Syllabus

This course is an introduction to empirical methods in high-frequency financial econometrics. Its focus is on understanding the core theory and applying its theorems to observed high-frequency financial data. Students will work with high-frequency data on different financial assets and create weekly project reports. The course is designed for students interested in obtaining a general understanding of high-frequency financial econometrics.

The course is mainly based on lecture notes. However, a more in-depth coverage of the topics we will discuss are available in the books:

• High-frequency Financial Econometrics by Ait-Sahalia and Jacod
• Discretization of Processes by Jacod and Protter

• Lecture Notes
• Papers from the literature (indicated in the lecture notes)

• This course is intended for Duke MA and PhD students.
• Undergraduate econometrics or advanced statistics is required. Specifically, students should be comfortable with the notions of asymptotic approximations.
• Programming skill in Matlab is required.
• Students must have the latest version of Matlab (2019a) installed, which is freely available via Duke OIT services.
• Students using the Windows operating system must have installed Git for Windows.
• Students must have installed the appropriate version of Latex for their operating system (Tex Live for Windows, Mactex for Mac).
• The following software packages can be learned over the semester: Latex, Bash and Git.

• Final grade will be based on weekly projects, a midterm and a final examination.
• Exams can be either 36-hour take home or in-class closed book with a note sheet allowed.
• Grade Division:
  • Projects: 40%
  • Midterm: 30%*
  • Final: 30%
• *If a student misses the midterm for any reason, then its weight is placed on the final examination. If a student attempts the midterm but fails to turn it in, then this student's midterm score is recorded as zero.

• Projects will be assigned on a weekly basis.
• Problem sets are individual. Each student must do the entire problem set. This includes: writing your own code, making your own plots, interpreting the results, and preparing a pdf report with Latex.
• The best way to learn the contents of the course and obtain an excellent grade is to do the hard work yourself.
• Grading of Projects:
  • Projects are due by midnight of the announced due date (usually a week after the project was posted).
  • No late projects are accepted (no exceptions).
  • Grading is done on a 0-10 scale.
  • Projects with excessive overlap with other student's answers will receive a zero grade. Students must uphold the Duke Community Standard.

We will cover the following topics:

• Simulation of jump diffusion processes
• Implied volatility
• Volatility signature plot
• Separating jump returns
• Truncated variance
• Inference for integrated variance
• Realized beta
• Bootstrapping standard errors
• Local variance estimator
• Jump regression
• Variance forecasting with AR, HAR and RQ models
• Black-Scholes options pricing
• Microstructure noise effects on realized variance
• Two-scale realized variance

The exams will take place on the following days:

• Midterm Date: October 1st
• Final Exam Date: November 26th