Deep Learning

Enhancing Portfolio Optimization with Deep Learning Insights ArXiv ID: 2601.07942 “View on arXiv” Authors: Brandon Luo, Jim Skufca Abstract Our work focuses on deep learning (DL) portfolio optimization, tackling challenges in long-only, multi-asset strategies across market cycles. We propose training models with limited regime data using pre-training techniques and leveraging transformer architectures for state variable inclusion. Evaluating our approach against traditional methods shows promising results, demonstrating our models’ resilience in volatile markets. These findings emphasize the evolving landscape of DL-driven portfolio optimization, stressing the need for adaptive strategies to navigate dynamic market conditions and improve predictive accuracy. ...

Deep Learning and Elicitability for McKean-Vlasov FBSDEs With Common Noise ArXiv ID: 2512.14967 “View on arXiv” Authors: Felipe J. P. Antunes, Yuri F. Saporito, Sebastian Jaimungal Abstract We present a novel numerical method for solving McKean-Vlasov forward-backward stochastic differential equations (MV-FBSDEs) with common noise, combining Picard iterations, elicitability and deep learning. The key innovation involves elicitability to derive a path-wise loss function, enabling efficient training of neural networks to approximate both the backward process and the conditional expectations arising from common noise - without requiring computationally expensive nested Monte Carlo simulations. The mean-field interaction term is parameterized via a recurrent neural network trained to minimize an elicitable score, while the backward process is approximated through a feedforward network representing the decoupling field. We validate the algorithm on a systemic risk inter-bank borrowing and lending model, where analytical solutions exist, demonstrating accurate recovery of the true solution. We further extend the model to quantile-mediated interactions, showcasing the flexibility of the elicitability framework beyond conditional means or moments. Finally, we apply the method to a non-stationary Aiyagari–Bewley–Huggett economic growth model with endogenous interest rates, illustrating its applicability to complex mean-field games without closed-form solutions. ...

Partial multivariate transformer as a tool for cryptocurrencies time series prediction ArXiv ID: 2512.04099 “View on arXiv” Authors: Andrzej Tokajuk, Jarosław A. Chudziak Abstract Forecasting cryptocurrency prices is hindered by extreme volatility and a methodological dilemma between information-scarce univariate models and noise-prone full-multivariate models. This paper investigates a partial-multivariate approach to balance this trade-off, hypothesizing that a strategic subset of features offers superior predictive power. We apply the Partial-Multivariate Transformer (PMformer) to forecast daily returns for BTCUSDT and ETHUSDT, benchmarking it against eleven classical and deep learning models. Our empirical results yield two primary contributions. First, we demonstrate that the partial-multivariate strategy achieves significant statistical accuracy, effectively balancing informative signals with noise. Second, we experiment and discuss an observable disconnect between this statistical performance and practical trading utility; lower prediction error did not consistently translate to higher financial returns in simulations. This finding challenges the reliance on traditional error metrics and highlights the need to develop evaluation criteria more aligned with real-world financial objectives. ...

Machine-learning a family of solutions to an optimal pension investment problem ArXiv ID: 2511.07045 “View on arXiv” Authors: John Armstrong, Cristin Buescu, James Dalby, Rohan Hobbs Abstract We use a neural network to identify the optimal solution to a family of optimal investment problems, where the parameters determining an investor’s risk and consumption preferences are given as inputs to the neural network in addition to economic variables. This is used to develop a practical tool that can be used to explore how pension outcomes vary with preference parameters. We use a Black-Scholes economic model so that we may validate the accuracy of network using a classical and provably convergent numerical method developed using the duality approach. ...

Exact Terminal Condition Neural Network for American Option Pricing Based on the Black-Scholes-Merton Equations ArXiv ID: 2510.27132 “View on arXiv” Authors: Wenxuan Zhang, Yixiao Guo, Benzhuo Lu Abstract This paper proposes the Exact Terminal Condition Neural Network (ETCNN), a deep learning framework for accurately pricing American options by solving the Black-Scholes-Merton (BSM) equations. The ETCNN incorporates carefully designed functions that ensure the numerical solution not only exactly satisfies the terminal condition of the BSM equations but also matches the non-smooth and singular behavior of the option price near expiration. This method effectively addresses the challenges posed by the inequality constraints in the BSM equations and can be easily extended to high-dimensional scenarios. Additionally, input normalization is employed to maintain the homogeneity. Multiple experiments are conducted to demonstrate that the proposed method achieves high accuracy and exhibits robustness across various situations, outperforming both traditional numerical methods and other machine learning approaches. ...

Increase Alpha: Performance and Risk of an AI-Driven Trading Framework ArXiv ID: 2509.16707 “View on arXiv” Authors: Sid Ghatak, Arman Khaledian, Navid Parvini, Nariman Khaledian Abstract There are inefficiencies in financial markets, with unexploited patterns in price, volume, and cross-sectional relationships. While many approaches use large-scale transformers, we take a domain-focused path: feed-forward and recurrent networks with curated features to capture subtle regularities in noisy financial data. This smaller-footprint design is computationally lean and reliable under low signal-to-noise, crucial for daily production at scale. At Increase Alpha, we built a deep-learning framework that maps over 800 U.S. equities into daily directional signals with minimal computational overhead. The purpose of this paper is twofold. First, we outline the general overview of the predictive model without disclosing its core underlying concepts. Second, we evaluate its real-time performance through transparent, industry standard metrics. Forecast accuracy is benchmarked against both naive baselines and macro indicators. The performance outcomes are summarized via cumulative returns, annualized Sharpe ratio, and maximum drawdown. The best portfolio combination using our signals provides a low-risk, continuous stream of returns with a Sharpe ratio of more than 2.5, maximum drawdown of around 3%, and a near-zero correlation with the S&P 500 market benchmark. We also compare the model’s performance through different market regimes, such as the recent volatile movements of the US equity market in the beginning of 2025. Our analysis showcases the robustness of the model and significantly stable performance during these volatile periods. Collectively, these findings show that market inefficiencies can be systematically harvested with modest computational overhead if the right variables are considered. This report will emphasize the potential of traditional deep learning frameworks for generating an AI-driven edge in the financial market. ...

Deep Learning for Conditional Asset Pricing Models ArXiv ID: 2509.04812 “View on arXiv” Authors: Hongyi Liu Abstract We propose a new pseudo-Siamese Network for Asset Pricing (SNAP) model, based on deep learning approaches, for conditional asset pricing. Our model allows for the deep alpha, deep beta and deep factor risk premia conditional on high dimensional observable information of financial characteristics and macroeconomic states, while storing the long-term dependency of the informative features through long short-term memory network. We apply this method to monthly U.S. stock returns from 1970-2019 and find that our pseudo-SNAP model outperforms the benchmark approaches in terms of out-of-sample prediction and out-of-sample Sharpe ratio. In addition, we also apply our method to calculate deep mispricing errors which we use to construct an arbitrage portfolio K-Means clustering. We find that the arbitrage portfolio has significant alphas. ...

FinCast: A Foundation Model for Financial Time-Series Forecasting ArXiv ID: 2508.19609 “View on arXiv” Authors: Zhuohang Zhu, Haodong Chen, Qiang Qu, Vera Chung Abstract Financial time-series forecasting is critical for maintaining economic stability, guiding informed policymaking, and promoting sustainable investment practices. However, it remains challenging due to various underlying pattern shifts. These shifts arise primarily from three sources: temporal non-stationarity (distribution changes over time), multi-domain diversity (distinct patterns across financial domains such as stocks, commodities, and futures), and varying temporal resolutions (patterns differing across per-second, hourly, daily, or weekly indicators). While recent deep learning methods attempt to address these complexities, they frequently suffer from overfitting and typically require extensive domain-specific fine-tuning. To overcome these limitations, we introduce FinCast, the first foundation model specifically designed for financial time-series forecasting, trained on large-scale financial datasets. Remarkably, FinCast exhibits robust zero-shot performance, effectively capturing diverse patterns without domain-specific fine-tuning. Comprehensive empirical and qualitative evaluations demonstrate that FinCast surpasses existing state-of-the-art methods, highlighting its strong generalization capabilities. ...

Deep Learning for Short Term Equity Trend Forecasting: A Behavior Driven Multi Factor Approach ArXiv ID: 2508.14656 “View on arXiv” Authors: Yuqi Luan Abstract This study proposes a behaviorally-informed multi-factor stock selection framework that integrates short-cycle technical alpha signals with deep learning. We design a dual-task multilayer perceptron (MLP) that jointly predicts five-day future returns and directional price movements, thereby capturing nonlinear market behaviors such as volume-price divergence, momentum-driven herding, and bottom reversals. The model is trained on 40 carefully constructed factors derived from price-volume patterns and behavioral finance insights. Empirical evaluation demonstrates that the dual-task MLP achieves superior and stable performance across both predictive accuracy and economic relevance, as measured by information coefficient (IC), information ratio (IR), and portfolio backtesting results. Comparative experiments further show that deep learning methods outperform linear baselines by effectively capturing structural interactions between factors. This work highlights the potential of structure-aware deep learning in enhancing multi-factor modeling and provides a practical framework for short-horizon quantitative investment strategies. ...

Generative Neural Operators of Log-Complexity Can Simultaneously Solve Infinitely Many Convex Programs ArXiv ID: 2508.14995 “View on arXiv” Authors: Anastasis Kratsios, Ariel Neufeld, Philipp Schmocker Abstract Neural operators (NOs) are a class of deep learning models designed to simultaneously solve infinitely many related problems by casting them into an infinite-dimensional space, whereon these NOs operate. A significant gap remains between theory and practice: worst-case parameter bounds from universal approximation theorems suggest that NOs may require an unrealistically large number of parameters to solve most operator learning problems, which stands in direct opposition to a slew of experimental evidence. This paper closes that gap for a specific class of {“NOs”}, generative {“equilibrium operators”} (GEOs), using (realistic) finite-dimensional deep equilibrium layers, when solving families of convex optimization problems over a separable Hilbert space $X$. Here, the inputs are smooth, convex loss functions on $X$, and outputs are the associated (approximate) solutions to the optimization problem defined by each input loss. We show that when the input losses lie in suitable infinite-dimensional compact sets, our GEO can uniformly approximate the corresponding solutions to arbitrary precision, with rank, depth, and width growing only logarithmically in the reciprocal of the approximation error. We then validate both our theoretical results and the trainability of GEOs on three applications: (1) nonlinear PDEs, (2) stochastic optimal control problems, and (3) hedging problems in mathematical finance under liquidity constraints. ...