[9:00 AM-9:30 AM] Registration and Tea
[9:40 AM-10:30 AM ] Prof Cornelis Oosterlee - Deep time-inconsistent Portfolio Optimization with Stocks and Options
Abstract: We take the perspective of a trader who is allowed to trade in a risk-free bond, in stocks, and also in options, to some extent. The trader is allowed to trade at a set of trading dates. An objective function is used to evaluate whether the investor is satisfied with the return. A high-quality, objective function is able to numerically represent the investor's preferences of how much risk is acceptable for a certain level of potential profit. Moreover, we introduce market friction aspects of incomplete markets and trading constraints. Regarding the market, we add transaction costs and a non-bankruptcy constraint; for the trading strategies, we introduce leverage constraints.
[10:35 AM-11:25 AM] Prof Sankarshan Basu - Applications of Quantitative Techniques in Finance
Abstract: The talk shall cover the applications of various quantitative techniques in the finance – both from an academic point of view as well as the industry-level applications. We shall also delve in to the current buzzword of “FINTECH” and discuss how quantitative techniques and technology as use din Finance for a long time and how some major technological advances have made such usage more commonplace. In particular, we shall be linking the applications with work that have been done in that space – work ranging from bond and option valuation to risk management issues in banks and financial institutions to interconnectedness issues within firms, in particular financial firms, and its resultant impact on the overall market to text analysis and its impact on financial markets. The talk will cover broad theoretical aspects in this space and its applications in practice and highlight some potential areas going forward.
[11:30 AM-12:05 PM] Dr Arjun Beri - An Overview of the Challenges in Counterparty Credit Risk Modelling
Abstract: TBD
[12:10 PM-12:45 PM] Anshul Jain - Leveraging Gen AI for Quantitative Investing
Abstract: TBD
[12:45 PM-02:00 PM] Break
[2:05PM-2:55PM] Prof Srikanth Iyer - Asymmetric Super-Heston-rough Volatility Model with Zumbach effect as a Scaling Limit of Quadratic Hawkes Processes
Abstract: Modeling price variation has always been of interest, from options pricing to risk management. It has been observed that the high- frequency financial market is highly volatile, and the volatility is rough. Moreover, we have the Zumbach effect, which means that past trends in the price process convey important information on future volatility. Microscopic price models based on the univariate quadratic Hawkes (hereafter QHawkes) process can capture the Zumbach effect and the rough volatility behavior at the macroscopic scale. But they fail to capture the asymmetry in the upward and downward movement of the price process. To incorporate asymmetry in price movement at the micro-scale and rough volatility and the Zumbach effect at the macroscale, we introduce the bivariate Modified-QHawkes process for upward and downward price movement. After suitable scaling and shifting, we show that the limit of the price process in the behaves as the so-called Super-Heston-rough model with the Zumbach effect.
[3:00PM-3:35PM] Arjun KM - Facilitating Ethical AI Systems in the Financial Services Industry: Responsible Lending and Beyond
Abstract: Financial institutions are increasingly capitalizing on cutting-edge technologies facilitated by the integration of Machine Learning methodologies. This deployment extends well beyond traditional applications like credit scoring, encompassing diverse fields such as personalized financial planning, recommendation engines, payment systems, Anti Money Laundering (AML) protocols, internal controls, and regulatory compliance measures at a magnitude unattainable by human counterparts. Large-scale automation capabilities and cost savings have made ML algorithms attractive for personal and corporate finance applications. These attributes not only present novel prospects for refining and tailoring customer experience for business but also accentuate the heightened risk of adverse outcomes if these systems are inadequately designed. The rise of algorithmic decision-making has thus spawned much research on fair machine learning (ML). It is imperative to ensure that any AI & ML models used in decision-making follow the principles of fairness, ethics, accountability, explainability, privacy, security, and governance. One of the key design mistakes behind harmful AI systems in practice today is an absence of explicit and precise ethical objectives or constraints. Only by encoding precise statements of ethical standards into our designs can we expect AI systems to make responsible decisions. The talk will cover broad theoretical aspects in this space and highlights its applications in practice.
[3:40PM-4:0PM] Aditya Nittoor - Estimation and Minimization of Execution Cost for Quoting Strategies
Abstract: Capital markets in India and elsewhere are seeing increased volatility in sub-second timeframes when compared to larger time scales. This increased volatility leads to increased instantaneous drawdowns, called slippage, in execution algorithms used by market makers and arbitrageurs. We propose two estimators of slippage, using both high-frequency order flow and the instantaneous limit order book. We forecast the distribution of slippage using Hawkes processes and introduce the Composite Liquidity Factor (CLF) to measure the slippage from the instantaneous limit order book. We empirically show the convergence of the two measures in markets that have high trading activity and discuss the artifacts of market microstructure in Indian capital markets which possibly explain the convergence.
[4:05PM-4:25PM] Sobin Joseph - Non-Parametric Estimation of Multi-dimensional Marked Hawkes Processes
Abstract: An extension of the Hawkes process, the Marked Hawkes process distinguishes itself by featuring variable jump size across each event, in contrast to the constant jump size observed in a Hawkes process without marks. While extensive literature has been dedicated to the non-parametric estimation of both the linear and non-linear Hawkes process, there remains a significant gap in the literature regarding the marked Hawkes process. In response to this, we propose a methodology for estimating the conditional intensity of the marked Hawkes process. We introduce two distinct models: Shallow Neural Hawkes with marks- for Hawkes processes with excitatory kernels and Neural Network with Non-Linear Hawkes with Marks- for non-linear Hawkes processes. Both these approaches take the past arrival times and their corresponding marks as the input to obtain the arrival intensity. This approach is entirely non-parametric, preserving the interpretability associated with the marked Hawkes process. To validate the efficacy of our method, we subject the method to synthetic datasets with known ground truth. Additionally, we apply our method to model cryptocurrency order book data, demonstrating its applicability to real-world scenarios.
[4:30PM-4:55PM] Sumanjay Dutta - Low Sample Statistical Factor Modelling for Asset Pricing
Abstract: Principal component analysis (PCA) and PCA-based statistical factor modeling are fundamental in multivariate analysis, relying on eigen-decomposition of the covariance matrix. Traditional solutions falter in high-dimensional settings where data dimensions approach or exceed the sample size due to ill-conditioned covariance matrices. Current literature suggests linear or nonlinear shrinkage of eigenvalues, while this paper introduces an alternative solution: estimating the sparse inverse covariance matrix using GGM-based methods. The eigenvalues of the inverse covariance matrix, reciprocals of the covariance matrix's eigenvalues, are employed in PCA and factor modeling. Extending low-sample PCAs to asset pricing factor modeling, the paper compares competing methods through synthetic experiments assessing eigenvalue properties and statistical factors. Empirical experiments gauge out-of-sample asset pricing using metrics like RMS-Alpha, Adjusted R-Squared, and Sphericity, with Model Confidence Set (MCS) ranking. GGM-based factor modeling outperforms covariance matrix estimation, aligning with synthetic experiment findings.
[5:00PM] End of Symposium and Tea