Asset Market Liquidity Risk Management Using Machine Learning Techniques

Asset market liquidity risk is a significant and perplexing subject and though the term market liquidity risk is used quite chronically in academic literature it lacks an unambiguous definition, let alone understanding of the proposed risk measures. To this end, this paper presents a review of contemporary thoughts and attempts vis-à-vis asset market/liquidity risk management. Furthermore, this research focuses on the theoretical aspects of asset liquidity risk and presents critically two reciprocal approaches to measuring market liquidity risk for individual trading securities, and discusses the problems that arise in attempting to quantify asset market liquidity risk at a portfolio level. This paper extends research literature related to the assessment of asset market/liquidity risk by providing a generalized theoretical modeling underpinning that handle, from the same perspective, market and liquidity risks jointly and integrate both risks into a portfolio setting without a commensurate increase of statistical postulations. As such, we argue that market and liquidity risk components are correlated in most cases and can be integrated into one single market/liquidity framework that consists of two interrelated sub-components. The first component is attributed to the impact of adverse price movements, while the second component focuses on the risk of variation in transactions costs due to bid-ask spreads and it attempts to measure the likelihood that it will cost more than expected to liquidate the asset position. We thereafter propose a concrete theoretical foundation and a new modeling framework that attempts to tackle the issue of market/liquidity risk at a portfolio level by combining two asset market/liquidity risk models. The first model is a re-engineered and robust liquidity horizon multiplier that can aid in producing realistic asset market liquidity losses during the unwinding period. The essence of the model is based on the concept of Liquidity-Adjusted Value-at-Risk (L-VaR) framework, and particularly from the perspective of trading portfolios that have both long and short trading positions. Conversely, the second model is related to the transactions cost of liquidation due to bid-ask spreads and includes an improved technique that tackles the issue of bid-ask spread volatility. As such, the model comprises a new approach to contemplating the impact of time-varying volatility of the bid-ask spread and its upshot on the overall asset market/liquidity risk using machine learning techniques