home / skills / luohy15 / financial-skills / trading-agents

trading-agents skill

safe

This skill coordinates multi-agent stock analysis workflows across analysis, debate, trading, and risk phases to deliver structured reports and risk-aware

npx playbooks add skill luohy15/financial-skills --skill trading-agents

Review the files below or copy the command above to add this skill to your agents.

Files (12)

SKILL.md

722 B

---
name: trading-agents
description: Multi-agent stock analysis workflow
author: roy
---

# Trading Agents

Multi-agent stock analysis workflow with 4 phases: Analysis → Research Debate → Trading → Risk.

**Usage:** `/trading-agents NVDA` or `/trading-agents AAPL analysis`

**Phases:**
- Analysis: news, market, fundamentals, social analysts (parallel)
- Research: bull/bear researchers + manager
- Trading: trader proposal
- Risk: risk debate + manager → final decision

**Sub-skills:** `news-analyst`, `market-analyst`, `fundamentals-analyst`, `social-analyst`, `bull-researcher`, `bear-researcher`, `research-manager`, `trader`, `risk-debate`, `risk-manager`

**Output:** `report/<TICKER>/<DATE>/<RANK>/*.md`

Overview

This skill coordinates a multi-agent workflow for stock analysis and trade decisioning. It runs four phases—Analysis, Research Debate, Trading, and Risk—to produce structured trade reports and recommendations. The skill combines specialist sub-agents for fast, parallel data gathering and for deliberative, manager-led decision making.

How this skill works

On command for a ticker (for example: /trading-agents NVDA), parallel analyst agents gather inputs: news, market data, fundamentals, and social sentiment. A research stage then runs opposing bull and bear researcher agents moderated by a research manager to synthesize viewpoints. The trading agent proposes execution ideas, and a risk debate with a risk manager produces a final decision and ranked output files.

When to use it

When you need a systematic, multi-perspective stock analysis for a single ticker.
For producing reproducible trade proposals with documented reasoning and risk review.
To compare bullish and bearish research before committing capital.
When you want consolidated outputs from news, market, fundamentals, and social signals.
For teams that require clear traceability of who/what contributed to a recommendation.

Best practices

Specify the ticker and optional phase (e.g., analysis) to run targeted workflows.
Provide time-context or news cutoffs to avoid stale inputs in analysis.
Review both bull and bear researcher outputs before accepting trade proposals.
Use the risk manager output to adjust position sizing and stop-loss parameters.
Store and version the generated reports for backtesting and audit trails.

Example use cases

A desk analyst runs the full workflow to produce a trade-ready recommendation for NVDA ahead of earnings.
A PM triggers only the Analysis phase to refresh market and sentiment inputs for AAPL.
A compliance reviewer inspects the risk debate summary to validate position limits and controls.
A quant team archives ranked reports to feed supervised models and evaluate analyst consensus.
A trading manager runs the Research Debate to resolve conflicting analyst views before execution.

FAQ

What outputs does the skill produce?

It writes structured report files organized by ticker, date, and rank that capture analyst findings, research debate notes, trade proposals, and risk decisions.

Can I run only part of the workflow?

Yes. You can invoke specific phases (for example: analysis) to run only the desired sub-agents and skip later stages.