Phase 1: Foundation hardening

.clang-format

@@ -0,0 +1,11 @@
+BasedOnStyle: Google
+IndentWidth: 4
+ColumnLimit: 100
+AllowShortFunctionsOnASingleLine: Inline
+AllowShortIfStatementsOnASingleLine: Never
+AllowShortLoopsOnASingleLine: false
+BreakBeforeBraces: Attach
+PointerAlignment: Left
+DerivePointerAlignment: false
+SortIncludes: CaseInsensitive
+IncludeBlocks: Regroup

.github/workflows/ci.yml

@@ -0,0 +1,28 @@
+name: CI
+on:
+  push:
+    branches: [main, "phase-*"]
+  pull_request:
+    branches: [main]
+
+jobs:
+  build-and-test:
+    strategy:
+      matrix:
+        os: [ubuntu-latest, macos-latest]
+    runs-on: ${{ matrix.os }}
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install dependencies (Ubuntu)
+        if: runner.os == 'Linux'
+        run: sudo apt-get update && sudo apt-get install -y libcurl4-openssl-dev
+
+      - name: Configure
+        run: cmake -B build -DCMAKE_BUILD_TYPE=Release
+
+      - name: Build
+        run: cmake --build build --parallel
+
+      - name: Test
+        run: ctest --test-dir build --output-on-failure

.gitignore

@@ -30,3 +30,14 @@
 *.exe
 *.out
 *.app
+
+# Build directories
+build/
+cmake-build-*/
+.cache/
+compile_commands.json
+
+# IDE
+.vscode/
+.idea/
+*.swp

CLAUDE.md

@@ -0,0 +1,72 @@
+# Barcola — Project Conventions
+
+## What This Is
+A C++17 quantitative trading toolkit being transformed from a learning project into a
+quant-grade portfolio piece. The end goal: a fully backtested mean reversion strategy
+on real market data, with verified math, professional engineering, and documented results.
+
+## Architecture
+- `src/core/` — Foundational types: PricePoint, PriceHistory, time utilities
+- `src/indicators/` — Technical indicators: SMA, EMA, RSI, Bollinger Bands
+- `src/analysis/` — Market analysis: mean reversion, correlation, market data analysis
+- `src/risk/` — Risk management: position sizing, slippage modeling, dynamic hedging
+- `src/simulation/` — Monte Carlo simulation
+- `src/backtest/` — (Phase 2) Backtesting engine, strategy interface, performance metrics
+- `tests/` — Google Test suite, one test file per module
+- `data/` — Static CSV files for reproducible testing (never depend on network for tests)
+
+## Build
+```
+cmake -B build -DCMAKE_BUILD_TYPE=Release
+cmake --build build --parallel
+ctest --test-dir build --output-on-failure
+```
+
+## C++ Standards & Idioms
+- **C++17 required.** Use `std::optional`, `std::string_view`, structured bindings,
+  `[[nodiscard]]`, `if constexpr` where appropriate.
+- **No raw `new`/`delete`.** Stack-allocate by default. Use `std::unique_ptr` only when
+  polymorphism or heap allocation is genuinely needed.
+- **Const correctness.** All getters are `const`. Pass by `const&` unless mutation is needed.
+  Prefer `const` local variables.
+- **RAII everywhere.** Resources (CURL handles, file handles) are managed by constructors/
+  destructors or scope guards, never manual cleanup.
+- **No `using namespace std;` in headers.** Allowed in .cpp files at function scope only.
+- **`#pragma once`** for all headers.
+- **Error handling:** Throw `std::runtime_error` or `std::invalid_argument` for unrecoverable
+  errors. Return `std::optional` for expected-empty cases. Never silently swallow failures.
+- **No magic numbers.** Use named constants (`constexpr double RISK_FREE_RATE = 0.05;`).
+
+## Math Correctness (Critical)
+- Every numerical function must handle edge cases: division by zero, empty inputs,
+  insufficient data points, NaN/infinity propagation.
+- All formulas must be verifiable against hand-calculated known values in tests.
+- When in doubt, return a safe sentinel (0.0 for undefined correlation, 100.0 for RSI
+  with no losses) and document the choice.
+
+## Testing Conventions
+- One test file per module: `tests/test_<module>.cpp`
+- Use `EXPECT_DOUBLE_EQ` for exact floating-point matches, `EXPECT_NEAR(a, b, 1e-10)`
+  for computed values with rounding.
+- Every bug fix gets a regression test that would have failed before the fix.
+- Test data comes from CSV files in `data/`, path injected via `BARCOLA_TEST_DATA_DIR`
+  compile definition. Tests must NEVER hit the network.
+- Name tests descriptively: `TEST(RSITest, AllGains_Returns100)`, not `TEST(RSI, Test1)`.
+
+## Git Conventions
+- Feature branches: `phase-N/description` (e.g., `phase-1/foundation-hardening`)
+- Commit messages: imperative mood, concise.
+- Each commit should compile and pass tests.
+
+## What NOT to Do
+- Do not add external trading libraries. Everything is built from scratch.
+- Do not depend on Yahoo Finance API for tests or CI. Use CSV files.
+- Do not optimize prematurely. Correctness first, then clarity, then performance.
+- Do not add features from later phases during Phase 1.
+
+## Roadmap Reference
+Phase 1: Foundation hardening (CMake, tests, CI, bug fixes, CSV support)
+Phase 2: Backtesting engine (strategy interface, backtest engine, performance metrics)
+Phase 3: Run mean reversion backtest + document results
+Phase 4: Second strategy (dual MA crossover) + comparison
+Phase 5: Polish, visualizations, README rewrite

CMakeLists.txt

@@ -0,0 +1,17 @@
+cmake_minimum_required(VERSION 3.16)
+project(Barcola VERSION 0.1.0 LANGUAGES CXX)
+
+set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_CXX_EXTENSIONS OFF)
+set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
+
+find_package(CURL REQUIRED)
+
+add_subdirectory(src)
+
+option(BARCOLA_BUILD_TESTS "Build tests" ON)
+if(BARCOLA_BUILD_TESTS)
+    enable_testing()
+    add_subdirectory(tests)
+endif()