Quantitative Investment Based on Fundamental Analysis – A Case Study of the Securities Industry

·

The integration of fundamental analysis and quantitative investing has emerged as a powerful approach in modern finance, combining the depth of traditional value investing with the precision of data-driven models. This article explores how fundamental factors can be systematically incorporated into quantitative strategies, using China’s securities industry as a real-world case study. By leveraging financial metrics, predictive modeling, and portfolio optimization techniques, we demonstrate a structured framework for identifying high-potential stocks and constructing efficient investment portfolios.

Understanding Fundamental-Based Quantitative Investing

Fundamental-based quantitative investing bridges the gap between traditional value analysis and algorithmic trading. Rather than relying solely on historical price patterns or subjective judgment, this method evaluates companies based on intrinsic financial health while applying statistical rigor to identify patterns that drive excess returns.

At its core, the strategy hinges on analyzing relationships between key financial indicators—such as earnings per share, price-to-earnings ratios, and net asset value—and future stock performance. By transforming qualitative insights into quantifiable data points, investors gain a more objective, scalable, and repeatable decision-making process.

This hybrid model offers several advantages:

👉 Discover how advanced analytics are reshaping modern investment strategies

Research Framework and Methodology

To evaluate the effectiveness of fundamental-based quantitative investing in the securities sector, we employ a three-stage analytical workflow: data preparation, stock selection via multi-criteria evaluation, and portfolio optimization.

Time Series Moving Average for Forecasting

Given the temporal dependencies in financial data, we apply a time series moving average model to forecast next-month values of six key indicators for 38 selected securities firms:

Using historical data from January 2020 to December 2021, we compute simple moving averages to smooth out noise and reveal underlying trends. For beta—a measure of volatility—we invert the value to align it with positive performance metrics (lower beta = lower risk = better).

These forecasts serve as inputs for subsequent ranking and scoring models.

Entropy-Weighted TOPSIS for Stock Ranking

To objectively rank stocks based on multiple criteria, we implement the Entropy-Weighted TOPSIS method, which combines two powerful techniques:

  1. Entropy weighting calculates objective weights for each indicator based on data dispersion. Metrics with higher variability across firms receive greater weight, reflecting their discriminatory power.
  2. TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) ranks alternatives by measuring their distance from an ideal best-case scenario and worst-case scenario.

Step-by-step Process:

  1. Normalize raw forecast data using min-max scaling.
  2. Compute entropy values for each metric to derive objective weights.
  3. Identify positive (best) and negative (worst) ideal solutions.
  4. Calculate Euclidean distances from each stock to both ideal points.
  5. Derive a composite score representing proximity to optimal performance.

This method avoids subjective weighting biases and enhances model transparency compared to arbitrary scoring systems.

After applying this model, the top five securities firms by composite score are:

  1. CITIC Securities
  2. GF Securities
  3. Guosen Securities
  4. Orient Securities
  5. Dongxing Securities

These names represent large-cap, well-established players in China’s brokerage landscape—consistent with expectations for fundamentally strong performers.

Portfolio Construction Using Mean-Variance Optimization

With a shortlist of high-scoring stocks, the next step is determining optimal allocation weights to balance return and risk. We use the Markowitz Mean-Variance Model, a cornerstone of modern portfolio theory introduced in 1952.

Key Concepts

By solving a multi-objective optimization problem—maximizing return while minimizing variance—we identify efficient portfolio combinations.

Empirical Results

Using LINGO and MATLAB software, we calculate:

The resulting efficient frontier graph illustrates the trade-off between risk (x-axis) and return (y-axis). Each point on the curve represents an optimally balanced portfolio.

For instance, at a 9% risk level (standard deviation), the model achieves a 6.5% expected monthly return—a compelling risk-adjusted outcome.

According to the model, one optimal combination involves allocating:

This split maximizes expected return under the given assumptions.

👉 Explore how data-driven portfolio design enhances investment outcomes

Real-World Performance vs. Model Predictions

Despite strong theoretical results, actual market performance diverged significantly from predictions.

In January 2022:

The actual portfolio loss was approximately –7.79%, contrary to the model's positive forecast.

Why the Discrepancy?

Several external factors explain the gap:

This highlights a critical insight: even sophisticated models cannot fully anticipate systemic shocks or behavioral market dynamics.

Core Keywords in Fundamental Quantitative Investing

To enhance search visibility and reader engagement, the following core keywords are naturally integrated throughout this article:

These terms reflect central themes in algorithmic equity investing and align with common investor search queries related to smart investing strategies.

Frequently Asked Questions

What is fundamental-based quantitative investing?

It’s an investment strategy that combines traditional financial statement analysis with statistical modeling to identify undervalued or high-potential stocks. Instead of relying purely on intuition or pure price-based algorithms, it uses fundamental metrics like EPS, P/E ratio, and ROE as inputs into systematic trading models.

How does entropy-weighted TOPSIS improve stock evaluation?

Unlike subjective weighting schemes, entropy weighting assigns importance based on data variation—ensuring that more discriminative metrics have greater influence. Combined with TOPSIS ranking, it provides an objective, transparent way to compare companies across multiple dimensions without human bias.

Can quantitative models predict market crashes?

Most traditional models—including mean-variance optimization—assume stable market conditions and normal return distributions. They often fail during extreme events like crashes or bubbles. Advanced approaches incorporating machine learning, sentiment analysis, or regime-switching models may offer better resilience but still carry uncertainty.

Why did the predicted returns differ so much from real results?

The model focused only on historical return patterns and volatility, ignoring macro-level influences such as policy changes, global events, or investor psychology. In early 2022, negative sentiment across Chinese equities led to broad declines—even among fundamentally sound firms—demonstrating the limits of isolated quantitative models.

Is fundamental quant investing suitable for individual investors?

Yes—with proper tools and discipline. Retail investors can adopt simplified versions using screeners based on P/E, EPS growth, or ROE thresholds. Platforms offering backtesting capabilities allow users to validate strategies before deployment.

How can I reduce prediction errors in quantitative models?

Improve accuracy by:

👉 Learn how real-time analytics empower smarter investment decisions

Final Thoughts and Future Directions

This study confirms that integrating fundamental analysis into quantitative frameworks provides a structured path toward informed investing. While no model guarantees success—especially in volatile markets—the process enhances objectivity and consistency.

Key takeaways include:

Future enhancements could involve integrating machine learning algorithms—such as random forests or neural networks—to capture non-linear relationships among financial variables. Additionally, expanding factor sets to include ESG metrics or analyst sentiment could further refine selection accuracy.

Ultimately, successful investing lies not in perfect prediction—but in disciplined execution, continuous learning, and adaptive strategy refinement.