Swap-Curve-Forecatsing

Swap Curve and its Application: An interest rate swap refers to the exchange of a floating interest rate for a fixed interest rate where one party will be the payer and the other will be the recipient of the fixed rate. The cash flow of the fixed rate leg of the swap is set when the trade is undertaken. The cash flow for the floating rate leg is set periodically on the rate reset dates, which are determined by the reset period of the floating rate leg. Interest rate payments between fixed and floating rate legs are netted at the end of each payment period and only the difference is exchanged. Interest rate swaps became an essential tool for many types of investors, as well as corporate treasurers, risk managers and banks, because they have so many potential uses which includes:

• Portfolio Management - Interest rate swaps allow portfolio managers to adjust interest rate exposure and offset the risks posed by interest rate volatility
• Speculation - Because swaps require little capital up front, they give fixed income traders a way to speculate on movements in interest rates while potentially avoiding the cost of long and short positions in Treasuries
• Corporate finance - Firms with floating rate liabilities, such as loans linked to LIBOR, can enter into swaps where they pay fixed and receive floating
• Risk management - For Banks, the bulk of fixed and floating interest rate exposures typically cancel each other out, but any remaining interest rate risk can be offset with interest rate swaps
• Rate-locks on bond issuance - When corporations decide to issue fixed-rate bonds, they usually lock in the current interest rate by entering into swap contracts

Question:

Since Swap curve plays important roles in bank’s treasury, it is very important to have a good prediction of it. A dataset consists of Swap rates at daily frequency for time period of 1995-07-14 to 2021-02-28. Different time maturities were laid in the dataset. Below is a snapshot of dataset. Maturities ranged from 1,2,3,6,9 months to 1,2,3,4,5,6,7,8,9,10 years. Every row denotes a swap curve spread across given maturities. Predict Swap Curve for March 19th, 2021 across given set of maturities.

This work propose two methods to forecast Swap Curve:

• 1. Statistical Method
• 2. Autoencoder

It is recommended to first go through the statistical method followed by autoencoders.

It is found that statistical method performs better than the autoencoder. The notebook with statistical method has the granular approach with which each time series should be dealt and notebook with autoencoders has the conclusion with statistical vs autoencoder method inferences