screener-result-analysis

name	screener-result-analysis
description	Analyze TradingView scanner or Screener results from the Rust `tv` CLI as research candidates. Use when explaining why rows matched, what patterns appear, or what to research next after scanner or Screener reads.

Screener Result Analysis

Use this skill when the user asks what scanner or Screener results mean, which rows stand out, or what to do next after a screen.

Core Rule

Treat scanner and Screener rows as research candidates, not buy or sell recommendations. Explain what the screen found and what to research next.

Start With Context

Before interpreting results, identify:

• source: scanner REST, visible full-page Screener rows, hotlist, or chart read; preserve source_category, requires_desktop, and non_mutating when they are present in the payload;
• filters, columns, sort, hotlist slug, or saved screen name;
• result count and whether the output is a full set or only the first rows;
• freshness notes from market-data-interpretation when quote fields are used.

If the task needs field discovery, use tv scanner metainfo --field <FIELD>. If the task moves from scanner rows into chart observation, use docs/observation-workflows.md to choose between tv readiness, tv observe chart, lower-level tv stream ..., and screenshots.

Explain Why Rows Matched

Tie conclusions to returned fields. Common angles:

• valuation: price, market cap, price/earnings, or other requested valuation fields;
• quality: margins, returns, debt, or stability fields when requested;
• growth: revenue, earnings, or performance fields when requested;
• momentum: performance, RSI, trend, or relative-volume fields when requested;
• liquidity and session behavior: volume, premarket, or postmarket fields;
• concentration: sector, industry, exchange, or market-cap clustering.

If a field is missing, say it is missing. Do not fill gaps from memory.

Next Research Steps

Choose follow-up reads based on the question:

• tv compare <SYMBOL>... when several known candidates need Desktop-free quote, info, and fundamentals evidence side by side. Use summary for quick resolution, section-success, field-coverage, coverage-status, and requested-order readback, then inspect raw items[] before comparing candidates. Per-item follow_up_hints name possible next evidence commands; they are not recommendations. Per-item missing_evidence names section gaps and stable follow-up kinds such as snapshot or chart_quote for routing more evidence. Treat snapshot, chart_quote, observe_chart, and screenshot as stable evidence-surface names; do not rename chart_quote to quote_chart.
• tv snapshot <SYMBOL> for a one-symbol Desktop-free evidence packet that combines quote, info, and fundamentals sections before chart observation. Read summary for coverage, missing_evidence[] for follow-up routing, and raw sections for the evidence itself.
• tv info <SYMBOL> for symbol metadata when that is the only needed section.
• tv quote <SYMBOL> or tv quotes <SYMBOL>... for scanner-backed quote checks.
• tv quote <SYMBOL> --source chart when the selected Desktop chart feed matters. Do not treat this as premarket or postmarket evidence; use scanner-backed reads when extended-hours fields matter. Desktop quote-session probes may expose pre/post field names for research, but they are not yet a stable Screener follow-up surface. Even when the quote session reports a post-market phase, do not treat its pre/post close fields as scanner extended_hours unless a later contract explicitly says so.
• tv ohlcv --summary only after switching to a chart when chart bars are needed.
• tv observe chart --duration-ms <MS> --heartbeat-ms <MS> when a bounded readiness-plus-last-bar observation window is more useful than one static chart read.
• tv screenshot --region chart --output <PATH> when visual chart evidence is useful.

Do not mutate Screener filters, columns, screens, or watchlists without user approval.

Output Shape

Prefer a short, educational structure:

1. What the screen asked for.
2. What pattern appeared.
3. Why the notable rows matched.
4. What the data cannot prove.
5. The next 2-4 research steps.

For comparisons or rankings, show the basis for the ranking and call out missing fields. Avoid turning a ranking into a recommendation.