Building crypto portfolios with agentic AI ArXiv ID: 2507.20468 “View on arXiv” Authors: Antonino Castelli, Paolo Giudici, Alessandro Piergallini Abstract The rapid growth of crypto markets has opened new opportunities for investors, but at the same time exposed them to high volatility. To address the challenge of managing dynamic portfolios in such an environment, this paper presents a practical application of a multi-agent system designed to autonomously construct and evaluate crypto-asset allocations. Using data on daily frequencies of the ten most capitalized cryptocurrencies from 2020 to 2025, we compare two automated investment strategies. These are a static equal weighting strategy and a rolling-window optimization strategy, both implemented to maximize the evaluation metrics of the Modern Portfolio Theory (MPT), such as Expected Return, Sharpe and Sortino ratios, while minimizing volatility. Each step of the process is handled by dedicated agents, integrated through a collaborative architecture in Crew AI. The results show that the dynamic optimization strategy achieves significantly better performance in terms of risk-adjusted returns, both in-sample and out-of-sample. This highlights the benefits of adaptive techniques in portfolio management, particularly in volatile markets such as cryptocurrency markets. The following methodology proposed also demonstrates how multi-agent systems can provide scalable, auditable, and flexible solutions in financial automation. ...

Function approximations for counterparty credit exposure calculations ArXiv ID: 2507.09004 “View on arXiv” Authors: Domagoj Demeterfi, Kathrin Glau, Linus Wunderlich Abstract The challenge to measure exposures regularly forces financial institutions into a choice between an overwhelming computational burden or oversimplification of risk. To resolve this unsettling dilemma, we systematically investigate replacing frequently called derivative pricers by function approximations covering all practically relevant exposure measures, including quantiles. We prove error bounds for exposure measures in terms of the $L^p$ norm, $1 \leq p < \infty$, and for the uniform norm. To fully exploit these results, we employ the Chebyshev interpolation and show exponential convergence of the resulting exposure calculations. As our main result we derive probabilistic and finite sample error bounds under mild conditions including the natural case of unbounded risk factors. We derive an asymptotic efficiency gain scaling with $n^{“1/2-\varepsilon”}$ for any $\varepsilon>0$ with $n$ the number of simulations. Our numerical experiments cover callable, barrier, stochastic volatility and jump features. Using 10,000 simulations, we consistently observe significant run-time reductions in all cases with speed-up factors up to 230, and an average speed-up of 87. We also present an adaptive choice of the interpolation degree. Finally, numerical examples relying on the approximation of Greeks highlight the merit of the method beyond the presented theory. ...

Temperature Measurement in Agent Systems ArXiv ID: 2507.08394 “View on arXiv” Authors: Christoph J. Börner, Ingo Hoffmann Abstract Models for spin systems, known from statistical physics, are applied analogously in econometrics in the form of agent-based models. The models discussed in the econophysics literature all use the state variable $T$, which, in physics, represents the temperature of a system. However, there is little evidence on how temperature can be measured in econophysics, so that the models can be applied. Only in idealized capital market applications has the relationship between temperature and volatility been demonstrated, allowing temperature to be determined through volatility measurements. The question remains how this can be achieved in agent systems beyond capital market applications. This paper focuses precisely on this question. It examines an agent system with two decision options in a news environment, establishes the measurement equation, and outlines the basic concept of temperature measurement. The procedure is illustrated using an example. In an application with competing subsystems, an interesting strategy for influencing the average opinion in the competing subsystem is presented. ...

Tensor train representations of Greeks for Fourier-based pricing of multi-asset options ArXiv ID: 2507.08482 “View on arXiv” Authors: Rihito Sakurai, Koichi Miyamoto, Tsuyoshi Okubo Abstract Efficient computation of Greeks for multi-asset options remains a key challenge in quantitative finance. While Monte Carlo (MC) simulation is widely used, it suffers from the large sample complexity for high accuracy. We propose a framework to compute Greeks in a single evaluation of a tensor train (TT), which is obtained by compressing the Fourier transform (FT)-based pricing function via TT learning using tensor cross interpolation. Based on this TT representation, we introduce two approaches to compute Greeks: a numerical differentiation (ND) approach that applies a numerical differential operator to one tensor core and an analytical (AN) approach that constructs the TT of closed-form differentiation expressions of FT-based pricing. Numerical experiments on a five-asset min-call option in the Black-Sholes model show significant speed-ups of up to about $10^{“5”} \times$ over MC while maintaining comparable accuracy. The ND approach matches or exceeds the accuracy of the AN approach and requires lower computational complexity for constructing the TT representation, making it the preferred choice. ...

To Trade or Not to Trade: An Agentic Approach to Estimating Market Risk Improves Trading Decisions ArXiv ID: 2507.08584 “View on arXiv” Authors: Dimitrios Emmanoulopoulos, Ollie Olby, Justin Lyon, Namid R. Stillman Abstract Large language models (LLMs) are increasingly deployed in agentic frameworks, in which prompts trigger complex tool-based analysis in pursuit of a goal. While these frameworks have shown promise across multiple domains including in finance, they typically lack a principled model-building step, relying instead on sentiment- or trend-based analysis. We address this gap by developing an agentic system that uses LLMs to iteratively discover stochastic differential equations for financial time series. These models generate risk metrics which inform daily trading decisions. We evaluate our system in both traditional backtests and using a market simulator, which introduces synthetic but causally plausible price paths and news events. We find that model-informed trading strategies outperform standard LLM-based agents, improving Sharpe ratios across multiple equities. Our results show that combining LLMs with agentic model discovery enhances market risk estimation and enables more profitable trading decisions. ...

A Regression-Based Share Market Prediction Model for Bangladesh ArXiv ID: 2507.18643 “View on arXiv” Authors: Syeda Tasnim Fabiha, Rubaiyat Jahan Mumu, Farzana Aktar, B M Mainul Hossain Abstract Share market is one of the most important sectors of economic development of a country. Everyday almost all companies issue their shares and investors buy and sell shares of these companies. Generally investors want to buy shares of the companies whose market liquidity is comparatively greater. Market liquidity depends on the average price of a share. In this paper, a thorough linear regression analysis has been performed on the stock market data of Dhaka Stock Exchange. Later, the linear model has been compared with random forest based on different metrics showing better results for random forest model. However, the amount of individual significance of different factors on the variability of stock price has been identified and explained. This paper also shows that the time series data is not capable of generating a predictive linear model for analysis. ...

Variable annuities: A closer look at ratchet guarantees, hybrid contract designs, and taxation ArXiv ID: 2507.07358 “View on arXiv” Authors: Jennifer Alonso-Garcia, Len Patrick Dominic M. Garces, Jonathan Ziveyi Abstract This paper investigates optimal withdrawal strategies and behavior of policyholders in a variable annuity (VA) contract with a guaranteed minimum withdrawal benefit (GMWB) rider incorporating taxation and a ratchet mechanism for enhancing the benefit base during the life of the contract. Mathematically, this is accomplished by solving a backward dynamic programming problem associated with optimizing the discounted risk-neutral expectation of cash flows from the contract. Furthermore, reflecting traded VA contracts in the market, we consider hybrid products providing policyholders access to a cash fund which functions as an intermediate repository of earnings from the VA and earns interest at a contractually specified cash rate. We contribute to the literature by revealing several significant interactions among taxation, the cash fund, and the benefit base update mechanism. When tax rates are high, the tax-shielding effect of the cash fund, which is taxed differently from ordinary withdrawals from the VA, plays a significant role in enhancing the attractiveness of the overall contract. Furthermore, the ratchet benefit base update scheme (in contrast to the ubiquitous return-of-premium specification in the literature) tends to discourage early surrender as it provides enhanced downside market risk protection. In addition, the cash fund discourages active withdrawals, with policyholders preferring to transfer the guaranteed withdrawal amount to the cash fund to leverage the cash fund rate. ...

From Rattle to Roar: Optimizer Showdown for MambaStock on S&P 500 ArXiv ID: 2508.04707 “View on arXiv” Authors: Alena Chan, Maria Garmonina Abstract We evaluate the performance of several optimizers on the task of forecasting S&P 500 Index returns with the MambaStock model. Among the most widely used algorithms, gradient-smoothing and adaptive-rate optimizers (for example, Adam and RMSProp) yield the lowest test errors. In contrast, the Lion optimizer offers notably faster training. To combine these advantages, we introduce a novel family of optimizers, Roaree, that dampens the oscillatory loss behavior often seen with Lion while preserving its training speed. ...

Large-scale portfolio optimization with variational neural annealing ArXiv ID: 2507.07159 “View on arXiv” Authors: Nishan Ranabhat, Behnam Javanparast, David Goerz, Estelle Inack Abstract Portfolio optimization is a routine asset management operation conducted in financial institutions around the world. However, under real-world constraints such as turnover limits and transaction costs, its formulation becomes a mixed-integer nonlinear program that current mixed-integer optimizers often struggle to solve. We propose mapping this problem onto a classical Ising-like Hamiltonian and solving it with Variational Neural Annealing (VNA), via its classical formulation implemented using autoregressive neural networks. We demonstrate that VNA can identify near-optimal solutions for portfolios comprising more than 2,000 assets and yields performance comparable to that of state-of-the-art optimizers, such as Mosek, while exhibiting faster convergence on hard instances. Finally, we present a dynamical finite-size scaling analysis applied to the S&P 500, Russell 1000, and Russell 3000 indices, revealing universal behavior and polynomial annealing time scaling of the VNA algorithm on portfolio optimization problems. ...

Beating the Best Constant Rebalancing Portfolio in Long-Term Investment: A Generalization of the Kelly Criterion and Universal Learning Algorithm for Markets with Serial Dependence ArXiv ID: 2507.05994 “View on arXiv” Authors: Duy Khanh Lam Abstract In the online portfolio optimization framework, existing learning algorithms generate strategies that yield significantly poorer cumulative wealth compared to the best constant rebalancing portfolio in hindsight, despite being consistent in asymptotic growth rate. While this unappealing performance can be improved by incorporating more side information, it raises difficulties in feature selection and high-dimensional settings. Instead, the inherent serial dependence of assets’ returns, such as day-of-the-week and other calendar effects, can be leveraged. Although latent serial dependence patterns are commonly detected using large training datasets, this paper proposes an algorithm that learns such dependence using only gradually revealed data, without any assumption on their distribution, to form a strategy that eventually exceeds the cumulative wealth of the best constant rebalancing portfolio. Moreover, the classical Kelly criterion, which requires independent assets’ returns, is generalized to accommodate serial dependence in a market modeled as an independent and identically distributed process of random matrices. In such a stochastic market, where existing learning algorithms designed for stationary processes fail to apply, the proposed learning algorithm still generates a strategy that asymptotically grows to the highest rate among all strategies, matching that of the optimal strategy constructed under the generalized Kelly criterion. The experimental results with real market data demonstrate the theoretical guarantees of the algorithm and its performance as expected, as long as serial dependence is significant, regardless of the validity of the generalized Kelly criterion in the experimental market. This further affirms the broad applicability of the algorithm in general contexts. ...