JP Morgan Quantitative Research

Task 1 – Investigate and Analyze Price Data

This repository contains files for Task 1 of the J.P. Morgan Quantitative Research Virtual Job Experience.
The goal is to analyze natural gas monthly price data and build a model to:

• Estimate gas prices on any given date (historical or future).
• Extrapolate prices 12 months into the future for storage contract pricing.
• Identify seasonal patterns in the price data.

---

📊 Problem Statement

Commodity storage contracts depend heavily on future price movements.
We were provided with monthly natural gas prices (Oct 2020 – Sep 2024), and the task was to:

1. Clean and analyze the dataset.
2. Build a function that takes a date as input and returns an estimated price.
3. Extend the forecast for 1 additional year beyond the given dataset.
4. Visualize trends, seasonality, and fitted model.

---

🛠️ Approach

I used a Sinusoidal Regression Model combined with a linear trend to capture:

• Trend: Long-term upward/downward movement of prices.
• Seasonality: Annual demand patterns (higher in winter, lower in summer).

Steps:

1. Converted dates into a time index (t in months).
2. Added sin and cos terms to model yearly seasonality.
3. Fitted a Linear Regression model:
   [ Price = a \cdot t + b \cdot \sin\left(\frac{2 \pi t}{12}\right) + c \cdot \cos\left(\frac{2 \pi t}{12}\right) ]
4. Forecasted 12 months ahead (Oct 2024 – Sep 2025).
5. Built a helper function estimate_price_sinusoidal(date) for interpolation/extrapolation.

---

📈 Results

• The model captures both the upward trend and seasonal variation in natural gas prices.
• Provides daily estimates by mapping any input date into the sinusoidal model.
• Produces a 1-year forecast beyond the dataset.

Example Output:

print(estimate_price_sinusoidal("2024-06-15"))  # interpolation
print(estimate_price_sinusoidal("2025-09-15"))  # extrapolation