🛢️ WTI & Brent Crude Oil — Price Analysis (2015–2026)

A Python/Jupyter analysis of WTI and Brent crude oil prices over a decade, covering price trends, volatility, return correlation, drawdowns, and seasonality. Data is pulled live from Yahoo Finance via yfinance.

---

📊 Outputs

Chart	Description
oil_price_dashboard.png	Price history, Brent–WTI spread, and 90-day annualised volatility
oil_return_correlation.png	Daily return distributions, scatter plot, and rolling 90-day correlation
oil_seasonality.png	Average price by month and by year
oil_drawdown.png	Cumulative indexed returns and drawdown (base = 2015-01-01)

---

🗂️ Project Structure

.
oil-price-analysis/
├── images/
│   ├── oil_price_dashboard.png
│   ├── oil_return_correlation.png
│   ├── oil_seasonality.png
│   └── oil_drawdown.png
├── notebooks/
│   └── oil_pipeline.ipynb
├── README.md
└── requirements.txt

---

⚙️ Setup & Usage

1. Clone the repository

git clone https://github.com/sandwidinarcisse/oil-price-analysis.git
cd oil-price-analysis

2. Create a virtual environment (recommended)

python -m venv .venv
source .venv/bin/activate      # macOS / Linux
.venv\Scripts\activate         # Windows

3. Install dependencies

pip install -r requirements.txt

4. Launch the notebook

jupyter notebook oil_prices.ipynb

Run all cells — the notebook fetches data from Yahoo Finance, processes it, and saves the four charts to disk.

---

🔬 Methodology

Data

• Source: Yahoo Finance via yfinance — tickers BZ=F (Brent) and CL=F (WTI)
• Period: 2015-01-01 → 2026-04-30 (daily frequency)

Cleaning (clean_oil_data)

• DatetimeIndex enforced and sorted
• Weekends / public holidays forward-filled (up to 3 days)
• Rows where both prices are missing are dropped
• Extreme outliers clipped at ±5 × IQR from the median

Feature Engineering (engineer_features)

Feature	Description
Spread	Brent Price − WTI Price
Brent_WTI_Ratio	Brent / WTI
*_Return	Daily percentage return
*_MA30d / *_MA90d	30-day and 90-day rolling mean
*_Vol30d / *_Vol90d	Annualised rolling volatility (σ × √252)
Year / Month	Calendar fields for seasonality

Key Events Annotated

Date	Event
2016-01-20	Oil price crash low
2018-10-03	4-year peak
2020-04-20	WTI goes negative
2022-03-08	Ukraine war price spike
2023-06-04	OPEC+ production cuts
2026-02-28	Iran war

---

📈 Key Findings

• Brent–WTI spread is structurally positive but narrow — Brent trades at a modest premium to WTI in most periods, typically in the low single digits ($/bbl). The spread only blew out dramatically in April 2020 when WTI futures went negative, creating an anomalous spike visible in the spread panel.
• WTI went negative on 20 April 2020 (–$37/bbl due to futures contract mechanics and storage constraints), causing a WTI drawdown of approximately –140% from its 2015 base — far exceeding Brent's worst drawdown of ~–70%. This is the largest spread divergence on record, not a typical COVID demand shock.
• Rolling 90-day correlation averages 0.919, not the commonly cited figure of 0.97. The two benchmarks move tightly together under normal conditions, but the correlation collapsed to ~0.35 in April 2020 — the single sharpest decorrelation in the dataset — precisely when WTI's contract mechanics decoupled from physical Brent pricing.
• WTI annualised volatility briefly exceeded 500% in April–May 2020, compared to Brent's ~100% peak over the same window, confirming the 2020 event was a WTI-specific structural dislocation rather than a symmetric commodity shock.
• 2022 was the cycle peak for both benchmarks, driven by the Ukraine war supply shock (annotated March 2022), with Brent reaching ~$128/bbl. Both benchmarks have trended lower since, settling in the $60–$80 range through 2024–2025.
• A new geopolitical event — "Iran war" — is annotated in early 2026, associated with a sharp price spike visible in the dashboard, bringing Brent back above $100/bbl briefly before retracing. This is the most recent macro shock in the dataset.

---

🛠️ Requirements

Package	Purpose
pandas	Data manipulation
numpy	Numerical computations
matplotlib	All visualisations
yfinance	Yahoo Finance data download
requests	HTTP (yfinance dependency)

Python 3.10+ recommended (developed on 3.12.6).

---

📄 License

MIT — free to use, modify, and distribute with attribution.