Heather E. Dempsey, M.S., D.B.A.

Quantitative research, machine learning, algorithmic trading, and financial markets are my passion. I hold a Masters in Computer Science & Information Technology as well as a Doctorate of Business Administration in Finance. I have over 3 years of professional experience in deep learning, predictive regression, statistics, data analysis, and relational database experience.

Skill Set:

Machine learning models:

Model design
Parameter tuning
Cross-validation

Monitoring and evaluation:

Risk management
Performance
Attribution

Predictions:

Risk factors
Prices & returns
Covariance

Portfolio optimization:

Asset allocation
Sector weights
Risk-return profile

Contact Infomation:
203.456.5557
Linkedin
dempseyh_e@yahoo.com
GitHub

Research Papers

Negative Nominal Interest Rates

[+] Abstract...

Predicting Corporate Bankruptcy

[+] Abstract...

This paper presents an empirical investigation which endeavors to answer the question of whether corporate bankruptcy1 can be accurately predicted based on publicly available financial information. In this study, a logistic regression model is specified to forecast the probability of failure three months ahead as a function of publicly available data. The sample includes quarterly financial data from Q12005-Q12015 on [insert number of firms] small, medium and large firms as measured by market capitalization calculated from the Center for Research in Securities and Prices (CRSP)/Compustat Merged Database. Firm-level accounting information from the balance sheet, cash flow, and income statement are taken for the years 2005 to 2015 to predict the probability of failure in 3 months’ time. This analysis provides government regulators, business analysts and researchers a framework through which to quantify the likelihood of corporate failure and intervene to obviate that failure. Cash flow and debt ratios, significant at the 1% level, are the most important factors in the likelihood of a firm’s continued viability. The profitability ratio, quick ratio and size are all significant at the 5% level, proving reliable indicators of the impact on the likelihood of failure with forecast accuracy, overall.

Predicting Corporate Bankruptcy

Pairs Trading with a Kalman Filter

[+] Abstract...

* These papers are in pdf format and require that Adobe Reader be installed to view and print out these papers. If you don't already have the Adobe Reader, you can download it for free by clicking on the button below.

Dissertation & Presentations

Dissertation Defense

[+] Open...

Pairs Trading

[+] Open...

Predicting Corporate Bankruptcy

[+] Open...

Presentation #1

[+] Open...

Residuals

[+] Open...

ROA

[+] Open...

US Stock Returns Forecasts vs. Random Walk Theory

[+] Open...

MATLAB Files

MATLAB

[+] Abstract...

OLS

[+] Abstract...

Test

[+] Abstract...

* These are MATLAB files and require MATLAB. If you don't already have MATLAB, you can download a trial version by clicking here. You'll have the option to purchase a copy or have the ability to access MATLAB online.

These MATLAB files are a work in progress. Feel free to download them for your own use. If you have any questions or technical difficulties, please feel free to contact me.

* These zip files require a zip archiver. If you don't already have a zip archiver, you can download 7-zip for free by clicking on the button below.