Paper: arXiv 2501.14736
Authors: Li-Chun Huang
Abstract
In this study, we applied the NEAT (NeuroEvolution of Augmenting Topologies) algorithm to stock trading using multiple technical indicators. Our approach focused on maximizing earning, avoiding risk, and outperforming the Buy & Hold strategy. We used progressive training data and a multi-objective fitness function to guide the evolution of the population towards these objectives. The results of our study showed that the NEAT model achieved similar returns to the Buy & Hold strategy, but with lower risk exposure and greater stability. We also identified some challenges in the training process, including the presence of a large number of unused nodes and connections in the model architecture. In future work, it may be worthwhile to explore ways to improve the NEAT algorithm and apply it to shorter interval data in order to assess the potential impact on performance.
Complexity vs Empirical Score
- Math Complexity: 6.0/10
- Empirical Rigor: 6.0/10 (capped from 7.5: transaction costs not modelled; see penalty flags)
- Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 7.5)
Why this score: The paper presents concrete empirical implementation details including dataset size, specific libraries, and a fitness function, resulting in high empirical rigor, while the mathematical complexity is moderate due to the use of standard technical indicator formulas rather than advanced derivations.
Research Flowchart
flowchart TD A["Research Goal: Apply NEAT Algorithm to Stock Trading for Max Earnings & Min Risk"] --> B["Methodology: Multi-Objective Fitness Function & Progressive Training Data"] B --> C["Data: Technical Indicators as Inputs & Equity Price History"] C --> D["Computational Process: NEAT Evolution of Population"] D --> E["Outcome: Model Trained & Tested vs. Buy & Hold Strategy"] E --> F["Key Findings: Comparable Returns, Lower Risk, Greater Stability"] F --> G["Challenge: High Sparsity (Unused Nodes/Connections)"] G --> H["Future Work: Apply to Shorter Interval Data & Optimize NEAT"]
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 NEAT-based stock trading strategy that uses multiple technical indicators and a multi‑objective fitness function to maximize earnings, minimize risk and outperform a buy‑and‑hold baseline.
Source sentence
In this study, we applied the NEAT (NeuroEvolution of Augmenting Topologies) algorithm to stock trading using multiple technical indicators.
Assumes 3
- The data are sampled at a daily frequency (1‑day interval). data
Source sentence
... has a 1‑day interval for each row, resulting in a dataset with 2580890 rows.
- Buy & Hold is used as the baseline for performance comparison. market
Source sentence
We compare the performance of our NEAT‑based approach to a buy and hold strategy, which serves as a baseline for comparison.
- The training process uses progressive training data (shorter windows early, longer later). model
Source sentence
In our case, the model is trained on 90‑days data in the first 1500 generations
Data
stocks · equities · daily
- 22 years of historical stock trading data market-prices public
Source sentence
The dataset used in this project sources from yahoo finance API, which consists of 22 years of historical stock trading data for the 503 constituents of the S&P 500 index, covering the period from 1999/12/31 to 2022/11/27.
Costs
Not modeled
Finds 3
annual return 18.8%
- The NEAT model achieved returns similar to the Buy & Hold strategy while exhibiting lower risk exposure and greater stability. performance
Source sentence
The results of our study showed that the NEAT model achieved similar returns to the Buy & Hold strategy, but with lower risk exposure and greater stability.
- Although the model’s average return was lower than Buy & Hold, it maintained approximately the same win rate and reduced risk due to shorter exposure time. performance
Source sentence
While the average return for the model’s strategy was lower, it had approximately the same win rate and lower risk exposure due to the shorter exposure time.
- Training revealed a large number of unused nodes, connections and inputs in the evolved neural networks. negative
Source sentence
We also identified some challenges in the training process, including the presence of a large number of unused nodes and connections in the model architecture.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 42 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 (7)
- Efficient evolution of neural network topologies (2002) — link
- SQLite (2020) — link
- neat-python
- Image Classification Using Generative Neuro Evolution for Deep Learning (2015) — link
- Predicting stock and stock price index movement using Trend Deterministic Data Preparation and machine learning techniques (2015) — link
- Using NEAT for continuous adaptation and teamwork formation in Pacman (2008) — link
- A Neuroevolution Approach to General Atari Game Playing (2014) — link
0 of 7 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.