Daily Stock Screener

Use this skill to turn a watchlist into a ranked shortlist of daily trade candidates.

It is designed for lightweight daily screening, not execution-grade alpha research.

When to use

• User asks to screen a stock list for today's opportunities.
• User wants a ranked shortlist based on trend, valuation, volume, and sentiment.
• A recurring job needs a repeatable daily watchlist brief.

Inputs

Confirm these when the user provides them. If they do not, use the defaults below and state that you did.

• watchlist: list of equities to screen
• market_index: benchmark for context
• pe_range: default 5-25
• rsi_range: default 30-70
• volume_multiplier: default 1.5
• sentiment_threshold: default 0.2
• report_format: default markdown

Safe defaults by market:

• US: benchmark SPY
• growth / tech-heavy: benchmark QQQ
• A-share: benchmark CSI 300
• Hong Kong: benchmark HSI

Tools to use

Primary tools:

1. market_snapshot for price, change, volume, average volume, flow hints, and momentum proxy

- For A-share lists, prefer TickFlow-backed realtime snapshot when configured.

1. market_fundamentals for valuation fields such as PE when available

- For A-share lists, TickFlow currently provides profile/share-cap basics first; PE/PB may still be unavailable and should be labeled explicitly.

1. market_news for recent headlines
2. marketsocialsentiment for crowd tone when relevant

- For A-share and Hong Kong lists, do not assume Reddit-like live coverage; treat social tone as optional and lower-confidence unless a market-native source is present.

1. market_brief for final market summary and scenario framing

Optional enrichment:

• If the user explicitly requires exact RSI, 20MA, or 50MA, load stock-info-explorer and use its local Yahoo-based script.
• If exact indicator coverage is unavailable for a symbol or market, do not invent values. Mark them as unavailable and continue with the remaining filters.

Read [references/screening-playbook.md](references/screening-playbook.md) when you need:

• the default scoring breakdown
• explicit downgrade behavior for missing indicators
• a reusable rejected / borderline classification pattern

Processing pipeline

Run this pipeline in order:

1. Normalize the watchlist and identify the market lane.
2. Fetch snapshot and fundamentals data for every symbol.
3. Apply hard filters:

- PE inside range, if PE exists - volume spike above volume_multiplier - trend / momentum proxies are not clearly bearish

1. Fetch recent headlines for the remaining names.
2. Score sentiment from headlines and social context.
3. Rank surviving names.
4. Generate a short market brief plus top candidates.

Screening rules

Use the user's exact thresholds when provided. Otherwise use the defaults below.

Hard filters

• P/E Ratio: keep names inside 5-25 when PE is available
• Volume Spike: prefer names with volume / avgVolume > 1.5
• Trend Filter: prefer names with positive momentum or clear inflow hints
• Sentiment Filter: prefer names with sentiment score above 0.2

Technical rules

If exact technical indicators are available:

• RSI: keep 30-70
• Trend Filter: price above 50MA
• Momentum: 20MA > 50MA

If exact technical indicators are not available:

• use market_snapshot momentum + flow hints as a proxy
• explicitly label technical precision as limited

Ranking

Use a transparent score. A simple default is:

score =
0.30 * trend_score
+ 0.25 * volume_score
+ 0.25 * sentiment_score
+ 0.20 * valuation_score

Interpretation:

• trend_score: price structure, momentum, inflow hint
• volume_score: volume spike vs average volume
• sentiment_score: headlines plus optional social tone
• valuation_score: PE inside range and not obviously stretched

If a factor is unavailable, reduce its weight and say so.

The more detailed scoring and downgrade notes live in [references/screening-playbook.md](references/screening-playbook.md).

Pairings

• Use market-report if the user wants a full write-up after the screener.
• Use news-intelligence if one candidate has a complex catalyst that needs deeper event analysis.
• Use sentiment-analysis if the user asks for a fuller sentiment breakdown.
• Use stock-data-sourcing if source coverage is uncertain for the target market.
• Use catalyst-tracker for earnings dates and known upcoming catalysts.

Output format

Default to Markdown:

# Daily Stock Screener Report

## Market Context
- Benchmark:
- Session:
- Screening assumptions:

## Top Candidates

### 1. <SYMBOL>
- Score:
- Price:
- PE:
- RSI:
- Volume vs Avg:
- Sentiment:
- Why it passed:
- Risk flags:

### 2. <SYMBOL>
- Score:
- Price:
- PE:
- RSI:
- Volume vs Avg:
- Sentiment:
- Why it passed:
- Risk flags:

## Rejected / Borderline
- <SYMBOL>: reason

## Market Summary
- Breadth / tone:
- Best setups:
- Main risks:

> Disclaimer: MarketBot provides research and analysis only, not financial advice.

If the user asks for JSON, return:

{
  "top_candidates": [],
  "sentiment_scores": {},
  "technical_metrics": {},
  "risk_flags": {},
  "market_summary": ""
}

Rules

• Do not fabricate PE, RSI, moving averages, insider activity, or earnings dates.
• Separate hard facts from model judgment.
• If fewer than two names survive, say that the screen is sparse instead of padding the list.
• When sentiment is mixed or stale, downgrade confidence.
• Keep the final brief concise and ranking-oriented.