Paper: arXiv 2510.15921

Authors: Amarendra Mohan, Ameer Tamoor Khan, Shuai Li, Xinwei Cao, Zhibin Li

Abstract

Cross-market portfolio optimization has become increasingly complex with the globalization of financial markets and the growth of high-frequency, multi-dimensional datasets. Traditional artificial neural networks, while effective in certain portfolio management tasks, often incur substantial computational overhead and lack the temporal processing capabilities required for large-scale, multi-market data. This study investigates the application of Spiking Neural Networks (SNNs) for cross-market portfolio optimization, leveraging neuromorphic computing principles to process equity data from both the Indian (Nifty 500) and US (S&P 500) markets. A five-year dataset comprising approximately 1,250 trading days of daily stock prices was systematically collected via the Yahoo Finance API. The proposed framework integrates Leaky Integrate-andFire neuron dynamics with adaptive thresholding, spike-timingdependent plasticity, and lateral inhibition to enable event-driven processing of financial time series. Dimensionality reduction is achieved through hierarchical clustering, while populationbased spike encoding and multiple decoding strategies support robust portfolio construction under realistic trading constraints, including cardinality limits, transaction costs, and adaptive risk aversion. Experimental evaluation demonstrates that the SNN-based framework delivers superior risk-adjusted returns and reduced volatility compared to ANN benchmarks, while substantially improving computational efficiency. These findings highlight the promise of neuromorphic computation for scalable, efficient, and robust portfolio optimization across global financial markets.

Complexity vs Empirical Score

  • Math Complexity: 9.0/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced neuromorphic theory including spiking neural networks, Leaky Integrate-and-Fire dynamics, and spike-timing-dependent plasticity, giving high mathematical complexity. It also demonstrates strong empirical rigor with a five-year dataset from Yahoo Finance, backtesting against ANN benchmarks, and reporting of risk-adjusted returns and volatility metrics.

Research Flowchart

  flowchart TD
  A["Research Goal: Apply SNNs to Cross-Market Portfolio Optimization"] --> B["Data Collection: 5-Year Equity Data<br>Nifty 500 & S&P 500 via Yahoo Finance"]
  B --> C["SNN Architecture Design<br>Leaky Integrate-and-Fire with STDP & Lateral Inhibition"]
  C --> D["Event-Driven Processing & Optimization<br>Hierarchical Clustering, Population Encoding, Risk Constraints"]
  D --> E["Experimental Evaluation<br>SNN vs. ANN Benchmarks"]
  E --> F["Findings: Superior Risk-Adjusted Returns<br>Reduced Volatility & Computational Efficiency"]

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 spiking neural network framework for cross‑market portfolio optimization that leverages neuromorphic computing to process equity data from Indian and US markets.

Source sentence
This study investigates the application of Spiking Neural Networks (SNNs) for cross‑market portfolio optimization, leveraging neuromorphic computing principles to process equity data from both the Indian (Nifty 500) and US (S&P 500) markets.

Assumes 6

  • Portfolio cardinality limits are imposed (Kmin ≤ number of non‑zero assets ≤ Kmax). execution
    Source sentence
    Portfolio cardinality refers to the number of assets held in the portfolio. Practical considerations such as transaction costs, monitoring requirements, and regulatory constraints often necessitate limits on portfolio size: Kmin ≤∥{i : wi > 0}∥≤Kmax
  • Transaction costs are proportional to the absolute change in asset weights. execution
    Source sentence
    We model proportional transaction costs as: C(w, wprev) = ctc
  • An adaptive risk‑aversion parameter λ(t) decays exponentially over optimization iterations. model
    Source sentence
    where λ0 is the initial risk aversion level, α controls the decay rate
  • Assets with less than 90% data completeness are removed from the sample. data
    Source sentence
    Assets exhibiting data completeness below 90% of the total observation period are systematically removed to ensure robust statistical inference.
Show 2 more
  • Daily closing prices are used as the primary price signal. data
    Source sentence
    The raw closing price data is conv…
  • The market returns are assumed to be non‑Gaussian and time‑varying, motivating the use of SNNs over traditional mean‑variance methods. market
    Source sentence
    their practical limitations become apparent when confronted with real‑world market dynamics characterized by non‑Gaussian return distributions, time‑varying volatility, and complex inter‑market dependencies

Data

Backtest sample: Indian (Nifty 500) and US (S&P 500) equities · multi-asset · daily

  • Nifty 500 and S&P 500 daily price dataset · Yahoo Finance (via yfinance Python library) market-prices public
    approximately 5 years (≈1,250 trading da · Nifty 500 and S&P 500 constituents
    Source sentence
    A five‑year dataset comprising approximately 1,250 trading days of daily stock prices was systematically collected via the Yahoo Finance API.

Recognised sources: Yahoo Finance

Costs

Modeled Costs are part of the reported headline results.

Proportional transaction costs with rate ctc applied to the absolute change in portfolio weights.

Source sentence
We model proportional transaction costs as: C(w, wprev) = ctc

Finds 5

Reported performance (Spiking Neural Network for cross‑market portfolio optimization): Sharpe 2.08 (benchmark 1.94) · total return 42.7%

  • The SNN framework delivers superior risk‑adjusted returns and lower volatility than ANN benchmarks. performance
    Source sentence
    Experimental evaluation demonstrates that the SNN‑based framework delivers superior risk‑adjusted returns and reduced volatility compared to ANN benchmarks
  • Mean rolling Sharpe ratio of the SNN is 2.082, exceeding the ANN by 7.1%. performance
    Source sentence
    The mean rolling Sharpe ratio of 2.082 exceeds the ANN performance by 7.1%
  • SNN portfolios have higher diversification: average cardinality 42.3 vs 38.7 assets, lower concentration index (0.043 vs 0.051). performance
    Source sentence
    Average Portfolio Cardinality 42.3 38.7 +3.6 assets
  • Training the SNN requires more epochs (250 vs 150) but achieves a final loss (negative Sharpe) of -2.81 vs -1.95 for ANN, a 44.1% improvement. performance
    Source sentence
    Training Epochs Required 250 150 +100 epochs … Final Loss (Negative Sharpe) -2.81 -1.95 -44.1%
  • Computational complexity of the SNN is O(N·S) versus O(N²) for ANN, indicating higher efficiency. performance
    Source sentence
    Computational Complexity O(N · S) O(N2) Sparse advantage

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 60 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.