A Sinusoidal Hull-White Model for Interest Rate Dynamics: Capturing Long-Term Periodicity in U.S. Treasury Yields ArXiv ID: 2506.06317 “View on arXiv” Authors: Amit Kumar Jha Abstract This study is motivated by empirical observations of periodic fluctuations in interest rates, notably long-term economic cycles spanning decades, which the conventional Hull-White short-rate model fails to adequately capture. To address this limitation, we propose an extension that incorporates a sinusoidal, time-varying mean reversion speed, allowing the model to reflect cyclic interest rate dynamics more effectively. The model is calibrated using a comprehensive dataset of daily U.S. Treasury yield curves obtained from the Federal Reserve Economic Data (FRED) database, covering the period from January 1990 to December 2022. The dataset includes tenors of 1, 2, 3, 5, 7, 10, 20, and 30 years, with the most recent yields ranging from 1.22% (1-year) to 2.36% (30-year). Calibration is performed using the Nelder-Mead optimization algorithm, and Monte Carlo simulations with 200 paths and a time step of 0.05 years. The resulting 30-year zero-coupon bond price under the proposed model is 0.43, compared to 0.47 under the standard Hull-White model. This corresponds to root mean squared errors of 0.12% and 0.14%, respectively, indicating a noticeable improvement in fit, particularly for longer maturities. These results highlight the model’s enhanced capability to capture long-term yield dynamics and suggest significant implications for bond pricing, interest rate risk management, and the valuation of interest rate derivatives. The findings also open avenues for further research into stochastic periodicity and alternative interest rate modeling frameworks. ...

Classifying and Clustering Trading Agents ArXiv ID: 2505.21662 “View on arXiv” Authors: Mateusz Wilinski, Anubha Goel, Alexandros Iosifidis, Juho Kanniainen Abstract The rapid development of sophisticated machine learning methods, together with the increased availability of financial data, has the potential to transform financial research, but also poses a challenge in terms of validation and interpretation. A good case study is the task of classifying financial investors based on their behavioral patterns. Not only do we have access to both classification and clustering tools for high-dimensional data, but also data identifying individual investors is finally available. The problem, however, is that we do not have access to ground truth when working with real-world data. This, together with often limited interpretability of modern machine learning methods, makes it difficult to fully utilize the available research potential. In order to deal with this challenge we propose to use a realistic agent-based model as a way to generate synthetic data. This way one has access to ground truth, large replicable data, and limitless research scenarios. Using this approach we show how, even when classifying trading agents in a supervised manner is relatively easy, a more realistic task of unsupervised clustering may give incorrect or even misleading results. We complete the results with investigating the details of how supervised techniques were able to successfully distinguish between different trading behaviors. ...

Forecasting Nigerian Equity Stock Returns Using Long Short-Term Memory Technique ArXiv ID: 2507.01964 “View on arXiv” Authors: Adebola K. Ojo, Ifechukwude Jude Okafor Abstract Investors and stock market analysts face major challenges in predicting stock returns and making wise investment decisions. The predictability of equity stock returns can boost investor confidence, but it remains a difficult task. To address this issue, a study was conducted using a Long Short-term Memory (LSTM) model to predict future stock market movements. The study used a historical dataset from the Nigerian Stock Exchange (NSE), which was cleaned and normalized to design the LSTM model. The model was evaluated using performance metrics and compared with other deep learning models like Artificial and Convolutional Neural Networks (CNN). The experimental results showed that the LSTM model can predict future stock market prices and returns with over 90% accuracy when trained with a reliable dataset. The study concludes that LSTM models can be useful in predicting financial time-series-related problems if well-trained. Future studies should explore combining LSTM models with other deep learning techniques like CNN to create hybrid models that mitigate the risks associated with relying on a single model for future equity stock predictions. ...

Replication of Reference-Dependent Preferences and the Risk-Return Trade-Off in the Chinese Market ArXiv ID: 2505.20608 “View on arXiv” Authors: Penggan Xu Abstract This study replicates the findings of Wang et al. (2017) on reference-dependent preferences and their impact on the risk-return trade-off in the Chinese stock market, a unique context characterized by high retail investor participation, speculative trading behavior, and regulatory complexities. Capital Gains Overhang (CGO), a proxy for unrealized gains or losses, is employed to explore how behavioral biases shape cross-sectional stock returns in an emerging market setting. Utilizing data from 1995 to 2024 and econometric techniques such as Dependent Double Sorting and Fama-MacBeth regressions, this research investigates the interaction between CGO and five risk proxies: Beta, Return Volatility (RETVOL), Idiosyncratic Volatility (IVOL), Firm Age (AGE), and Cash Flow Volatility (CFVOL). Key findings reveal a weaker or absent positive risk-return relationship among high-CGO firms and stronger positive relationships among low-CGO firms, diverging from U.S. market results, and the interaction effects between CGO and risk proxies, significant and positive in the U.S., are predominantly negative in the Chinese market, reflecting structural and behavioral differences, such as speculative trading and diminished reliance on reference points. The results suggest that reference-dependent preferences play a less pronounced role in the Chinese market, emphasizing the need for tailored investment strategies in emerging economies. ...

Hybrid Models for Financial Forecasting: Combining Econometric, Machine Learning, and Deep Learning Models ArXiv ID: 2505.19617 “View on arXiv” Authors: Dominik Stempień, Robert Ślepaczuk Abstract This research systematically develops and evaluates various hybrid modeling approaches by combining traditional econometric models (ARIMA and ARFIMA models) with machine learning and deep learning techniques (SVM, XGBoost, and LSTM models) to forecast financial time series. The empirical analysis is based on two distinct financial assets: the S&P 500 index and Bitcoin. By incorporating over two decades of daily data for the S&P 500 and almost ten years of Bitcoin data, the study provides a comprehensive evaluation of forecasting methodologies across different market conditions and periods of financial distress. Models’ training and hyperparameter tuning procedure is performed using a novel three-fold dynamic cross-validation method. The applicability of applied models is evaluated using both forecast error metrics and trading performance indicators. The obtained findings indicate that the proper construction process of hybrid models plays a crucial role in developing profitable trading strategies, outperforming their individual components and the benchmark Buy&Hold strategy. The most effective hybrid model architecture was achieved by combining the econometric ARIMA model with either SVM or LSTM, under the assumption of a non-additive relationship between the linear and nonlinear components. ...

Comparative analysis of financial data differentiation techniques using LSTM neural network ArXiv ID: 2505.19243 “View on arXiv” Authors: Dominik Stempień, Janusz Gajda Abstract We compare traditional approach of computing logarithmic returns with the fractional differencing method and its tempered extension as methods of data preparation before their usage in advanced machine learning models. Differencing parameters are estimated using multiple techniques. The empirical investigation is conducted on data from four major stock indices covering the most recent 10-year period. The set of explanatory variables is additionally extended with technical indicators. The effectiveness of the differencing methods is evaluated using both forecast error metrics and risk-adjusted return trading performance metrics. The findings suggest that fractional differentiation methods provide a suitable data transformation technique, improving the predictive model forecasting performance. Furthermore, the generated predictions appeared to be effective in constructing profitable trading strategies for both individual assets and a portfolio of stock indices. These results underline the importance of appropriate data transformation techniques in financial time series forecasting, supporting the application of memory-preserving techniques. ...

Distributionally Robust Deep Q-Learning ArXiv ID: 2505.19058 “View on arXiv” Authors: Chung I Lu, Julian Sester, Aijia Zhang Abstract We propose a novel distributionally robust $Q$-learning algorithm for the non-tabular case accounting for continuous state spaces where the state transition of the underlying Markov decision process is subject to model uncertainty. The uncertainty is taken into account by considering the worst-case transition from a ball around a reference probability measure. To determine the optimal policy under the worst-case state transition, we solve the associated non-linear Bellman equation by dualising and regularising the Bellman operator with the Sinkhorn distance, which is then parameterized with deep neural networks. This approach allows us to modify the Deep Q-Network algorithm to optimise for the worst case state transition. We illustrate the tractability and effectiveness of our approach through several applications, including a portfolio optimisation task based on S&{“P”}~500 data. ...

Stochastic Price Dynamics in Response to Order Flow Imbalance: Evidence from CSI 300 Index Futures ArXiv ID: 2505.17388 “View on arXiv” Authors: Chen Hu, Kouxiao Zhang Abstract We conduct modeling of the price dynamics following order flow imbalance in market microstructure and apply the model to the analysis of Chinese CSI 300 Index Futures. There are three findings. The first is that the order flow imbalance is analogous to a shock to the market. Unlike the common practice of using Hawkes processes, we model the impact of order flow imbalance as an Ornstein-Uhlenbeck process with memory and mean-reverting characteristics driven by a jump-type Lévy process. Motivated by the empirically stable correlation between order flow imbalance and contemporaneous price changes, we propose a modified asset price model where the drift term of canonical geometric Brownian motion is replaced by an Ornstein-Uhlenbeck process. We establish stochastic differential equations and derive the logarithmic return process along with its mean and variance processes under initial boundary conditions, and evolution of cost-effectiveness ratio with order flow imbalance as the trading trigger point, termed as the quasi-Sharpe ratio or response ratio. Secondly, our results demonstrate horizon-dependent heterogeneity in how conventional metrics interact with order flow imbalance. This underscores the critical role of forecast horizon selection for strategies. Thirdly, we identify regime-dependent dynamics in the memory and forecasting power of order flow imbalance. This taxonomy provides both a screening protocol for existing indicators and an ex-ante evaluation paradigm for novel metrics. ...

Enhancing Meme Token Market Transparency: A Multi-Dimensional Entity-Linked Address Analysis for Liquidity Risk Evaluation ArXiv ID: 2506.05359 “View on arXiv” Authors: Qiangqiang Liu, Qian Huang, Frank Fan, Haishan Wu, Xueyan Tang Abstract Meme tokens represent a distinctive asset class within the cryptocurrency ecosystem, characterized by high community engagement, significant market volatility, and heightened vulnerability to market manipulation. This paper introduces an innovative approach to assessing liquidity risk in meme token markets using entity-linked address identification techniques. We propose a multi-dimensional method integrating fund flow analysis, behavioral similarity, and anomalous transaction detection to identify related addresses. We develop a comprehensive set of liquidity risk indicators tailored for meme tokens, covering token distribution, trading activity, and liquidity metrics. Empirical analysis of tokens like BabyBonk, NMT, and BonkFork validates our approach, revealing significant disparities between apparent and actual liquidity in meme token markets. The findings of this study provide significant empirical evidence for market participants and regulatory authorities, laying a theoretical foundation for building a more transparent and robust meme token ecosystem. ...

Interpretable Machine Learning for Macro Alpha: A News Sentiment Case Study ArXiv ID: 2505.16136 “View on arXiv” Authors: Yuke Zhang Abstract This study introduces an interpretable machine learning (ML) framework to extract macroeconomic alpha from global news sentiment. We process the Global Database of Events, Language, and Tone (GDELT) Project’s worldwide news feed using FinBERT – a Bidirectional Encoder Representations from Transformers (BERT) based model pretrained on finance-specific language – to construct daily sentiment indices incorporating mean tone, dispersion, and event impact. These indices drive an XGBoost classifier, benchmarked against logistic regression, to predict next-day returns for EUR/USD, USD/JPY, and 10-year U.S. Treasury futures (ZN). Rigorous out-of-sample (OOS) backtesting (5-fold expanding-window cross-validation, OOS period: c. 2017-April 2025) demonstrates exceptional, cost-adjusted performance for the XGBoost strategy: Sharpe ratios achieve 5.87 (EUR/USD), 4.65 (USD/JPY), and 4.65 (Treasuries), with respective compound annual growth rates (CAGRs) exceeding 50% in Foreign Exchange (FX) and 22% in bonds. Shapley Additive Explanations (SHAP) affirm that sentiment dispersion and article impact are key predictive features. Our findings establish that integrating domain-specific Natural Language Processing (NLP) with interpretable ML offers a potent and explainable source of macro alpha. ...