Newest papers

MLP-Initialized Reinforcement Learning for Exit Timing in SPX Short-Straddle Portfolios

Jianning Wang, Robert Ślepaczuk, Paweł Sakowski · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.13126 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

Este artigo estuda o cronograma de saída para portfólios SPX de curta faixa usando um quadro de aprendizagem supervisionado e reforço de dois estágios. Episódios históricos geram primeiro rótulos de retenção/saída óptimos ex post para treinar um perceptron de camada múltipla (MLP) usando apenas variáveis de estado disponíveis simultaneamente na implantação. O MLP treinado inicia os atores de aprendizagem de reforço que são finamente ajustados com uma recompensa de negociação consciente do risco. A avaliação de amostra em 2024-2026 mostra que o MLP supervisionado supera substancialmente a participação passiva e as regras de saída representativas de tempo fixo e baseadas em lucro e perda. Ao longo de sete sementes independentes, os Actor-Criticages iniciados

Machine translation (Meta M2M100); the English original follows.

Abstract

This paper studies exit timing for SPX short-straddle portfolios using a two-stage supervised and reinforcement-learning framework. Historical episodes first generate ex-post optimal hold/exit labels for training a multilayer perceptron (MLP) using only contemporaneously available state variables at deployment. The trained MLP then initializes reinforcement-learning actors that are fine-tuned with a risk-aware trading reward. Out-of-sample evaluation over 2024-2026 shows that the supervised MLP substantially outperforms passive holding and representative fixed-time and profit-and-loss-based exit rules. Across seven independent seeds, MLP-initialized Actor-Critic averages 37.54% cumulative return and a Sharpe ratio of 1.705, compared with 35.01% and 1.651 for the supervised MLP. Actor-Critic trained from scratch averages only 0.18% return. These findings show that oracle-guided supervised learning provides a stable foundation for reward-based policy fine-tuning in a sparse and path-dependent option-exit problem.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Sharpe ratio, Artificial intelligence, Computer science, Machine learning, Reinforcement learning, Supervised learning

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange