Paper: arXiv 2601.03927

Authors: Vrinda Dhingra, Amita Sharma, Anubha Goel

Abstract

Index tracking, also known as passive investing, has gained significant traction in financial markets due to its cost-effective and efficient approach to replicating the performance of a specific market index. This review paper provides a comprehensive overview of the various modeling approaches and strategies developed for index tracking, highlighting the strengths and limitations of each approach. We categorize the index tracking models into three broad frameworks: optimization-based models, statistical-based models and machine learning based data-driven approach. A comprehensive empirical study conducted on the S&P 500 dataset demonstrates that the tracking error volatility model under the optimization-based framework delivers the most precise index tracking, the convex co-integration model, under the statistical-based framework achieves the strongest return-risk balance, and the deep neural network with fixed noise model within the data-driven framework provides a competitive performance with notably low turnover and high computational efficiency. By combining a critical review of the existing literature with comparative empirical analysis, this paper aims to provide insights into the evolving landscape of index tracking and its practical implications for investors and fund managers.

Complexity vs Empirical Score

  • Math Complexity: 6.5/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper involves optimization-based models with quadratic programming and combinatorial constraints (NP-hard problems) alongside statistical and machine learning frameworks, indicating significant mathematical density. It also presents a comprehensive empirical study on the S&P 500 with specific model comparisons and performance metrics, demonstrating strong data and implementation focus.

Research Flowchart

  flowchart TD
  A["Research Goal: Review Index Tracking Models<br>and Evaluate Performance"] --> B["Methodology: Categorize Models<br>into 3 Frameworks"]
  
  B --> C{"Data Input: S&P 500 Dataset"}
  
  C --> D["Computational Process 1:<br>Optimization-based Models"]
  C --> E["Computational Process 2:<br>Statistical-based Models"]
  C --> F["Computational Process 3:<br>Machine Learning/Deep Learning"]
  
  D --> G["Outcome 1: Tracking Error Volatility<br>Model achieves highest precision"]
  E --> H["Outcome 2: Convex Co-integration<br>Model offers best risk-return balance"]
  F --> I["Outcome 3: DNN with Fixed Noise<br>Model provides low turnover & high 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 comprehensive review of index‑tracking models across three paradigms and an empirical comparison on the S&P 500, highlighting that the tracking‑error variance model is most accurate, the convex co‑integration model has the best risk‑adjusted performance, and a deep neural networ

Source sentence
This review paper provides a comprehensive overview of the various modeling approaches and strategies developed for index tracking, highlighting the strengths and limitations of each approach.

Assumes 4

  • Both the index and constituent asset price series are integrated of order one (I(1)). statistical
    Source sentence
    The approach assumes that both the index and the constituent asset price series are integrated of order one, i.e., I(1), and that their logarithms can be linearly related through a co‑integrating vector.
  • Residuals from the co‑integration regression are stationary when the null of a unit root is rejected at the 5% level. statistical
    Source sentence
    Subsets for which the null of a unit root is rejected at the 5% level (p < 0.05) are retained as candidate co‑integrated portfolios.
  • The MSE tracking‑error function is convex and quadratic, leading to a convex quadratic programming problem. model
    Source sentence
    This objective function is convex and quadratic, resulting in the following convex quadratic programming problem: (MSE) Min
  • All underlying formulations are convex programs, so solutions are globally optimal when convergence occurs. model
    Source sentence
    It is worth noting that, since all underlying formulations are convex programs, the solutions reported are globally optimal when convergence occurs.

Data

  • S&P 500 dataset · Bloomberg benchmark commercial
    12 October 2012 to 8 November 2022 · S&P 500 constituents
    Source sentence
    We characterize the S&P 500 index and its constituents over the study period (12 October 2012 to 8 November 2022) using three descriptive statistics...

Costs

Partial Costs appear only as a robustness check, a flat fee without spread or impact, or outside the headline numbers.

Turnover is used as a proxy for transaction costs; the paper reports turnover ratios but does not embed explicit cost numbers in the performance results.

Source sentence
Turnover, a proxy for transaction costs, further differentiates the paradigms.

Finds 5

  • The tracking‑error variance (TEV) model achieves the lowest tracking error and highest correlation with the index among optimization‑based formulations. performance
    Source sentence
    TEV model delivers the most accurate replication, achieving the lowest tracking error and the highest correlation with the index.
  • The convex co‑integration (Cvx CoInt) model attains the strongest risk‑adjusted performance, with superior Sharpe, Sortino, and information ratios. performance
    Source sentence
    The convex cointegration (Cvx CoInt) model attains the strongest risk‑adjusted performance, as reflected in superior Sharpe, Sortino, and information ratios.
  • The deep neural network with fixed noise (DNNF) provides competitive replication accuracy while delivering substantially lower turnover and faster computation. performance
    Source sentence
    The deep neural network with fixed noise (DNNF) achieves competitive replication accuracy while offering substantial advantages in terms of turnover and computational efficiency.
  • Optimization‑based portfolios have higher turnover (median ~1.20) compared to statistical (median ~1.18) and data‑driven models (median ~0.87). performance
    Source sentence
    Turnover, a proxy for transaction costs, further differentiates the paradigms. Optimization‑based portfolios fall in the range 1.14–1.91 (median = 1.20)
  • Statistical models, especially Cvx CoInt, are computationally intensive (≈18 hours), whereas regression‑based and penalized regression models solve in seconds; data‑driven models are generally the most efficient. performance
    Source sentence
    All optimization models (except SES) are computationally intensive, requiring 17–18 hours on average per evaluation window

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 28 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (98)

  1. An evolutionary heuristic for the index tracking problem (2003) — link
  2. Mixed-integer programming approaches for index tracking and enhanced indexation (2009) — link
  3. Combining random forest and multicollinearity modeling for index tracking (2022) — link
  4. Active versus passive investing: Evidence from the 2009-2017 market (2018)
  5. The Shift from Active to Passive Investing: Risks to Financial Stability? (2020) — link
  6. Passive Investment Strategies and Efficient Markets (2003) — link
  7. A particle swarm optimization heuristic for the index tacking problem (2010) — link
  8. Optimal index tracking under transaction costs and impulse control (1998)
and 90 more
  1. Empirical insights on indexing: how capitalization, stratification and weighting can affect tracking error (1998)
  2. Portfolio optimization and index tracking for the shipping stock and freight markets using evolutionary algorithms (2013) — link
  3. Performance replication of the Spot Energy Index with optimal equity portfolio selection: Evidence from the UK, US and Brazilian markets (2014) — link
  4. A linear model for tracking error minimization (1999) — link
  5. Kernel Search: An application to the index tracking problem (2012) — link
  6. Clustering of financial time series with application to index and enhanced index tracking portfolio (2005) — link
  7. Tracking S&P 500 Index Funds (2001) — link
  8. Optimal portfolio selection and dynamic benchmark tracking (2005) — link
  9. Bond portfolio optimization problems and their applications to index tracking: A partial optimization approach (1996) — link
  10. A hybrid optimization approach to index tracking (2009) — link
  11. Factor based index tracking (2006) — link
  12. Nonnegative-lasso and application in index tracking (2014) — link
  13. Robust portfolio selection for index tracking (2012) — link
  14. Differential evolution and combinatorial search for constrained index-tracking (2009) — link
  15. Simultaneous pursuit of out-of-sample performance and sparsity in index tracking portfolios (2013) — link
  16. A methodology for index tracking based on time-series clustering (2004) — link
  17. A constrained cluster-based approach for tracking the S&P 500 index (2017) — link
  18. Index tracking and enhanced indexing using cointegration and correlation with endogenous portfolio selection (2017) — link
  19. On the index tracking and the statistical arbitrage choosing the stocks by means of cointegration: the role of stock picking (2015) — link
  20. Solving the index tracking problem based on a convex reformulation for cointegration (2020) — link
  21. Index tracking based on deep neural network (2019) — link
  22. Index tracking through deep latent representation learning (2020) — link
  23. Index tracking with cardinality constraints: A stochastic neural networks approach (2020)
  24. Neural network with fixed noise for index-tracking portfolio optimization (2021) — link
  25. Factor-based robust index tracking (2017) — link
  26. Optimal construction and rebalancing of index-tracking portfolios (2018) — link
  27. Cardinality versus q-norm constraints for index tracking (2014) — link
  28. Sparse Portfolios for High-Dimensional Financial Index Tracking (2018) — link
  29. Convex Optimization (2004)
  30. Support vector regression as conditional value-at-risk minimization with application to financial time-series analysis (2010) — link
  31. Conditional value-at-risk for general loss distributions (2002) — link
  32. A mean/variance analysis of tracking error (1992)
  33. Co-integration and error correction: Representation, estimation, and testing (1987) — link
  34. Some properties of time series data and their use in econometric model specification (1981) — link
  35. Portfolio selection (1952)
  36. Index tracking with controlled number of assets using a hybrid heuristic combining genetic algorithm and non-linear programming (2017) — link
  37. Meta-heuristic based decision support for portfolio optimization with a case study on tracking error minimization in passive portfolio management (2003) — link
  38. Nonnegative-lasso and application in index tracking (2014) — link
  39. Nonnegative Elastic Net and application in index tracking (2014) — link
  40. Lasso-based index tracking and statistical arbitrage long-short strategies (2020) — link
  41. Quantile regression for index tracking and enhanced indexation (2013) — link
  42. Efficient sparse portfolios based on composite quantile regression for high-dimensional index tracking (2020) — link
  43. Creating better tracking portfolios with quantiles (2022) — link
  44. Sparse Portfolio Selection via Topological Data Analysis based Clustering — → 2401.16920
  45. Sparse Index Tracking via Topological Learning — → 2310.09578
  46. Network-based index tracking using asset dependency structures (2026) — link
  47. Exact and heuristic approaches for the index tracking problem with UCITS constraints (2013) — link
  48. An efficient optimization approach for a cardinality-constrained index tracking problem (2016) — link
  49. An index tracking model with stratified sampling and optimal allocation (2018) — link
  50. Minimizing the tracking error of cardinality constrained portfolios (2018) — link
  51. Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk (1964)
  52. Efficient Capital Markets: A Review of Theory and Empirical Work (1970)
  53. Index tracking and enhanced indexing using mixed conditional value-at-risk (2018) — link
  54. A systematic literature review on solution approaches for the index tracking problem (2024) — link
  55. Artificial intelligence and machine learning in finance: Identifying foundations, themes, and research clusters from bibliometric analysis (2021) — link
  56. Optimal hedging using cointegration (1999) — link
  57. Stock-Index Tracking Optimization Using Auto-Encoders (2020) — link
  58. Risk-allocation-based index tracking (2023) — link
  59. Index optimization replication algorithm by using the soft subspace clustering method (2014) — link
  60. Index Tracking Based on Dynamic Time Warping and Constrained k-medoids Clustering (2021) — link
  61. Sparse Portfolio Selection via Topological Data Analysis based Clustering — → 2401.16920
  62. Accurate solution of the Index Tracking problem with a hybrid simulated annealing algorithm — → 2303.13282
  63. A fuzzy index tracking portfolio selection model (2005) — link
  64. Robust and sparse portfolio model for index tracking (2019) — link
  65. Lasso-based index tracking and statistical arbitrage long-short strategies (2020) — link
  66. The Performance of Mutual Funds in the Period 1945–1964 (1968)
  67. Support Vector Regression for Time Series Analysis (2011) — link
  68. Splitting criteria for ordinal decision trees: an experimental study (2024)
  69. Regression-Based Index Tracking Versus Clustering-Based Index Tracking: An Empirical Study (2025) — link
  70. Risk reduced sparse index tracking portfolio: A topological data analysis approach (2026) — link
  71. A robust possibilistic programming approach for selecting a clustering-based index tracking portfolio (2025) — link
  72. A benchmark-asset principal component factorization for index tracking on large investment universes (2025) — link
  73. Index Tracking Based on Norm-Constraints and Regularization (2025) — link
  74. Enhancing Sparse Index-Tracking Portfolios Using Deep Learning Models (2025) — link
  75. Harmony Search Based Metaheuristic for the Index Tracking Problem (2024) — link
  76. Optimizing index tracking: A Random Matrix Theory approach to portfolio selection (2025) — link
  77. Risk measure index tracking model (2022) — link
  78. A mixed 0-1 LP for index tracking problem with CVaR risk constraints (2012) — link
  79. Index tracking by using sparse support vector regression (2017) — link
  80. Combining random forest and multicollinearity modeling for index tracking (2022) — link
  81. Reinforcement Learning for Financial Index Tracking — → 2308.02820
  82. An integrative extraction approach for index-tracking portfolio construction and forecasting under a deep learning framework (2024) — link
  83. Fast and stable portfolios through Huber’s criterion for constrained index tracking (2025) — link
  84. How the index fund was born (2011)
  85. SPDR S&P 500 ETF Trust (SPY) (1993) — link
  86. Deep learning for enhanced index tracking (2024)
  87. Reinforcement Learning for Financial Index Tracking — → 2308.02820
  88. Index tracking with cardinality constraints: A stochastic neural networks approach (2020)
  89. Index tracking using shapley additive explanations and one-dimensional pointwise convolutional autoencoders (2024)
  90. State-dependent stock selection in index tracking: a machine learning approach (2022)

6 of 98 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.