Impermanent Loss and Loss-vs-Rebalancing in Automated Market Makers: Statistical Properties and Insights
For Brownian motion with a given volatility, impermanent loss (IL) and loss-versus-rebalancing (LVR) have identical expectation values but vastly differing distribution functions.
PDSim: A Shiny App for Simulating and Estimating Polynomial Diffusion Models of Commodity Futures
PDSim is an R package that enables users to simulate commodity futures prices using the polynomial diffusion model and estimate the model parameters via the Extended Kalman Filter or Unscented Kalman Filter.
Robust Portfolio Selection under Time-Dependent and State-Dependent Confidence Set with Bayesian Learning
The investor, who is ambiguity-averse, seeks to maximize the expected utility of terminal wealth under the worst-case scenario of the unknown drift of the risky asset. The confidence set for the drift is time-dependent and state-dependent, and is updated through Bayesian learning.
Markowitz Portfolio Optimization with Adaptive Expected Return Level
The authors propose a novel portfolio optimization model that adaptively optimizes the expected return level and the portfolio simultaneously, relieving the trouble of deciding the return level during an investment and making the model more adaptive to the ever-changing financial market.
다중 에이전트 및 자기 적응형 프레임워크를 활용한 동적 포트폴리오 위험 관리를 위한 딥 강화 학습 프레임워크 개발
다중 에이전트 강화 학습 기반의 MASA 프레임워크를 통해 포트폴리오 수익률과 위험 간의 균형을 동적으로 달성하고자 함
안정적인 공식적 알파 요인 마이닝을 위한 분산 제한 REINFORCE
본 논문은 공식적 알파 요인 마이닝을 위한 새로운 강화학습 알고리즘인 QuantFactor REINFORCE(QFR)를 제안한다. QFR은 기존 PPO 알고리즘의 한계를 극복하고, 안정적이고 효율적인 공식적 알파 요인 생성을 가능하게 한다.
Efficient Mining of Steady Formulaic Alpha Factors using Variance-bounded REINFORCE
The proposed QuantFactor REINFORCE (QFR) algorithm can efficiently mine steady formulaic alpha factors that outperform previous methods in terms of correlation with asset returns and ability to generate excess profits.
Efficient Pricing of Bermudan Swaptions using Deep Joint Learning Techniques
This paper proposes an efficient deep learning-based approach to price Bermudan swaptions, a complex financial derivative, by combining sophisticated neural network concepts like differential machine learning, Monte Carlo simulation-based training, and joint learning.
Accelerating Parallel-in-Time Integration of the Two-Asset Black-Scholes Equation using a Fourier Neural Operator as Coarse Propagator
A physics-informed Fourier Neural Operator (PINO) can serve as an effective and computationally efficient coarse propagator for the parallel-in-time integration of the two-asset Black-Scholes equation using the Parareal method, enabling better overall speedup compared to both purely spatial parallelization and space-time parallelization with a numerical coarse propagator.
Efficient Simulation of Optimal Investment Strategies in Competitive Financial Markets with Heterogeneous Agents
The core message of this paper is to propose an efficient deep learning-based method for computing the equilibrium strategies and utilities in a graphon game formulation of a utility maximization problem under relative performance concern among a large population of heterogeneous financial agents.
© 2024 by Linnk AI