amquant 0.1.2

A lightweight library for loading market data and computing engineered features.

AMQuant - Quick Start Guide

Overview

AMQuant is an open-source Python library for quantitative finance that provides:

• Market data loading
• OHLCV time series management
• Feature engineering
• A clean Python API for research, backtesting and quantitative analysis

Unlike the original script-based implementation, AMQuant is now a fully installable Python package available through PyPI.

---

1. Create a Python Virtual Environment

It is recommended to work inside a virtual environment.

Linux / macOS

python3 -m venv .venv
source .venv/bin/activate

Windows

python -m venv .venv
.venv\Scripts\activate

---

2. Install AMQuant

From PyPI

pip install amquant

3. Usage Example

import amquant

# Load market data for an explicit date range.
# fromdate / todate accept "YYYY-MM-DD" strings, date/datetime objects,
# or raw unix seconds.
result = amquant.load_market_data(
    fromdate="2024-01-01",
    todate="2024-06-01",
)

# Loading summary
print(f"Loaded : {result.ok}")
print(f"Failed : {result.failed}")
print(f"Skipped: {result.skipped}")

# Access raw market data
aapl_series = result.raw_series["AAPL"]

# Access engineered features
aapl_features = result.feature_sets["AAPL"]

# Alternatively, if update_globals=True (default)
aapl_series = amquant.RAW_SERIES["AAPL"]
aapl_features = amquant.FEATURE_SETS["AAPL"]

If fromdate / todate are omitted, load_market_data() defaults to a trailing 1-year window (today − 365 days → today).

Downloading a Single Symbol

Every instrument in the universe is fetched through the same download(what, fromdate, todate) call under the hood. You can use it directly if you just want one symbol without going through the full universe/feature pipeline:

from amquant.yahoo_finance import YahooFinanceClient

client = YahooFinanceClient()
series = client.download("AAPL", "2024-01-01", "2024-06-01")

print(series.symbol, len(series.bars))

Returned Object

load_market_data() returns a LoadResult dataclass containing both the loaded data and execution statistics.

Attribute	Description
result.raw_series	Dictionary mapping Symbol → Series for successfully loaded instruments.
result.feature_sets	Dictionary mapping Symbol → FeatureSet containing the engineered features.
result.ok	Number of instruments successfully loaded and featurized.
result.failed	Number of instruments that failed during loading or feature computation.
result.skipped	Number of skipped instruments (missing or unavailable data).

Note: By default (update_globals=True), the loaded data is also mirrored into amquant.RAW_SERIES and amquant.FEATURE_SETS for interactive exploration in a Python REPL or Jupyter notebook. For production code, using result.raw_series and result.feature_sets is recommended.

Contributing & Collaboration

AMQuant is an open-source project and welcomes contributions from developers, quantitative researchers, and Python enthusiasts.

Getting Started

1. Fork the repository

Create your own fork of the AMQuant repository, then clone it locally:

git clone https://github.com/rayenamer/AmQuant
cd amquant

2. create a virtual environment

python3 -m venv venv

source venv/bin/activate

3. install development dependencies

python -m pip install --upgrade pip

pip install -e ".[dev]"

4. Write some good code (please keep the AI slop away)

The Vision Behind AMQuant

AMQuant is designed as the data-engineering foundation of a quantitative research stack.