Files
trading/AGENT_HANDOFF.md

103 lines
3.7 KiB
Markdown
Raw Permalink Normal View History

# Agent Handoff Note
This file is for Claude (or any AI assistant) picking up this project mid-stream.
Read this before touching anything in this repo.
---
## What This Project Is
Bill and his mom Colette are learning to trade together with the goal of passing a prop firm challenge and managing a firm account for income.
Bill is a data scientist with ML and cybersecurity/scripting skills.
Colette is new to trading but actively watches financial news and markets.
They operate as a two-person team:
- **Bill** = quant/engineer — builds models, scripts, backtests, automation
- **Colette** = market observer — spots macro themes, news catalysts, trade ideas
Colette surfaces ideas. Bill validates them with data. They decide together.
---
## Current State
- TradingView installed on Bill's Ubuntu Linux machine (`~/trading/` is the project root)
- Both tracks are in **Phase 1 / Sprint 01** — nothing has been built yet
- No prop firm has been chosen yet (decision due end of Sprint 01)
- No asset class has been decided yet (stocks, futures, or forex)
---
## Directory Structure
```
~/trading/
├── README.md ← project overview
├── ROADMAP.md (in shared/) ← timeline, prop firm options
├── AGENT_HANDOFF.md ← this file
├── shared/
│ ├── ROADMAP.md
│ ├── curriculum/
│ │ ├── README.md ← team model overview
│ │ ├── bill-track.md ← Bill's 5-phase quant curriculum
│ │ └── colette-track.md ← Colette's 5-phase observer curriculum
│ ├── sprints/ ← shared sprint files (currently empty)
│ └── resources/ ← links, PDFs, reference material
├── bill/
│ ├── sprints/sprint-01.md ← data pipeline + Pine Script setup
│ ├── journal/template.md
│ ├── backtests/
│ ├── watchlists/
│ ├── notes/
│ └── scripts/ ← Python scripts live here
└── colette/
├── sprints/sprint-01.md ← chart literacy (TradingView basics)
├── journal/template.md
├── backtests/
├── watchlists/
└── notes/
```
---
## Key Decisions Already Made
- No HFT — target is algorithmic swing/day trading on daily or hourly timeframes
- Paper trading first, then simulated prop challenge passes, then real challenge
- Colette's track avoids heavy math — she communicates ideas in a structured format, Bill backtests them
- Sprint cadence: separate sprints per person, reviewed together
---
## What To Do When Helping
**For Bill:** He can handle Python, ML, Pine Script, scripting, and data pipelines.
Jump straight to implementation — he doesn't need concepts explained from scratch.
Frame suggestions in terms of data quality, edge validation, and risk math.
**For Colette:** She reads this repo too. Keep language clear and non-condescending.
Her intuition about markets and news is an asset, not a liability.
When she asks questions, connect the answer to something she already understands.
---
## Next Steps (as of project init)
1. Bill: complete Sprint 01 — build `fetch_data.py` data pipeline
2. Colette: complete Sprint 01 — TradingView chart literacy exercises
3. Both: decide asset class focus and target prop firm
4. Bill: set up TradingView alerts → webhook pipeline
5. First collaborative milestone: Colette flags a setup, Bill backtests it
---
## Technical Environment
- OS: Ubuntu Linux
- Shell: bash
- Python: available (venv to be set up in `bill/scripts/`)
- TradingView: installed as .deb
- Git remote: Gitea at borgswarm:3000
- SSH key: `~/.ssh/id_ed25519.pub`