ARR Investment Partners - Portfolio Error Detection

Repository Link: https://github.com/RaynorEtienn/Financial-Data-Analysis

Introduction for Reviewers

Welcome! This repository contains my solution for the Portfolio Error Detection task. I have taken on the role of a "detective" to uncover inconsistencies and errors in the provided portfolio dataset.

Where to start:

• Main Analysis: The primary entry point is notebooks/analysis.ipynb. This notebook demonstrates the execution of the validation engine, visualizes the identified errors (price spikes, reconciliation breaks, etc.), and provides a summary of the findings.
• Architecture: The solution is designed as a modular Python library located in src/. Each class of error is handled by a dedicated validator in src/validators/, ensuring the code is scalable and easily extensible for new types of checks.
• Quality Assurance: Comprehensive unit tests are provided in tests/ to verify the logic of each validator.

---

Project Overview

This project is designed to analyze a hypothetical portfolio dataset to identify data integrity issues, inconsistencies, and calculation errors. It serves as a technical assessment for the Quantitative Developer role at ARR Investment Partners.

Problem Statement

The input dataset contains daily portfolio positions and transaction details. The goal is to detect errors such as:

• Inconsistent prices (day-to-day jumps).
• Mismatches between holding quantity changes and trade information.
• Incorrect calculated ratios (weights, values).
• Discrepancies between trade prices and holding prices.
• Others?

Architecture

The project follows a modular architecture to ensure scalability and maintainability:

1. Data Ingestion Layer: Responsible for loading raw data (Excel/CSV) and standardizing it into a usable format (e.g., Pandas DataFrame).
2. Validation Engine: A collection of specialized validators, each responsible for a specific type of error check.
   • PriceValidator: Checks for unrealistic price movements.
   • ReconciliationValidator: Reconciles positions with trades.
   • CalculationValidator: Verifies derived metrics (Market Value, Weights).
   • ConsistencyValidator: Checks cross-referencing data (Trade Price vs. Market Price).
   • DataCompletenessValidator: Checks for missing critical data points (Prices, FX rates).
   • FXConsistencyValidator: Checks for inconsistent exchange rates across assets.
   • StaticDataValidator: Checks for consistency in static data (Sector, Currency) over time.
3. Reporting Layer: Aggregates findings from all validators and produces a structured report.
4. Presentation: A Jupyter Notebook demonstrating the usage of the library and highlighting the findings.

Project Structure

.
├── data/               # Input data and generated reports
├── notebooks/          # Jupyter notebooks for analysis and presentation
├── src/                # Source code
│   ├── __init__.py
│   ├── data_loader.py  # Data ingestion
│   ├── validators/     # Error detection logic
│   │   ├── __init__.py
│   │   ├── base.py
│   │   ├── price.py
│   │   ├── quantity.py
│   │   └── ...
│   └── utils.py        # Helper functions
├── tests/              # Unit tests
├── requirements.txt    # Project dependencies
└── README.md           # Project documentation

Setup

1. Clone the repository.
2. Install dependencies:

   pip install -r requirements.txt

3. Place the input data file in the data/ directory.

Testing

To run the unit tests, execute the following command from the project root:

python -m pytest

Usage

Run the analysis via the Jupyter Notebook in notebooks/analysis.ipynb or execute the main script (to be implemented).