Curriculum — Bottie build path

This page is the syllabus: what each chapter covers, how long to budget, what you should produce as evidence of understanding, and how to combine chapters into learning paths.

Chapter index

# Chapter Guided time You can demonstrate
0 New to Python & APIs — keys for Binance, Finnhub, … 2–4 h Python installed; know where API keys are created on provider sites.
1 Overview & architecture 3–4 h A diagram and a spoken walkthrough from HTTP to engine.
2 Environment & first run 4–6 h Clean venv, import render_api, /health 200, smoke exit 0.
3 API & backend services 5–7 h TestClient script hitting 3+ routes; explain CORS.
4 Engine & execution 6–8 h Trace one bar in logs; explain HOLD vs BUY thresholds.
5 ML & ensemble 6–8 h Load or stub a model; change env path; compare logs.
6 Quantum & sentiment 4–5 h Toggle ENABLE_QUANTUM; find sentiment route + env keys.
7 Hardening & operations 5–7 h Security checklist + green pytest on your machine.

Full track: roughly 35–50 hours with Chapter 0 included, or 33–45 hours if you skip Chapter 0 (read + labs + note-taking), not counting optional notebooks you author yourself.

Meta: Note for creators — pricing and donations

When providers change: The tool as-is — adapting when things change (expect URLs and API rules to drift; learn the process, not one frozen screenshot).

Outcomes by chapter (detail)

Chapter 1 — Overview

  • Concepts: clients (dashboard, curl), FastAPI surface, engine loop, ML as a plug-in, external data.
  • Deliverable: your own architecture sketch (Mermaid, Excalidraw, or paper).

Chapter 2 — Environment

  • Concepts: venv isolation, requirements.txt, Windows vs Unix activation, .env hygiene.
  • Deliverable: screenshot or log snippet proving import render_api and scripts/smoke.py success.

Chapter 3 — API

  • Concepts: render_api.py vs api/app.py, routers, TestClient, representative JSON payloads.
  • Deliverable: small Python file or notebook with 3 documented API calls.

Chapter 4 — Engine

  • Concepts: SignalGenerator, ExecutionManager, portfolio targets, paper executor, cooldowns.
  • Deliverable: annotated log transcript for10+ smoke iterations.

Chapter 5 — ML

  • Concepts: predict_proba contract, 3-class vector, SafeModel, optional on-disk model.
  • Deliverable: before/after comparison of signal logs with model present vs absent.

Chapter 6 — Quantum & sentiment

  • Concepts: optional Qiskit stack, classical fallback, Finnhub / sentiment configuration.
  • Deliverable: one-page note on what you enable in production vs dev.

Chapter 7 — Hardening

  • Concepts: secrets, testnet, rate limits, logging, tests, deployment smoke.
  • Deliverable: personal runbook (bullet list) for “start safe session”.

Learning paths

Path A — “Dashboard and API only”

Chapters: 1 → 2 → 3 → 7

Goal: Run and harden the HTTP layer; understand routes the frontend calls; minimal engine depth.

Stretch: Add one read-only diagnostic route and document it in Chapter 3 style.

Path B — “Signals and paper trading”

Chapters: 1 → 2 → 4 → 5 → 7

Goal: Deep engine + ML path; treat API as a thin client unless you need Path A depth.

Stretch: Change feature extraction in signal_generator.py (carefully) and re-run smoke.

Path C — “Full stack”

Chapters: 1 → 7 in order.

Goal: Teach or productize a “build the bot” narrative end-to-end.

Stretch: One new chapter (e.g. “Chapter 8 — Backtesting hook”) using the same folder pattern.

Path D — “Ops and teaching”

Chapters: 1 → 2 → 7 → then 3–6 as reference.

Goal: Prioritize safe operations and documentation for students; implement features only when needed.

How this pairs with Berta Chapters

For each new topic you want Berta to help write:

  1. Create chapters/chapter-NN-slug/README.md with objectives and file pointers (Berta works best when the repo context is explicit).
  2. Ask for notebooks under chapters/chapter-NN-slug/notebooks/ with progressive difficulty (intro → intermediate → stretch).
  3. Add exercises with a solutions/ folder if you sell a solutions pack.
  4. Keep canonical code in the repo root; chapters link to it—avoid duplicating large code blocks that drift.

Attribute generated prose if you publish it: Generated by Berta AI plus human review for accuracy against this repo.

Suggested weekly pace (self-study)

Week Chapters Notes
1 1–2 Solid footing beats rushing.
2 3–4 Heaviest reading; re-run smoke often.
3 5–6 ML + optional features; expect dependency friction.
4 7 + review Tests, checklist, your own appendix chapter.

Next: Getting started or Chapter 1.