diff --git a/docs/plans/2026-02-22-sprint-plan.md b/docs/plans/2026-02-22-sprint-plan.md
new file mode 100644
index 0000000..d05111c
--- /dev/null
+++ b/docs/plans/2026-02-22-sprint-plan.md
@@ -0,0 +1,134 @@
+# Trading Bot — Sprint Plan
+
+## Sprint 1: Foundation
+**Goal:** Working project skeleton with database, shared libraries, and infrastructure containers.
+
+**Acceptance Criteria:**
+- [ ] `docker compose up -d postgres redis` boots healthy containers
+- [ ] `python -m alembic upgrade head` applies all migrations cleanly
+- [ ] `python -m pytest tests/` — all unit tests pass
+- [ ] Shared libraries importable: `shared.config`, `shared.redis_streams`, `shared.telemetry`, `shared.models`, `shared.schemas`, `shared.db`
+
+**Deliverables:**
+- `pyproject.toml` with all dependency groups
+- `docker-compose.yml` with postgres+timescaledb, redis, ollama
+- `.env.example`, `.gitignore`
+- `shared/` package: config, redis_streams, telemetry, db
+- `shared/models/` — all SQLAlchemy models (trading, news, learning, auth, timeseries)
+- `shared/schemas/` — all Pydantic schemas
+- `alembic/` — initial migration with hypertables
+- Unit tests for redis_streams, models, schemas
+
+**Tasks (Plan Tasks 1-4):**
+- Task 1: Python monorepo setup + shared utilities
+- Task 2: Docker Compose infrastructure
+- Task 3: Database models + Alembic migrations
+- Task 4: Pydantic schemas
+
+---
+
+## Sprint 2: Data Pipeline
+**Goal:** News articles flow from RSS/Reddit through sentiment analysis into scored Redis stream.
+
+**Acceptance Criteria:**
+- [ ] News fetcher polls RSS feeds and publishes `RawArticle` to `news:raw` stream
+- [ ] Sentiment analyzer consumes `news:raw`, scores with FinBERT, publishes to `news:scored`
+- [ ] Broker abstraction interface defined, Alpaca implementation can fetch account/positions
+- [ ] All unit tests pass
+
+**Deliverables:**
+- `shared/broker/` — abstract interface + Alpaca implementation
+- `services/news-fetcher/` — RSS + Reddit sources
+- `services/sentiment-analyzer/` — FinBERT + Ollama fallback + ticker extraction
+- Unit tests for broker, news fetcher, sentiment analyzer
+
+**Tasks (Plan Tasks 5-7):**
+- Task 5: Brokerage abstraction layer
+- Task 6: News fetcher service
+- Task 7: Sentiment analyzer service
+
+---
+
+## Sprint 3: Trading Core
+**Goal:** System generates trade signals from market data + sentiment and executes paper trades.
+
+**Acceptance Criteria:**
+- [ ] 3 strategies (momentum, mean reversion, news-driven) produce correct signals for known inputs
+- [ ] Signal generator combines strategies via weighted ensemble, publishes to `signals:generated`
+- [ ] Trade executor applies risk checks, submits orders via broker, records trades in DB
+- [ ] All unit tests pass
+
+**Deliverables:**
+- `shared/strategies/` — base interface + 3 strategy implementations
+- `services/signal-generator/` — market data consumer + weighted ensemble
+- `services/trade-executor/` — risk manager + execution loop
+- Unit tests for strategies, signal generator, trade executor
+
+**Tasks (Plan Tasks 8-10):**
+- Task 8: Strategy implementations
+- Task 9: Signal generator service
+- Task 10: Trade executor service
+
+---
+
+## Sprint 4: Intelligence
+**Goal:** System learns from trade outcomes and can replay historical data for backtesting.
+
+**Acceptance Criteria:**
+- [ ] Learning engine computes P&L, adjusts strategy weights respecting all guardrails
+- [ ] Backtester replays historical data through same strategies, produces equity curves + metrics
+- [ ] Weight adjustments logged in `learning_adjustments` table
+- [ ] All unit tests pass
+
+**Deliverables:**
+- `services/learning-engine/` — evaluator + weight adjuster
+- `backtester/` — engine, simulated broker, data loader, metrics
+- Unit tests for learning engine, backtester
+
+**Tasks (Plan Tasks 11-12):**
+- Task 11: Learning engine service
+- Task 12: Backtesting engine
+
+---
+
+## Sprint 5: API & Dashboard
+**Goal:** Web interface for monitoring and controlling the trading bot.
+
+**Acceptance Criteria:**
+- [ ] Passkey registration and login works end-to-end
+- [ ] All API endpoints return correct data (portfolio, trades, signals, strategies, news)
+- [ ] Control endpoints pause/resume trading, force-close positions
+- [ ] Dashboard renders all 5 views with real-time WebSocket updates
+- [ ] All tests pass (backend + frontend)
+
+**Deliverables:**
+- `services/api-gateway/` — FastAPI with passkey auth, all routes, WebSocket
+- `dashboard/` — React app with auth, portfolio, trades, strategies, news, backtest views
+- Unit tests for API routes, auth flow
+
+**Tasks (Plan Tasks 13-16):**
+- Task 13: API Gateway — Auth
+- Task 14: API Gateway — Trading endpoints
+- Task 15: Dashboard — Setup & Auth
+- Task 16: Dashboard — Trading views
+
+---
+
+## Sprint 6: Integration & Deployment
+**Goal:** Full system runs in Docker Compose, passes integration tests.
+
+**Acceptance Criteria:**
+- [ ] `docker compose up` starts all services, all health checks pass
+- [ ] Smoke test script passes (hits key endpoints, verifies responses)
+- [ ] Integration tests pass (news pipeline end-to-end, trading flow end-to-end)
+- [ ] Default strategies seeded in database
+
+**Deliverables:**
+- `docker/` — Dockerfiles for all services
+- Updated `docker-compose.yml` with all application services
+- `scripts/smoke-test.sh`, `scripts/seed_strategies.py`
+- `tests/integration/` — pipeline and trading flow tests
+
+**Tasks (Plan Tasks 17-18):**
+- Task 17: Dockerfiles & full Docker Compose
+- Task 18: Integration testing & seed data