Paper: arXiv 2512.07860
Authors: Mohammed Alruqimi, Luca Di Persio
Abstract
This paper investigates an optimal integration of deep learning with financial models for robust asset price forecasting. Specifically, we developed a hybrid framework combining a Long Short-Term Memory (LSTM) network with the Merton-Lévy jump-diffusion model. To optimise this framework, we employed the Grey Wolf Optimizer (GWO) for the LSTM hyperparameter tuning, and we explored three calibration methods for the Merton-Levy model parameters: Artificial Neural Networks (ANNs), the Marine Predators Algorithm (MPA), and the PyTorch-based TorchSDE library. To evaluate the predictive performance of our hybrid model, we compared it against several benchmark models, including a standard LSTM and an LSTM combined with the Fractional Heston model. This evaluation used three real-world financial datasets: Brent oil prices, the STOXX 600 index, and the IT40 index. Performance was assessed using standard metrics, including Mean Squared Error (MSE), Mean Absolute Error(MAE), Mean Squared Percentage Error (MSPE), and the coefficient of determination (R2). Our experimental results demonstrate that the hybrid model, combining a GWO-optimized LSTM network with the Levy-Merton Jump-Diffusion model calibrated using an ANN, outperformed the base LSTM model and all other models developed in this study.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced stochastic calculus including Lévy processes, jump-diffusion SDEs, and fractional Brownian motion, while demonstrating empirical rigor through backtesting on three real-world datasets with multiple performance metrics and hyperparameter tuning.
Research Flowchart
flowchart TD
A["Research Goal:<br>Optimal DL-Financial Model Integration"] --> B["Input Data:<br>Brent Oil, STOXX 600, IT40 Indices"]
B --> C["Framework Development:<br>LSTM + Merton-Lévy Jump-Diffusion"]
C --> D["Hybrid Optimization"]
D --> E["Performance Evaluation<br>MSE, MAE, MSPE, R²"]
subgraph D ["Hybrid Optimization"]
D1["LSTM Tuning:<br>Grey Wolf Optimizer GWO"]
D2["Model Calibration:<br>ANN, MPA, TorchSDE"]
end
D --> F["Key Finding:<br>GWO-LSTM + ANN Calibration<br>outperforms benchmarks"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a hybrid framework that combines a Long Short-Term Memory (LSTM) network with the Merton‑Lévy jump‑diffusion model, tuned by a Grey Wolf Optimizer and calibrated via neural networks.
Source sentence
we developed a hybrid framework combining a Long Short-Term Memory (LSTM) network with the Merton-Lévy jump-diffusion model.
Assumes 2
- LSTM models implicitly assume local stationarity within their memory horizon. model
Source sentence
LSTMs, while capable of learning temporal dependencies, implicitly assume local stationarity within their memory horizon.
- Data preprocessing uses linear scaling (StandardScaler and Min‑MaxScaler), which assumes linear transformations are sufficient for the price series. data
Source sentence
The approaches we developed for the present paper preprocess data using StandardScaler and Min‑MaxScaler, but these linear transformations fail to address regime-dependent volatility or trending means.
Data
- Brent oil price dataset market-prices unknown
Source sentence
The evaluation metric for the Crude Brent oil dataset resulting from the application of the three calibration methods is presented in Table 1.
- STOXX 600 index dataset market-prices unknown
Source sentence
The STOXX Europe 600 dataset (STOXX 600) Table 3 shows the results for the STOXX 600 dataset with the three calibration methods.
- IT40 index dataset market-prices unknown
Source sentence
Experimental validation on Brent oil prices, the STOXX 600 index, and the IT40 index demonstrated that our hybrid model outperformed benchmark methods.
Costs
Not applicable
Finds 2
- The hybrid LSTM‑Levy model calibrated with a neural network achieves MAE 0.00045 and R² 0.996976 on the Brent oil dataset, outperforming the standard LSTM (MAE 0.00210, R² 0.985889). performance
Source sentence
LSTM-L´evy NN 0.0004500 0.0000045 0.0021213 0.0000152 0.996976 ... LSTM -
- Across all three datasets, the GWO‑optimized hybrid model calibrated via ANN consistently outperforms standard LSTM and LSTM‑Fractional Heston benchmarks on multiple error metrics. predictive-accuracy
Source sentence
Our experimental results demonstrate that the hybrid model, combining a GWO-optimized LSTM network with the Levy-Merton Jump-Diffusion model calibrated using an ANN, outperformed the base LSTM model and all other models developed in this study.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 7 related papers). Coverage is partial, so this is neither replication nor refutation.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (25)
- Option pricing when underlying stock returns are discontinuous (1976) — link
- Multistep Brent oil price forecasting with a multi-aspect meta-heuristic optimization and ensemble deep learning model (2024)
- Analysis of Financial Time Series (2010) — link
- Investment under uncertainty (2012)
- Valuation of Commodity Futures and Options Under Stochastic Convenience Yields, Interest Rates, and Jump Diffusions in the Spot (1998)
- The Pricing of Options and Corporate Liabilities (1973) — link
- Empirical analysis of crude oil dynamics using affine vs. non-affine jump-diffusion models (2024) — link
- The jump-risk premia implicit in options: evidence from an integrated time-series study (2002) — link
and 17 more
- A Closed-Form Solution for Options with Stochastic Volatility with Applications to Bond and Currency Options (1993)
- Lévy processes and infinitely divisible distributions (1999)
- Post-‘87 crash fears in the S&P 500 futures option market (2000) — link
- An oil futures volatility forecast perspective on the selection of high-frequency jump tests (2022) — link
- Jumps and stochastic volatility in oil prices: Time series evidence (2011) — link
- A generalized Schwartz model for energy spot prices — Estimation using a particle MCMC method (2018) — link
- Jump-diffusion modelling of the gold and crude oil futures prices and predictive analysis of their economic impact (2022) — link
- Deep Calibration With Artificial Neural Network: A Performance Comparison on Option Pricing Models — → 2303.08760
- A neural network-based framework for financial model calibration (2019) — link
- A hybrid deep learning approach by integrating LSTM-ANN networks with GARCH model for copper price volatility prediction (2020) — link
- Enhancing Multi-step Brent Oil Price Forecasting with Ensemble Multi-scenario Bi-GRU Networks (2024)
- A hybrid deep learning approach by integrating LSTM-ANN networks with GARCH model for copper price volatility prediction (2020) — link
- Jump Diffusion-Informed Neural Networks with Transfer Learning for Accurate American Option Pricing under Data Scarcity (2024) — link
- Mean–variance portfolio optimization using machine learning-based stock price prediction (2021) — link
- A hybrid model for carbon price forecastingusing GARCH and long short-term memory network (2021) — link
- Marine Predators Algorithm: A nature-inspired metaheuristic (2020) — link
- Scalable gradients for stochastic differential equations (2020)
1 of 25 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.