treat forward-looking statements as hypotheses.
- Non-covered instruments — instruments not in the EODHD + Finnhub coverage universe may return limited or no context.
- Very recent events — articles published in the last hour may not yet be in the vector store.
Was this page helpful?