Paper: arXiv 2508.13429

Authors: Paulo André Lima de Castro

Abstract

Autonomous trading strategies have been a subject of research within the field of artificial intelligence (AI) for aconsiderable period. Various AI techniques have been explored to develop autonomous agents capable of trading financial assets. These approaches encompass traditional methods such as neural networks, fuzzy logic, and reinforcement learning, as well as more recent advancements, including deep neural networks and deep reinforcement learning. Many developers report success in creating strategies that exhibit strong performance during simulations using historical price data, a process commonly referred to as backtesting. However, when these strategies are deployed in real markets, their performance often deteriorates, particularly in terms of risk-adjusted returns. In this study, we propose an AI-based strategy inspired by a classical investment paradigm: Value Investing. Financial AI models are highly susceptible to lookahead bias and other forms of bias that can significantly inflate performance in backtesting compared to live trading conditions. To address this issue, we conducted a series of computational simulations while controlling for these biases, thereby reducing the risk of overfitting. Our results indicate that the proposed approach outperforms major Brazilian market benchmarks. Moreover, the strategy, named AlphaX, demonstrated superior performance relative to widely used technical indicators such as the Relative Strength Index (RSI) and Money Flow Index (MFI), with statistically significant results. Finally, we discuss several open challenges and highlight emerging technologies in qualitative analysis that may contribute to the development of a comprehensive AI-based Value Investing framework in the future

Complexity vs Empirical Score

  • Math Complexity: 4.0/10
  • Empirical Rigor: 6.0/10 (capped from 7.5: transaction costs not modelled; see penalty flags)
  • Quadrant: Street Traders — empirical and applied, lower math complexity (effective empirical rigor capped from 7.5)

Why this score: The paper applies standard value investing metrics (P/E, P/B) and AI models without heavy mathematical derivations, but it emphasizes bias-controlled backtesting and performance metrics against Brazilian benchmarks, indicating a practical, implementation-focused approach.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Develop AI Value Investing Strategy<br>for Brazilian Stock Market"] --> B
  subgraph B ["Methodology & Data"]
      B1["Historical Price Data<br>Brazilian Equities"] --> B2{"Computational Simulations"}
      B2 --> B3["Value Investing Logic + AI<br>Reinforcement Learning"]
      B4["Bias Control &<br>Backtesting Validation"] --> B2
  end
  B --> C
  subgraph C ["Key Findings"]
      C1["Outperforms Major<br>Brazilian Benchmarks"]
      C2["Superior to Technical<br>Indicators RSI & MFI"]
      C3["Statistically Significant<br>Risk-Adjusted Returns"]
  end
  C --> D["Outcome:<br>AlphaX Strategy Framework<br>Ready for Deployment"]

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 an AI‑based value investing strategy for Brazil called AlphaX.

Source sentence
In this study, we propose an AI‑based strategy inspired by a classical investment paradigm: Value Investing.

Assumes 6

  • Financial statements become available two months after quarter end. data
    Source sentence
    are available two months after the end of the quarter and not at the end of the quarter.
  • Positions are closed and opened on the day of disclosure at the average of the day’s high and low. execution
    Source sentence
    position change operations are made on the day of disclosure at the average price of the day (average of the maximum and minimum of the day).
  • Investments are held for at least three months; recommendations are made quarterly. market
    Source sentence
    Recommendations are made per quarter, since companies’ balance sheet data is released with this frequency. However, it is recommended that the investor aim for longer periods, making adjustments to the portfolio quarterly if needed.
  • Maximum of four assets may be selected for investment at any time. model
    Source sentence
    In our tests, we set X to 4, so up to four assets may be selected to investment.
Show 2 more
  • Selected companies must have at least five years of historical data and high liquidity. data
    Source sentence
    In order to be selected, a company must have a long historical track record (a minimum of five years) and high liquidity (listed on B3’s main inde…
  • The strategy aims to outperform the Selic rate and Ibovespa benchmarks. market
    Source sentence
    Its primary objective is to achieve returns that exceed the two main benchmarks of the Brazilian market—Selic (closely aligned with CDI which is Brazil’s benchmark interest rate) and Ibovespa.

Data

Backtest sample: 2020-10 – 2025-03 · Brazilian equities · equities

  • B3 daily price data · B3 market-prices public
    Feb 2021 to May 2025 · Stocks traded on B3 meeting liquidity criteria
    Source sentence
    Market data comes from B3 which is composed by open, maximum, minimum and close daily prices. The target assets were selected from stock traded on B3.
  • CVM financial statements · CVM fundamentals public
    Feb 2021 to May 2025 · Companies listed on B3 with at least five years of history
    Source sentence
    ... It gets financial statements data about the target companies from CVM including Revenue, Operating Expenses, Gross Profit, EBIT, Net Income, Assets, Liabilities, Equity, Operation Cash Flow and so on.

Costs

Not modeled A trading or execution result is reported without costs.

Source sentence
position change operations are made on the day of disclosure at the average price of the day (average of the maximum and minimum of the day).

Finds 3

Reported performance (AlphaX): Sharpe 0.98 · annual return 16.9% · total return 97.9% · max drawdown 20.8%

  • AlphaX outperforms major Brazilian benchmarks and technical indicators (RSI, MFI) in return, drawdown and risk‑adjusted metrics. performance
    Source sentence
    AlphaX has shown superior performance compared to all of them in terms of Return (CAGR, Total Return), Risk (Max Drawdown) and Risk‑adjusted returns (An. Sharpe Ratio and An. Sortino), as presented in Table III.
  • AlphaX’s estimated Sharpe Ratio is above zero with 97.4% confidence (PSR(0)) and above 0.01 with 95% confidence, outperforming all compared technical strategies. statistical
    Source sentence
    In Table IV, the results show that AlphaX has an estimated Sharpe Ratio above zero, PSR(0), with 97.4% level of confidence and above 0.01 with 95% level of confidence.
  • MFI achieved the second‑best performance with an estimated Sharpe ratio above zero at 81% confidence, but no technical strategy reached 95% confidence. performance
    Source sentence
    MFI presented the second best performance with an estimated Sharpe ratio above zero with just 81% level of confidence. No technical strategy reaches a 95% level of confidence.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 19 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.