Skip navigation
Please use this identifier to cite or link to this item:
Title: Using Treasury Note Option Volatility to Forecast Underlying Price Movements on Non-Farm Payroll Dates
Authors: Snowden, Harrison
Advisors: Almgren, Robert F.
Department: Operations Research and Financial Engineering
Certificate Program: Center for Statistics and Machine Learning
Class Year: 2022
Abstract: Monthly Nonfarm Payroll (NFP) Announcements are high-volatility macroeconomic events generally preceded by significant high-frequency trading activity. This exploratory research project involves forecasting underlying price movements on Non-farm Payroll Announcement dates by applying various machine learning methods to novel feature sets constructed using front-term implied volatility surfaces. Although there already exists considerable research exploring the relationship between the options market and the expectancy of spot price movement following high-volatility events, much of the available literature focuses on equity options and earnings announcements; by contrast, this study examines the interpolated implied volatility surfaces of options on 10-year treasury note futures using robust intraday quote data from Chicago Mercantile Exchange. We find that machine learning algorithms trained on this data demonstrate a limited but non-trivial ability to generate meaningful predictions about the magnitude and, to a lesser extent, the direction of spot price movements. Finally, we conclude that a simple multivariate regression model exhibits the best predictive power and performance on validation data sets. This observation is used to propose potentially profitable volatility trading strategies on NFP dates.
Type of Material: Princeton University Senior Theses
Language: en
Appears in Collections:Operations Research and Financial Engineering, 2000-2022

Files in This Item:
File Description SizeFormat 
SNOWDEN-HARRISON-THESIS.pdf2.66 MBAdobe PDF    Request a copy

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.