PythonFinanceAI

KO | EN

gitlite — search

#javascript #python #ai #llm #go #golang #machine-learning #java #windows #cli #vue #awesome-list #pytorch #open-source

PythonFinanceAI

(★ 20)

A collection of Python notebooks demonstrating the integration of AI with financial strategies.

PythonFinanceAI 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# PythonFinanceAI

**Reproducible experiments in machine learning and quantitative finance** — factor modeling, portfolio optimization, time-series econometrics, and systematic trading strategies, implemented in Python with real market data.

Every notebook is standalone and runs end to end: data is pulled live via `yfinance` (no API keys required), methods are implemented explicitly rather than hidden behind wrappers, and results are visualized and discussed inline. All notebooks ship fully executed, so results and charts are readable directly on GitHub.

## Methodological standards

Every notebook follows the same contract:

- **Stated objective and approach** — a header explains what is being tested, how, and why it matters, before any code runs.
- **Documented methodology** — estimators, objective functions and test statistics are written out (with equations where it helps), including annualization conventions and the exact meaning of each reported metric.
- **No look-ahead** — signals, weights and model fits use only information available at decision time; train/test separations are explicit.
- **Honest statistics** — known pitfalls are addressed in the text where they arise: multiple testing when scanning lags or pairs, estimated-parameter bias in goodness-of-fit tests, overlapping-window dependence in horizon analyses, and estimation error in optimized portfolios.
- **Stated limitations** — each notebook closes with a conclusions section that separates what the evidence supports from what it does not (costs, slippage, sample size, survivorship).

---

## Highlights

| Notebook | What it demonstrates |
|---|---|
| [04 — Alpha Research & Factor Modeling](04_AlphaResearch_FactorModeling.ipynb) | End-to-end factor research pipeline: return decomposition, factor construction, PCA-based risk models with scikit-learn, and portfolio construction via convex optimization (`cvxpy`). The deepest notebook in the repo. |
| [03 — Smart Beta & Portfolio Optimization](03_SmartBeta_PortfolioOptimization.ipynb) | Smart-beta weighting schemes vs. cap-weighted benchmarks, with tracking-error control formulated and solved as a convex program. |
| [10 — Cointegration Across Multiple Asset Pairs](10_Cointegration_Multiple_Asset_Pairs.ipynb) | Systematic scan for cointegrated pairs (Engle-Granger), spread construction, and stationarity diagnostics with `statsmodels`. |
| [05 — Portfolio Optimization: Theory vs. Reality](05_PortfolioOptimization_Theory_Reality.ipynb) | Mean-variance optimization confronted with estimation error and out-of-sample behavior — where Markowitz breaks in practice and why. |

## Full index

**Systematic strategies & backtesting**
- `01` Momentum strategy on single stocks — signal construction and evaluation
- `02` Breakout strategy — rule-based entries/exits with technical indicators
- `09` Pair trading — spread mean-reversion, entry/exit thresholds, P&L analysis
- `11` Automated detection of support and resistance levels
- `13` Long BTC / Short IBOV — cross-asset relative-value strategy analysis

**Portfolio construction & factor models**
- `03` Smart beta and portfolio optimization (cvxpy)
- `04` Alpha research and factor modeling (scikit-learn + cvxpy)
- `05` Portfolio optimization: theory vs. reality
- `12` Asset rebalancing — turnover, drift, and rebalancing policies

**Time-series econometrics**
- `06` Autocorrelation of financial returns, explained
- `07` Cross-correlation between tech stocks
- `08` Ljung-Box Q-test for serial dependence
- `10` Cointegration across multiple asset pairs

## Stack

Python · pandas · NumPy · SciPy · statsmodels · scikit-learn · cvxpy · yfinance · Matplotlib / Plotly / Seaborn

## Getting started

```bash
git clone https://github.com/hmssantos/PythonFinanceAI.git
cd PythonFinanceAI
pip install -r requirements.txt
jupyter notebook
```

Open any notebook and run all cells — market data is fetched automatically at runtime (notebook 10 downloads its B3 ticker universe from the public [brapi](https://brapi.dev) API when the local CSV is absent). Results will vary slightly with the data window; that is by design, not a bug.

Requirements are pinned where it matters: `pandas` 2.x (several idioms used here changed in pandas 3) and `yfinance` ≥ 1.0 (the notebooks use its current API: `auto_adjust`, `multi_level_index`, explicit `period`).

## Companion articles

Long-form walkthroughs of several of these notebooks are on [Medium](https://medium.com/@hugomichaelisss).

## Author

**Hugo Santos** — Senior Data Scientist working on ML and causal inference for marketing and finance. Building [Aureum](https://aureuminvest.com), an ML-driven portfolio intelligence platform.

[LinkedIn](https://www.linkedin.com/in/hugo-santos-b3449016a) · [GitHub](https://github.com/hmssantos)

## Disclaimer

This repository is for research and educational purposes only. Nothing here is investment advice.

PythonFinanceAI

PythonFinanceAI 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

PythonFinanceAI

PythonFinanceAI 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README