本研究以「退休目標導向」為核心，建構多年期資產配置之評估與決策框架，旨在於既定投資年限與終值目標下，兼顧「每期投入強度」與「資產成長軌跡」，以提升最終達標把握度。為此，我們聚焦兩項需解決的關鍵議題：其一，於相同資產池與交易條件下，衡量動態與靜態資產配置策略之長期績效與風險輪廓；其二，在目標導向中，評估不同投資年限如何改變投資人之投入需求與達標機率，並據以回推出滿足既定信心水準所需之最低起始投入資金。方法上，本文採用一致之月度再平衡、含交易成本與通膨平減之實務口徑，透過歷史樣本回測與蒙地卡羅模擬雙軌評量，將一般績效指標轉譯為「達標率」與「最低所需投入資金」等決策語言。主要發現顯示：動態策略在時間資本充裕時，能透過更好地捕捉市場上漲訊號與報酬與較低的換手成本，提升達標機率並降低所需投入資金；相對地，靜態策略提供更平滑之風險輪廓與較淺尾部風險，但為維持同等達標把握，通常需承擔較高且持續的投入。綜合之下，本研究以可部署之強化學習架構結合目標導向評量，為退休前資產累積提供具體且可落地的量化依據。

This study develops a goal-based framework for multi-period portfolio allocation toward retirement wealth accumulation. Under a fixed horizon and a predefined terminal target, we jointly address two key decisions: (i) the required periodic contribution and (ii) the portfolio growth path, with the goal of maximizing the probability of target attainment.
We examine two questions. First, how does a dynamic asset-allocation policy perform relative to a static mean–variance strategy in long-horizon return and risk? The comparison uses the same investable universe and consistent assumptions on transaction costs and inflation. Second, within a goal-based setting, how does the investment horizon affect contribution needs and success probabilities? We also infer the minimum starting monthly contribution required to achieve a chosen confidence level.
Methodologically, we apply a unified monthly rebalancing protocol incorporating trading frictions and CPI deflation. Evaluation combines historical backtesting with Monte Carlo simulations, translating conventional metrics into goal-based terms—namely, the hit rate (probability of reaching the target) and minimum initial contribution. The dynamic policy is realized through a deployable deep reinforcement learning model, benchmarked against constrained mean–variance optimization (MVO).
Results show that, with sufficient time horizon, the dynamic policy delivers higher upside capture and lower turnover, improving success probabilities and reducing required contributions relative to MVO. In contrast, the static approach offers smoother risk but generally demands larger, more persistent contributions for comparable confidence. Integrating a deployable DRL policy with goal-based evaluation thus provides practical, evidence-based guidance for pre-retirement wealth accumulation and contribution planning.

1. Lusardi, A., & Mitchell, O. S. (2011). Financial literacy and retirement planning in the United States. Journal of Pension Economics and Finance, 10(4), 509–525.
2. van Rooij, M. C. J., Lusardi, A., & Alessie, R. J. M. (2012). Financial literacy, retirement planning and household wealth. The Economic Journal, 122(560), 449–478.
3. OECD. (2024). Supporting the new generation of retail investors in France. OECD Business and Finance Policy Papers.
4. Vanguard. (2020). How America invests: Investor action guide. Vanguard Research.
5. Grable, J., & Lytton, R. H. (1999). Financial risk tolerance revisited: The development of a risk assessment instrument. Financial Services Review, 8(3), 163–181.
6. OECD. (2023). New retail investors in France: Attitudes, knowledge and behaviours. OECD Publishing.
7. Browning, C. G., Cheng, Y., & Finke, M. S. (2016). Spending in retirement: Determining the consumption gap. Journal of Financial Planning, 29(2), 42–53.
8. Markowitz, H. (1952). Portfolio selection. The Journal of Finance, 7(1), 77–91.
9. Elton, E. J., & Gruber, M. J. (1997). Modern portfolio theory, 1950 to date. Journal of Banking & Finance, 21(11), 1743–1759.
10. Jagannathan, R., & Ma, T. (2003). Risk reduction in large portfolios: Why imposing the wrong constraints helps. The Journal of Finance, 58(4), 1651–1683.
11. Ledoit, O., & Wolf, M. (2012). Nonlinear shrinkage estimation of large-dimensional covariance matrices. The Annals of Statistics, 40(2), 1024–1060.
12. Nevins, D. (2004). Goals-based investing: Integrating traditional and behavioral finance. The Journal of Wealth Management, 6(1), 8–23.
13. Das, S., Markowitz, H., Scheid, J., & Statman, M. (2010). Portfolio optimization with mental accounts. Journal of Financial and Quantitative Analysis, 45(2), 311–334.
14. Giron, K. M., Mulvey, J., Milhau, V., & Suri, A. (2018). Applying goal-based investing principles to the retirement problem. EDHEC-Risk Institute.
15. Gu, S., Kelly, B., & Xiu, D. (2020). Empirical asset pricing via machine learning. The Review of Financial Studies, 33(5), 2223–2273.
16. Moody, J., & Saffell, M. (2001). Learning to trade via direct reinforcement. IEEE Transactions on Neural Networks, 12(4), 875–889.
17. Nakayama, Y., & Sawaki, T. (2023). Causal inference on investment constraints and non-stationarity in dynamic portfolio optimization through reinforcement learning. arXiv preprint arXiv:2311.04946.
18. Sood, S., Zhang, L., & Wang, Y. (2023). Deep reinforcement learning for optimal portfolio allocation: A comparative study with mean-variance optimization. In Proceedings of the ICAPS 2023 Workshop on Financial Planning (FinPlan 2023). Association for the Advancement of Artificial Intelligence. https://icaps23.icaps-conference.org/papers/finplan/FinPlan23_paper_4.pdf
19. Xiao, X. (2023). Quantitative investment decision model based on PPO algorithm. Highlights in Science, Engineering and Technology, 34, 16–24.
20. Schulman, J., et al. (2017). Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347.
21. Huang, G., Zhou, X., & Song, Q. (2024). A deep reinforcement learning framework for dynamic portfolio optimization: Evidence from China's stock market. arXiv preprint arXiv:2412.18563.
22. Espiga-Fernández, F., García-Sánchez, Á., & Ordieres-Meré, J. (2024). A systematic approach to portfolio optimization: A comparative study of reinforcement learning agents, market signals, and investment horizons. Algorithms, 17(12), 570.
23. Oshingbesan, A., et al. (2022). Model-free reinforcement learning for asset allocation. arXiv preprint arXiv:2209.10458.
24. Kim, J., et al. (2023). Deep reinforcement learning for asset allocation: Reward clipping. arXiv preprint arXiv:2301.05300.
25. Bengen, W. P. (1994). Determining withdrawal rates using historical data. Journal of Financial Planning, 7(4), 171–180.
26. Ledoit, O., & Wolf, M. (2020). Analytical nonlinear shrinkage of large-dimensional covariance matrices. The Annals of Statistics, 48(5), 3043–3065.