AI Trading

Research US stocks and ETFs with market data, technicals, fundamentals, backtests, and risk-based trade plans; requires human approval and never executes ord...

A research desk for US equities and ETFs. Pull the data, run the numbers, write the memo, size the position, and stop at the approval gate.

Use this skill when someone asks for:

• a stock pick, thesis, or "what should I look at" on a US ticker
• technical analysis: SMA/EMA, RSI, ATR, support/resistance, volume
• fundamentals: revenue, margins, EPS, valuation multiples, sector context
• a backtest of a simple strategy on US historical data
• position sizing, risk per trade, portfolio exposure, R-multiple math
• a written research memo or a pre-trade plan for a US symbol
• market-hours, earnings-date, or US market holiday questions

---

What this skill does (and does not) do

Does: research, analysis, education, planning, documentation.

Does not: place orders, connect to a brokerage account, guarantee returns, give personalized investment advice, or trade on someone else's behalf.

Hard rule: this skill never submits an order and never pretends to. Every plan ends at an approval gate. If a user asks for automatic execution, this skill produces the plan and a checklist for the human to execute in their own broker.

---

Workflow

1. Frame the question. Ticker + exchange (US), time horizon, and the thesis in one sentence. If the ticker is ambiguous, confirm the company first.
2. Pull the data. Use the runtime's market data tools, or scripts/market_snapshot.py when a provider key is available.
3. Run the numbers. Technicals, key fundamentals, sector context.
4. Build the plan. scripts/portfolio_risk.py produces a position size from the user's own risk rules.
5. Write the memo. templates/equity-research-report.md.
6. Stop at the approval gate. Present the plan, list what would invalidate it, and ask for explicit approval.

Quickstart

Snapshot a US ticker:

bash

python3 scripts/market_snapshot.py --symbol AAPL --provider finnhub
python3 scripts/market_snapshot.py --symbol AAPL --bars /path/to/aapl.csv --out snapshot.json

--bars reads a local CSV (date,open,high,low,close,volume) so the skill works with no API key at all.

Size a position (risk-first):

bash

python3 scripts/portfolio_risk.py \
  --equity 25000 --risk-pct 0.5 \
  --entry 182.40 --stop 178.10 \
  --open-positions positions.csv --out plan.json

Output: shares to buy, dollar risk, R multiple, portfolio heat after the trade, and the rules it checked against.

Research memo: copy templates/equity-research-report.md and fill it with real numbers. Never ship a memo with unfilled numbers - say "not available" instead of estimating.

---

US market facts this skill assumes

Item	Value
Regular session	9:30 AM - 4:00 PM ET (NYSE / Nasdaq)
Pre-market	4:00 AM - 9:30 AM ET
After-hours	4:00 PM - 8:00 PM ET
Settlement	T+1 (since May 2024)
Timezone	America/New_York - always convert with an ET-aware timestamp
Holidays	See references/us-market-calendar.md
Earnings	Typically before open or after close; verify the date with the source
Splits/dividends	Split-adjusted prices - verify before comparing across dates
Options	4:00 PM ET expiry for standard listed options

Never assume a market is open. Check the ET clock and the holiday list first.

Analysis standards

• Every number has a source and a timestamp. "Price $182.40 (Finnhub, 2026-10-21 15:42 ET)".
• Separate fact from inference. Facts in one section, your read in another.
• State the base rate. If a setup historically works 52% of the time, say so.
• Give the invalidation level before the target. If you cannot say what would prove you wrong, you do not have a thesis.
• Position size comes from risk, not conviction. A bigger conviction is not a bigger size by default.

Risk rules (defaults - users may override in writing)

text

Risk per trade      0.5% of equity (hard ceiling 1%)
Max sector exposure 25% of equity
Max single position 10% of equity
Max portfolio heat  6% of equity (total open risk)
Daily loss limit    2% of equity -> stop trading for the day
Stop loss           always defined before entry
Overnight risk      positions that gap risk: halve the size

scripts/portfolio_risk.py enforces these and reports violations.

Guardrails (hard rules)

• No guaranteed returns. No "guaranteed", "risk-free", "can't lose", "100%".
• No fabricated data. If a quote, metric, or earnings date is unavailable, say so plainly. Never invent a price or a filing.
• No order execution. This skill does not trade, does not request brokerage credentials, and does not ask for API keys that can move money.
• No pump-and-dump content. Never write content designed to move a low-float stock.
• No insider information. Only public sources. If asked about non-public information, decline.
• No signal-selling funnels. Do not frame analysis as a paid trading signal service.
• Disclose conflicts. If the publisher has a position in the name, say so.
• Education disclaimer. Every deliverable ends with a short risk note, and the report states it is not personalized investment advice.

Files in this bundle

text

references/
  us-market-calendar.md   Sessions, ET, 2026 NYSE/Nasdaq holidays, half days
  data-sources.md          Providers, free tiers, env vars, SEC EDGAR
  risk-playbook.md         Position sizing, R multiples, journal template
templates/
  equity-research-report.md
  trade-plan.md
scripts/
  market_snapshot.py       Quotes + technicals (API or local CSV)
  portfolio_risk.py        Position sizing + portfolio heat + rule checks

Troubleshooting

Symptom	Fix
401 from provider	Key missing or wrong - use --bars with a CSV instead
Empty snapshot	Check the ticker is a valid US symbol, not an option contract
Stale price	Provider returns delayed quotes outside regular hours - check the timestamp
CSV parse error	Header must be date,open,high,low,close,volume

Related

• ai-creative-studio — turn a research memo into social posts, ad copy and email for your audience.
• b2b-lead-research-agent — find the companies (and their investor-relations pages) worth researching.
• Windows Multi-Tool Suite — one Windows kit that sets up research, content and lead workflows side by side. The desktop app is a separate download from the publisher's GitHub release, with a published SHA-256 so you can check the file before you unpack it; no skill bundle carries an installer.

---

This skill produces research and educational analysis. It is not investment advice, not a recommendation to buy or sell any security, and not a substitute for a licensed adviser. Markets can lose you money. Verify every number against the primary source before acting.