CyberSec Research

Propose a deep learning driven multi factor investment model optimization method for risk control. By constructing a deep learning model based on Long Short Term Memory (LSTM) and combining it with a multi factor investment model, we optimize factor selection and weight determination to enhance the model's adaptability and robustness to market changes. Empirical analysis shows that the LSTM model is significantly superior to the benchmark model in risk control indicators such as maximum retracement, Sharp ratio and value at risk (VaR), and shows strong adaptability and robustness in different market environments. Furthermore, the model is applied to the actual portfolio to optimize the asset allocation, which significantly improves the performance of the portfolio, provides investors with more scientific and accurate investment decision-making basis, and effectively balances the benefits and risks.

The proliferation of artificial intelligence (AI) in financial services has prompted growing demand for tools that can systematically detect AI-related disclosures in corporate filings. While prior approaches often rely on keyword expansion or document-level classification, they fall short in granularity, interpretability, and robustness. This study introduces FinAI-BERT, a domain-adapted transformer-based language model designed to classify AI-related content at the sentence level within financial texts. The model was fine-tuned on a manually curated and balanced dataset of 1,586 sentences drawn from 669 annual reports of U.S. banks (2015 to 2023). FinAI-BERT achieved near-perfect classification performance (accuracy of 99.37 percent, F1 score of 0.993), outperforming traditional baselines such as Logistic Regression, Naive Bayes, Random Forest, and XGBoost. Interpretability was ensured through SHAP-based token attribution, while bias analysis and robustness checks confirmed the model's stability across sentence lengths, adversarial inputs, and temporal samples. Theoretically, the study advances financial NLP by operationalizing fine-grained, theme-specific classification using transformer architectures. Practically, it offers a scalable, transparent solution for analysts, regulators, and scholars seeking to monitor the diffusion and framing of AI across financial institutions.

Constructing the Implied Volatility Surface (IVS) is a challenging task in quantitative finance due to the complexity of real markets and the sparsity of market data. Structural models like Stochastic Alpha Beta Rho (SABR) model offer interpretability and theoretical consistency but lack flexibility, while purely data-driven methods such as Gaussian Process regression can struggle with sparse data. We introduce SABR-Informed Multi-Task Gaussian Process (SABR-MTGP), treating IVS construction as a multi-task learning problem. Our method uses a dense synthetic dataset from a calibrated SABR model as a source task to inform the construction based on sparse market data (the target task). The MTGP framework captures task correlation and transfers structural information adaptively, improving predictions particularly in data-scarce regions. Experiments using Heston-generated ground truth data under various market conditions show that SABR-MTGP outperforms both standard Gaussian process regression and SABR across different maturities. Furthermore, an application to real SPX market data demonstrates the method's practical applicability and its ability to produce stable and realistic surfaces. This confirms our method balances structural guidance from SABR with the flexibility needed for market data.

Forecasting central bank policy decisions remains a persistent challenge for investors, financial institutions, and policymakers due to the wide-reaching impact of monetary actions. In particular, anticipating shifts in the U.S. federal funds rate is vital for risk management and trading strategies. Traditional methods relying only on structured macroeconomic indicators often fall short in capturing the forward-looking cues embedded in central bank communications. This study examines whether predictive accuracy can be enhanced by integrating structured data with unstructured textual signals from Federal Reserve communications. We adopt a multi-modal framework, comparing traditional machine learning models, transformer-based language models, and deep learning architectures in both unimodal and hybrid settings. Our results show that hybrid models consistently outperform unimodal baselines. The best performance is achieved by combining TF-IDF features of FOMC texts with economic indicators in an XGBoost classifier, reaching a test AUC of 0.83. FinBERT-based sentiment features marginally improve ranking but perform worse in classification, especially under class imbalance. SHAP analysis reveals that sparse, interpretable features align more closely with policy-relevant signals. These findings underscore the importance of integrating textual and structured signals transparently. For monetary policy forecasting, simpler hybrid models can offer both accuracy and interpretability, delivering actionable insights for researchers and decision-makers.

Financial institutions increasingly adopt customer-centric strategies to enhance profitability and build long-term relationships. While Customer Lifetime Value (CLV) is a core metric, its calculations often rely solely on single-entity data, missing insights from customer activities across multiple firms. This study introduces the Potential Customer Lifetime Value (PCLV) framework, leveraging Open Banking (OB) data to estimate customer value comprehensively. We predict retention probability and estimate Potential Contribution Margins (PCM) from competitor data, enabling PCLV calculation. Results show that OB data can be used to estimate PCLV per competitor, indicating a potential upside of 21.06% over the Actual CLV. PCLV offers a strategic tool for managers to strengthen competitiveness by leveraging OB data and boost profitability by driving marketing efforts at the individual customer level to increase the Actual CLV.

Extending Buehler et al.'s 2019 Deep Hedging paradigm, we innovatively employ deep neural networks to parameterize convex-risk minimization (CVaR/ES) for the portfolio tail-risk hedging problem. Through comprehensive numerical experiments on crisis-era bootstrap market simulators -- customizable with transaction costs, risk budgets, liquidity constraints, and market impact -- our end-to-end framework not only achieves significant one-day 99% CVaR reduction but also yields practical insights into friction-aware strategy adaptation, demonstrating robustness and operational viability in realistic markets.

We propose a hybrid quantum-classical reinforcement learning framework for sector rotation in the Taiwan stock market. Our system employs Proximal Policy Optimization (PPO) as the backbone algorithm and integrates both classical architectures (LSTM, Transformer) and quantum-enhanced models (QNN, QRWKV, QASA) as policy and value networks. An automated feature engineering pipeline extracts financial indicators from capital share data to ensure consistent model input across all configurations. Empirical backtesting reveals a key finding: although quantum-enhanced models consistently achieve higher training rewards, they underperform classical models in real-world investment metrics such as cumulative return and Sharpe ratio. This discrepancy highlights a core challenge in applying reinforcement learning to financial domains -- namely, the mismatch between proxy reward signals and true investment objectives. Our analysis suggests that current reward designs may incentivize overfitting to short-term volatility rather than optimizing risk-adjusted returns. This issue is compounded by the inherent expressiveness and optimization instability of quantum circuits under Noisy Intermediate-Scale Quantum (NISQ) constraints. We discuss the implications of this reward-performance gap and propose directions for future improvement, including reward shaping, model regularization, and validation-based early stopping. Our work offers a reproducible benchmark and critical insights into the practical challenges of deploying quantum reinforcement learning in real-world finance.

We propose a novel two-stage framework to detect lead-lag relationships in the Chinese A-share market. First, long-term coupling between stocks is measured via daily data using correlation, dynamic time warping, and rank-based metrics. Then, high-frequency data (1-, 5-, and 15-minute) is used to detect statistically significant lead-lag patterns via cross-correlation, Granger causality, and regression models. Our low-coupling modular system supports scalable data processing and improves reproducibility. Results show that strongly coupled stock pairs often exhibit lead-lag effects, especially at finer time scales. These findings provide insights into market microstructure and quantitative trading opportunities.

We examine strategically incorporating broad stock market leveraged exchange-traded funds (LETFs) into investment portfolios. We demonstrate that easily understandable and implementable strategies can enhance the risk-return profile of a portfolio containing LETFs. Our analysis shows that seemingly reasonable investment strategies may result in undesirable Omega ratios, with these effects compounding across rebalancing periods. By contrast, relatively simple dynamic strategies that systematically de-risk the portfolio once gains are observed can exploit this compounding effect, taking advantage of favorable Omega ratio dynamics. Our findings suggest that LETFs represent a valuable tool for investors employing dynamic strategies, while confirming their well-documented unsuitability for passive or static approaches.

We present a reinforcement learning (RL)-driven framework for optimizing block-preconditioner sizes in iterative solvers used in portfolio optimization and option pricing. The covariance matrix in portfolio optimization or the discretization of differential operators in option pricing models lead to large linear systems of the form $\mathbf{A}\textbf{x}=\textbf{b}$. Direct inversion of high-dimensional portfolio or fine-grid option pricing incurs a significant computational cost. Therefore, iterative methods are usually used for portfolios in real-world situations. Ill-conditioned systems, however, suffer from slow convergence. Traditional preconditioning techniques often require problem-specific parameter tuning. To overcome this limitation, we rely on RL to dynamically adjust the block-preconditioner sizes and accelerate iterative solver convergence. Evaluations on a suite of real-world portfolio optimization matrices demonstrate that our RL framework can be used to adjust preconditioning and significantly accelerate convergence and reduce computational cost. The proposed accelerated solver supports faster decision-making in dynamic portfolio allocation and real-time option pricing.

Despite significant advancements in machine learning for derivative pricing, the efficient and accurate valuation of American options remains a persistent challenge due to complex exercise boundaries, near-expiry behavior, and intricate contractual features. This paper extends a semi-analytical approach for pricing American options in time-inhomogeneous models, including pure diffusions, jump-diffusions, and Levy processes. Building on prior work, we derive and solve Volterra integral equations of the second kind to determine the exercise boundary explicitly, offering a computationally superior alternative to traditional finite-difference and Monte Carlo methods. We address key open problems: (1) extending the decomposition method, i.e. splitting the American option price into its European counterpart and an early exercise premium, to general jump-diffusion and Levy models; (2) handling cases where closed-form transition densities are unavailable by leveraging characteristic functions via, e.g., the COS method; and (3) generalizing the framework to multidimensional diffusions. Numerical examples demonstrate the method's efficiency and robustness. Our results underscore the advantages of the integral equation approach for large-scale industrial applications, while resolving some limitations of existing techniques.

The key objective of this paper is to develop an empirical model for pricing SPX options that can be simulated over future paths of the SPX. To accomplish this, we formulate and rigorously evaluate several statistical models, including neural network, random forest, and linear regression. These models use the observed characteristics of the options as inputs -- their price, moneyness and time-to-maturity, as well as a small set of external inputs, such as the SPX and its past history, dividend yield, and the risk-free rate. Model evaluation is performed on historical options data, spanning 30 years of daily observations. Significant effort is given to understanding the data and ensuring explainability for the neural network. A neural network model with two hidden layers and four neurons per layer, trained with minimal hyperparameter tuning, performs well against the theoretical Black-Scholes-Merton model for European options, as well as two other empirical models based on the random forest and the linear regression. It delivers arbitrage-free option prices without requiring these conditions to be imposed.

We propose a quantum machine learning framework for approximating solutions to high-dimensional parabolic partial differential equations (PDEs) that can be reformulated as backward stochastic differential equations (BSDEs). In contrast to popular quantum-classical network hybrid approaches, this study employs the pure Variational Quantum Circuit (VQC) as the core solver without trainable classical neural networks. The quantum BSDE solver performs pathwise approximation via temporal discretization and Monte Carlo simulation, framed as model-based reinforcement learning. We benchmark VQCbased and classical deep neural network (DNN) solvers on two canonical PDEs as representatives: the Black-Scholes and nonlinear Hamilton-Jacobi-Bellman (HJB) equations. The VQC achieves lower variance and improved accuracy in most cases, particularly in highly nonlinear regimes and for out-of-themoney options, demonstrating greater robustness than DNNs. These results, obtained via quantum circuit simulation, highlight the potential of VQCs as scalable and stable solvers for highdimensional stochastic control problems.

We explore credit risk pricing by modeling equity as a call option and debt as the difference between the firm's asset value and a put option, following the structural framework of the Merton model. Our approach proceeds in two stages: first, we calibrate the asset volatility using the Black-Scholes-Merton (BSM) formula; second, we recover implied mean return and probability surfaces under the physical measure. To achieve this, we construct a recombining binomial tree under the real-world (natural) measure, assuming a fixed initial asset value. The volatility input is taken from a specific region of the implied volatility surface - based on moneyness and maturity - which then informs the calibration of drift and probability. A novel mapping is established between risk-neutral and physical parameters, enabling construction of implied surfaces that reflect the market's credit expectations and offer practical tools for stress testing and credit risk analysis.

Here we demonstrate how we can use Small Volatility Approximation in calibration of Multi-Factor HJM model with deterministic correlations, factor volatilities and mean reversals. It is noticed that quality of this calibration is very good and it does not depend on number of factors.

Local stochastic volatility refers to a popular model class in applied mathematical finance that allows for "calibration-on-the-fly", typically via a particle method, derived from a formal McKean-Vlasov equation. Well-posedness of this limit is a well-known problem in the field; the general case is largely open, despite recent progress in Markovian situations. Our take is to start with a well-defined Euler approximation to the formal McKean-Vlasov equation, followed by a newly established half-step-scheme, allowing for good approximations of conditional expectations. In a sense, we do Euler first, particle second in contrast to previous works that start with the particle approximation. We show weak order one for the Euler discretization, plus error terms that account for the said approximation. The case of particle approximation is discussed in detail and the error rate is given in dependence of all parameters used.

Financial decision-making presents unique challenges for language models, demanding temporal reasoning, adaptive risk assessment, and responsiveness to dynamic events. While large language models (LLMs) show strong general reasoning capabilities, they often fail to capture behavioral patterns central to human financial decisions-such as expert reliance under information asymmetry, loss-averse sensitivity, and feedback-driven temporal adjustment. We propose FinHEAR, a multi-agent framework for Human Expertise and Adaptive Risk-aware reasoning. FinHEAR orchestrates specialized LLM-based agents to analyze historical trends, interpret current events, and retrieve expert-informed precedents within an event-centric pipeline. Grounded in behavioral economics, it incorporates expert-guided retrieval, confidence-adjusted position sizing, and outcome-based refinement to enhance interpretability and robustness. Empirical results on curated financial datasets show that FinHEAR consistently outperforms strong baselines across trend prediction and trading tasks, achieving higher accuracy and better risk-adjusted returns.

This paper addresses the challenge of model uncertainty in quantitative finance, where decisions in portfolio allocation, derivative pricing, and risk management rely on estimating stochastic models from limited data. In practice, the unavailability of the true probability measure forces reliance on an empirical approximation, and even small misestimations can lead to significant deviations in decision quality. Building on the framework of Klibanoff et al. (2005), we enhance the conventional objective - whether this is expected utility in an investing context or a hedging metric - by superimposing an outer "uncertainty measure", motivated by traditional monetary risk measures, on the space of models. In scenarios where a natural model distribution is lacking or Bayesian methods are impractical, we propose an ad hoc subsampling strategy, analogous to bootstrapping in statistical finance and related to mini-batch sampling in deep learning, to approximate model uncertainty. To address the quadratic memory demands of naive implementations, we also present an adapted stochastic gradient descent algorithm that enables efficient parallelization. Through analytical, simulated, and empirical studies - including multi-period, real data and high-dimensional examples - we demonstrate that uncertainty measures outperform traditional mixture of measures strategies and our model-agnostic subsampling-based approach not only enhances robustness against model risk but also achieves performance comparable to more elaborate Bayesian methods.