Prototype of Pricing Model-Quantitative Research
- Radhin krishna
- Jul 6, 2024
- 4 min read
JPMorgan Chase & Co. - Quantitative Research Job Simulation from Forage
Introduction
I had the opportunity to immerse myself in the Quantitative Research (QR) program at JPMorgan Chase. This journey allowed me to delve into financial modeling, data analytics, and portfolio management.
At the core of JPMorgan Chase, QR stands as a global leader. QR is instrumental in sculpting the financial sector alongside traders, marketers, and risk managers.
In this program, I assumed the role of a QR professional, confronting real-life challenges reflective of their daily operations. My guides were Data Analysis, Python, Derivatives, Critical Thinking, Statistics, Credit, and Dynamic Programming. I engaged in inventory optimization, investigated electronic trading strategies, and refined valuation models, gaining new insights daily that honed my abilities and quickened my adaptability.
In Brief, I was assigned to create 2 prototypes one is for calculating :
1.Prototype of a pricing model For a Natural gas company
2. strategically bucket customers with various FICO scores(estimate the probability of default for a borrower). (In part 2 Article)
Prototype of a pricing model For a Natural gas company
1.Aim
This task in the project aims to develop a methodology for improved pricing of natural gas storage contracts within the commodity trading desk. Current data from external feeds lacks sufficient granularity for accurate pricing. This report outlines an approach to extrapolate existing data while considering seasonal trends, ultimately leading to more precise contract valuation.
2. Project Scope
The project focuses on:
Analyzing historical natural gas price data along with seasonal trends.
Developing a data extrapolation model to generate higher-resolution data points.
Utilizing the enhanced data to estimate future gas prices for accurate storage contract pricing.
3. Methodology
3.1 Data Acquisition
We will acquire historical natural gas price data from reputable external sources. This data will include each point in the data set that corresponds to the purchase price of natural gas at the end of a month, from 31st October 2020 to 30th September 2024.
3.2 Data Analysis
Seasonality Identification: We will employ statistical methods like time series analysis to identify seasonal patterns in historical gas prices. Techniques such as seasonal decomposition (e.g., STL decomposition) will help isolate seasonal components from the overall trend and cyclical fluctuations.
Data Cleaning and Preprocessing: The data will be cleaned to address missing values and outliers. Techniques like imputation and winsorization might be used to address these issues.
The date format was in mm-dd-yy, which is not the standard format for input, so I converted it into the standard format by rearranging the positions and expanding the year into four digits using the datetime module.
3.3 Data Extrapolation Model Development
Several data extrapolation models can be considered:
Data visualization: To identify Time series properties of the pricing it decomposed into individual components and created a line plot
Statistical Methods: Regression analysis with seasonal components can be used to model historical price trends and seasonality. This approach allows for the prediction of future prices based on historical data and identified seasonal patterns.
A primary inference indicates that seasonality, occurring consistently, contributes to the increase in natural gas prices during the winter season. Additionally, there is a trend of constant linear growth in prices over time.
3.4 Model selection
Thus the data shows a linear trend and a steady seasonality i suspect it will fit linear model-polynomial regression ( the internship's standard answer which uses the sine function with linear regression is also provided here but in terms of accuracy polynomial regression is slightly better)
3.5 Model Evaluation and Selection
The performance of various models will be evaluated using metrics like Mean Squared Error (MSE) or Mean Absolute Error (MAE). The model with the lowest error will be selected for further analysis.
3.6 Contract Pricing
The chosen data extrapolation model will estimate future gas prices at any given date. This information and the terms of the storage contract (storage fees, injection/withdrawal dates) will be used to price the contract accurately.
4. Project Deliverables
A well-documented data extrapolation model for natural gas prices.
A prototype system or script that can be used to price storage contracts based on the model's output.
A comprehensive report outlining the methodology, model development, and results.
The retailer can input the date he wants to predict the price and get the result closer to it
6. Conclusion
By enhancing the available data through extrapolation and incorporating seasonality, we can achieve more accurate pricing of natural gas storage contracts. This will provide the trading desk with a valuable tool for informed decision-making when entering into storage contracts.
Implementing the Model
1.Aim
Employing the aforementioned model and considering diverse financial obligations and costs affords the retailer a precise forecast of pricing strategies and the expected profit margin.
2. Key Concept
The core principle of the model is calculating the net profit achievable through a storage contract. This is achieved by subtracting all associated costs from the potential revenue generated by selling gas at a higher price than the purchase price.
3. Cash Flow Considerations
The model takes into account the following cash flows:
Revenue: Derived from selling the stored gas at a future date.
Purchase Cost: The cost of acquiring the gas for storage.
Storage Cost: Fees associated with storing the gas (fixed or variable).
Injection/Withdrawal Costs: Costs incurred during gas injection and withdrawal processes.
4. Model Inputs
The model requires the following user inputs for accurate valuation:
Injection/Withdrawal Dates: Dates for injecting and withdrawing gas from storage.
Purchase/Sale Prices: Prices at which gas is bought and sold on the respective dates.
Storage Costs: Fixed or variable fees associated with storage.
Injection/Withdrawal Costs: Costs per unit of gas injected or withdrawn.
Storage Capacity: Maximum volume of gas that can be stored.
5. Model Output
The model's output is the net value of the storage contract, representing the potential profit after considering all expenses.
6. Benefits
This model offers valuable insights for:
Evaluating the feasibility of storage contracts: By calculating the net value, the model helps determine if a storage contract is potentially profitable.
Comparing contract options: The model allows for a side-by-side comparison of different contract scenarios with varying costs and purchase/sale prices.
Negotiating with clients: The model's output can be used to establish a fair market value for the storage contract during negotiations with clients.
7. Limitations
Assumptions: The model assumes no interest rate fluctuations, market holidays, or delays in transportation.
Price Prediction: The model relies on user-provided purchase and sale prices. Incorporating future price forecasts could further refine the model.
8. Conclusion
This simplified model provides a foundation for evaluating natural gas storage contract values. By considering relevant cash flows and user-defined parameters, the model helps assess the potential profitability of such contracts and aids in informed decision-making.
Comments