Author: Amin Salighehdar
Degree: Ph.D. in Financial Engineering
Year: 2018
Advisory Committee: Dr. Ionut¸ Florescu, Dr. Dragos Bozdog, Dr. Khaldoun Khashanah, Dr. Zhenyu Cui, Dr. Charles Suffel

Abstract: Accurate measurement of any phenomenon is always a challenge for a scientist. This issue becomes critical when a researcher deals with a multi-dimensional phenomenon. In these situations, creating a combined indicator might be helpful in providing more insight into characterizing a phenomenon. The general idea of creating a combined indicator is to properly select individual measurements and combine them by assigning appropriate weights to each one of them. In this dissertation, we study multiple methods to create a combined indicator from separate measurements. We apply our study to analyze two different types of phenomena. The first phenomenon is a storm surge which is a naturally occurring event in the presence of hurricane. We use multiple statistical methodologies to create new ensemble models capable of better forecasting storm surges. The second phenomenon is lack of liquidity in financial markets. Liquidity is an essential feature of a financial market and it is used as an indicator for smooth operation of an economy. Unlike storm surges, liquidity is not directly observable. There are multiple proxies proposed by researchers to quantify liquidity in a financial market. We look into different statistical methods to create a new liquidity index called LIX. We evaluate the performance of the proposed comprehensive indicator. In the storm surge study, our findings show that using a de-biasing technique will generally improve the overall performance and produce more accurate predictions of the storm surge. In the context of financial study, we evaluate the performance of the LIX for several events; the U.S presidential election in 2016, the French presidential election in 2017, Flash Crash on May 6, 2010 and February 5, 2018. Our empirical analysis across multiple events indicate that considering a single measure to fully capture the dynamics of market liquidity is insufficient; similar to the conclusion drawn from the storm surge study. Our commonality analysis reveals that there is a strong correlation between existing liquidity measures. Further, multiple day analysis shows that the relationship between LIX, volatility, and assets return is very relevant during a financial event.