Abstract:We present a novel process for generating synthetic datasets tailored to assess asset allocation methods and construct portfolios within the fixed income universe. Our approach begins by enhancing the CorrGAN model to generate synthetic correlation matrices. Subsequently, we propose an Encoder-Decoder model that samples additional data conditioned on a given correlation matrix. The resulting synthetic dataset facilitates in-depth analyses of asset allocation methods across diverse asset universes. Additionally, we provide a case study that exemplifies the use of the synthetic dataset to improve portfolios constructed within a simulation-based asset allocation process.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is that in fixed - income asset allocation, due to the limited and biased actual historical data, the asset allocation methods perform poorly during the out - of - sample period. Specifically, traditional methods of constructing portfolios based on historical data are often prone to over - fitting and ignore market scenarios that may occur but have not actually occurred. To solve this problem, the paper proposes a new method of generating synthetic datasets to enhance the evaluation and optimization of asset allocation methods. Through this method, researchers can generate more extensive data that can better reflect the asset performance under different market scenarios, thereby improving the robustness and effectiveness of asset allocation strategies in unknown situations. ### Main contributions of the paper: 1. **Generating synthetic correlation matrices**: First, use the Generative Adversarial Network (GAN) model to generate synthetic correlation matrices that can capture the interdependencies between assets. 2. **Generating additional asset characteristics**: Then, use an encoder - decoder model to generate additional asset characteristics such as volatility, expected return, and future realized return based on the generated correlation matrices. 3. **Comprehensive evaluation**: Through the generated synthetic datasets, conduct in - depth analysis and evaluation of asset allocation methods, especially in the field of fixed - income assets. ### Specific steps of the paper: 1. **Data generation**: - **Generating correlation matrices**: Use the improved CorrGAN model to generate synthetic correlation matrices. - **Generating other characteristics**: Use the encoder - decoder model to generate volatility, expected return, and future realized return that are consistent with the generated correlation matrices. 2. **Model evaluation**: - **Performance of the generating model**: Evaluate the performance of the generating model by comparing the generated correlation matrices with the correlation matrices of the actual data. - **Application of synthetic datasets**: Through case studies, demonstrate the application and advantages of synthetic datasets in asset allocation. 3. **Case study**: - **Objective**: Construct a portfolio that can outperform the benchmark in the long run while maintaining a low Tracking Error Volatility (TEV). - **Method**: Use the simulation method based on synthetic datasets to optimize asset weights to minimize the 1 - month deviation of the portfolio relative to the benchmark, while ensuring that the expected return is higher than or equal to the expected return of the benchmark plus an additional target excess return. - **Result**: The experimental results show that the portfolio constructed using synthetic datasets performs better during the out - of - sample period than the portfolio constructed only using the actual datasets. ### Conclusion: The paper effectively solves the problems of insufficient and biased historical data in fixed - income asset allocation by generating synthetic datasets, and improves the robustness and effectiveness of asset allocation strategies in unknown market scenarios. This method provides new tools and ideas for research and practice in the financial field.

Improved Data Generation for Enhanced Asset Allocation: A Synthetic Dataset Approach for the Fixed Income Universe

A Modified CTGAN-Plus-Features Based Method for Optimal Asset Allocation

Generative Adversarial Networks Applied to Synthetic Financial Scenarios Generation

Generation of Realistic Synthetic Financial Time-series

Generation and Simulation of Synthetic Datasets with Copulas

A supervised generative optimization approach for tabular data

Generative Synthesis of Insurance Datasets

Synthetic Data Applications in Finance

Targeted synthetic data generation for tabular data via hardness characterization

GANs and synthetic financial data: calculating VaR*

Boosting Data Analytics With Synthetic Volume Expansion

Synthesizing Property & Casualty Ratemaking Datasets using Generative Adversarial Networks

Synthetic Data Generation in Cybersecurity: A Comparative Analysis

SYNC: A Copula based Framework for Generating Synthetic Data from Aggregated Sources

FinDiff: Diffusion Models for Financial Tabular Data Generation

Generative Machine Learning for Multivariate Equity Returns

FedSyn: Synthetic Data Generation using Federated Learning

A General Framework for Portfolio Construction Based on Generative Models of Asset Returns

Evaluating utility in synthetic banking microdata applications

Portfolio Optimization using Predictive Auxiliary Classifier Generative Adversarial Networks with Measuring Uncertainty

Asset Allocation Model for a Robo-Advisor Using the Financial Market Instability Index and Genetic Algorithms