Paper: arXiv 2411.13564
Authors: Krishna Neupane, Igor Griva
Abstract
According to The Exchange Act, 1934 unlawful insider trading is the abuse of access to privileged corporate information. While a blurred line between “routine” the “opportunistic” insider trading exists, detection of strategies that insiders mold to maneuver fair market prices to their advantage is an uphill battle for hand-engineered approaches. In the context of detailed high-dimensional financial and trade data that are structurally built by multiple covariates, in this study, we explore, implement and provide detailed comparison to the existing study (Deng et al. (2019)) and independently implement automated end-to-end state-of-art methods by integrating principal component analysis to the random forest (PCA-RF) followed by a standalone random forest (RF) with 320 and 3984 randomly selected, semi-manually labeled and normalized transactions from multiple industry. The settings successfully uncover latent structures and detect unlawful insider trading. Among the multiple scenarios, our best-performing model accurately classified 96.43 percent of transactions. Among all transactions the models find 95.47 lawful as lawful and \(98.00\) unlawful as unlawful percent. Besides, the model makes very few mistakes in classifying lawful as unlawful by missing only 2.00 percent. In addition to the classification task, model generated Gini Impurity based features ranking, our analysis show ownership and governance related features based on permutation values play important roles. In summary, a simple yet powerful automated end-to-end method relieves labor-intensive activities to redirect resources to enhance rule-making and tracking the uncaptured unlawful insider trading transactions. We emphasize that developed financial and trading features are capable of uncovering fraudulent behaviors.
Complexity vs Empirical Score
- Math Complexity: 4.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Street Traders — empirical and applied, lower math complexity
Why this score: The paper employs established machine learning techniques (PCA, Random Forest) with relatively accessible mathematical foundations, resulting in a moderate math score. However, it demonstrates high empirical rigor through the use of a real-world financial dataset (3,304 transactions), explicit cross-validation, detailed performance metrics (96.43% accuracy, confusion matrix), and a focus on feature importance analysis, making it highly backtest-ready.
Research Flowchart
flowchart TD
A["Research Goal:<br>Detect unlawful insider trading"] --> B["Data Acquisition & Processing"]
subgraph B ["Data/Inputs"]
B1["320 & 3984 Transactions<br>Multi-industry data"]
B2["Labeling & Normalization"]
end
B --> C["Key Methodology"]
subgraph C ["Methodology"]
C1["PCA-RF Model"]
C2["Standalone RF Model"]
end
C --> D["Computational Process"]
subgraph D ["Process"]
D1["PCA Dimensionality Reduction"]
D2["Random Forest Classification"]
D3["Gini Impurity Feature Ranking"]
end
D --> E["Key Findings/Outcomes"]
subgraph E ["Results"]
E1["96.43% Classification Accuracy"]
E2["95.47% Lawful, 98% Unlawful<br>Detection Rates"]
E3["2% False Positive Rate<br>Ownership/Governance Features<br>Most Predictive"]
end
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 end‑to‑end method that integrates principal component analysis with a random forest (PCA‑RF) followed by a standalone random forest for detecting unlawful insider trading.
Source sentence
integrating principal component analysis to the random forest (PCA-RF) followed by a standalone random forest (RF) with 320 and 3984 randomly selected, semi-manually labeled and normalized transactions
Assumes 4
- Experiments are conducted on a balanced dataset with a 0.5:0.5 ratio of lawful to unlawful transactions. data
Source sentence
We experiment in the balanced dataset with 0.5: 0.5 ratio of lawful to unlawful respectively for 320 and 3984 transactions.
- Qualitative features with subjective prefixes are excluded to avoid spuriousness. other
Source sentence
With the possibility of subjective interpretations, it may inadvertently introduce spuriousness.
- Random selection of transactions from different industrial groups improves objectivity compared to selecting within a single industry. data
Source sentence
In contrast to DCZ study, for wider applicability, we randomly selected transactions from different industrial groups thereby boosting objectivity.
- One‑hot encoding assumes independence (orthogonality) of categorical features, which may break down as cardinality grows. model
Source sentence
(that is, orthogonality) assumption of one-hot may no longer hold as the feature space becomes crowded and correlation between the features is inevitable.
Data
- Insider transaction dataset alternative unknown
Source sentence
We experiment in the balanced dataset with 0.5: 0.5 ratio of lawful to unlawful respectively for 320 and 3984 transactions.
Compustat SEC EDGAR
Costs
Not applicable
Finds 3
- The best‑performing model accurately classified 96.43% of transactions. performance
Source sentence
our best-performing model accurately classified 96.43 percent of transactions.
- Test accuracy averaged 99.13% with PCA and 98.13% without PCA. predictive-accuracy
Source sentence
high very test accuracy averaging at 99.13 (with PCA), 98.13 (without PCA) in Table 5.
- The model achieved high specificity, correctly identifying unlawful transactions while producing few false negatives. performance
Source sentence
High Specificity: The model correctly identified unlawful as unlawful which excludes the lawful transactions resulting higher number of true unlawful (true negatives) and a low number of unlawful transactions that are wrongly classified as lawful.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 17 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
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.