Pith. sign in

REVIEW 3 major objections 6 minor 20 references

An unsupervised K-Means++ pipeline flags 2.02 percent of roughly one million Dhaka Stock Exchange trades as suspicious and sorts them into spoofing, pump-and-dump, insider trading, fake breakout, or unclassified.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

K-Means++ plus percentile and price-change heuristics flag 2.02% of ~1M DSE trades as suspicious and assign mostly spoofing or unclassified labels, with only a 0.561 silhouette score as validation.

T0 review reviewed 2026-07-11 challenge →

load-bearing objection Solid engineering screen of ~1M DSE trades with a hybrid K-Means++ + heuristic pipeline, but the fraud-detection claim is circular and unsupported by any external validation. the 3 major comments →

arxiv 2607.04184 v1 pith:EVDLNAIH submitted 2026-07-05 cs.AI cs.LG

A Clustering-Based Framework for Identifying Suspicious Trading Patterns in Capital Market

Asif Zaman , Romona Magdalene Sarkar , Sabiha Khair Ohi , Iftekharul Mobin This is my paper
classification cs.AI cs.LG
keywords market manipulationK-Means++ clusteringunsupervised fraud detectionstock marketspoofinganomaly detectionDhaka Stock Exchangebehavioral heuristics
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that market manipulation can be surfaced without labeled fraud examples by clustering daily stock features and then filtering outliers with simple trading heuristics. Using about one million trades from the Dhaka Stock Exchange spanning 2012–2024, the authors build nine rolling-window indicators, run K-Means++ into five clusters, and mark a trade suspicious only when it is far from its cluster center and also shows extreme price or volume behavior. The method labels 2.02 percent of trades as suspicious; among those, spoofing accounts for just over half while the rest split among known manipulation patterns or remain unclassified. A silhouette score of 0.561 is offered as evidence that the clusters are coherent even though no ground-truth labels exist. A sympathetic reader cares because the same lightweight pipeline can be dropped onto other unlabeled exchanges and can produce risk scores and fraud-type tags that regulators or platforms can inspect further.

Core claim

A hybrid unsupervised pipeline that first forms natural trading clusters with K-Means++ and then intersects distance-based outliers with percentile and price-change heuristics recovers a small, interpretable set of suspicious trades (2.02 percent) that can be further labeled as spoofing, pump-and-dump, insider trading, fake breakout, or unclassified, all without any confirmed fraud labels.

What carries the argument

The Stock Market Manipulation Detection (SMMD) algorithm: K-Means++ (k=5) on standardized 30-day features, followed by a 95th-percentile distance threshold conjoined with behavioral rules (price move >10 percent or volume/trade spikes above the 95th percentile) and a five-day lookahead for pattern labeling.

Load-bearing premise

That being far from a K-Means cluster center plus crossing fixed price or volume thresholds is a reliable stand-in for real market manipulation when no confirmed fraud cases exist to check the false-positive rate.

What would settle it

Obtain a set of independently verified manipulation cases (or regulatory sanctions) from the same 2012–2024 DSE period and measure what fraction of them fall inside the 2.02 percent flagged set versus how many flagged trades have no corresponding sanction.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper proposes an unsupervised Stock Market Manipulation Detection (SMMD) pipeline that applies K-Means++ (k=5 chosen by elbow) to nine 30-day rolling technical features extracted from ~1.02M daily DSE trades (2012–2024). Structural outliers are defined as points whose Euclidean distance to the assigned centroid exceeds the 95th percentile of distances; a trade is labeled suspicious only if it is also a structural outlier and satisfies at least one fixed behavioral rule (|ΔP%|>10, volume/trade/turnover spikes above the 95th percentile). Flagged trades are then typed by a second table of the same price/volume/lookahead heuristics into spoofing (51.10%), pump-and-dump (0.10%), insider trading (0.55%), fake breakout (1.43%), and unclassified (46.83%), for an overall flag rate of 2.02%. Symbol-level suspicion scores and adaptive risk bins are derived from flag frequency and distance percentile rank. Cluster quality is reported via Silhouette scores of 0.561 (DSE) and 0.292 (NSE); no labeled fraud cases are available.

Significance. If the hybrid distance-plus-heuristic flags were shown to recover genuine manipulation events at usable precision, the work would supply a lightweight, label-free screening tool for emerging markets such as the DSE, where supervised detectors are impractical. The engineering effort (feature construction, dual-exchange visualization, risk scoring) is concrete and potentially reusable. However, the manuscript currently offers no external validation that the flags correspond to real market abuse; the reported percentages are therefore definitional outputs of the chosen thresholds rather than measured detection performance. The contribution is therefore best viewed as a transparent heuristic pipeline whose practical value remains unproven.

major comments (3)
  1. Abstract and §IV.G present Silhouette 0.561 as confirmation of fraud-detection performance. Silhouette only quantifies separation of the engineered feature clusters; it does not measure precision, recall, or false-positive rate of the subsequent hybrid flags. With no ground-truth labels, known-event checks, synthetic injection, or pure-rule ablation, the claim that the pipeline “identifies fraudulent trades” is unsupported. At minimum the abstract and evaluation sections must restate the metric as a clustering-quality diagnostic and remove any implication that it validates fraud detection.
  2. Algorithm 1 (anomaly block) and Table III define both the suspicious label and the fraud-type labels by the same fixed price/volume/lookahead rules conjoined with a 95th-percentile distance cut. Consequently the reported 2.02% rate and the 51.10%/0.10%/etc. breakdown are largely definitional. Contribution 3 further cites accuracy 0.987 and silhouette 0.965 from an external reference [7] as if they were obtained on the present data. Either an independent validation (regulator cases, news-event alignment, or controlled synthetic injection) or a clear reframing as a pure heuristic screening tool is required before the central claim can stand.
  3. Nearly half (46.83%) of the flagged trades remain “unclassified.” Combined with the circular definition of the remaining classes, this large residual undermines the claim of “interpretable fraud-type categorization aligned with real-world manipulation patterns” (contribution 4). The manuscript should either refine the rule set so that the residual is small or explicitly treat the unclassified mass as an open limitation rather than a successful categorization result.
minor comments (6)
  1. Contribution 3 asserts that “K-Means outperforms DBSCAN, OPTICS, and hierarchical clustering in accuracy (0.987), silhouette score (0.965)”; these numbers are taken from [7] and are not measured on the DSE/NSE data used here. The sentence should be rewritten or moved to related work.
  2. Inconsistency in year ranges: Algorithm 1 Require line mentions Excel sheets 2010–2020/2021–2024 and exclusion of 2010–2011, while the abstract and body consistently state 2012–2024. Clarify the exact date window.
  3. Equation (2) introduces a free weight α for the suspicion score, yet no value (or sensitivity analysis) is reported; the conclusion later alludes to a 60/40 split without derivation. State the chosen α and justify it.
  4. Figures 3 and 4 caption dates differ (2012–2024 vs 2012–2025); align captions with the data actually used.
  5. Several references contain placeholder page numbers (XX–XX) and incomplete venue information; these should be completed before camera-ready.
  6. Typographical issues: “LITERATUREREVIEW” and “RESEARCHMETHODOLOGY” lack spaces; “deals” appears for “trades” in §IV.H; “varying verification rate” in the conclusion is unclear.

Circularity Check

2 steps flagged

Suspicious rates and fraud-type breakdowns are definitional counts of the paper's own distance+heuristic predicates, not independent detections.

specific steps
  1. self definitional [Abstract; Algorithm 1 Anomaly Detection block (lines 21-27); §IV.A]
    "if d_i > t_d ∧ (|ΔP%|>10 ∨ S_V > t_V ∨ S_T > t_T) then Suspicious←True ... The method highlights 2.02% of trades as suspicious ... Despite the lack of ground truth, the model's performance is confirmed by a Silhouette Score of 0.561."

    Suspicious is defined exactly as the conjunction of (Euclidean distance to K-Means++ centroid > data-driven 95th-percentile threshold) and (at least one fixed behavioral rule). The 2.02% figure is therefore the measure of that predicate on the dataset; it cannot be an independent detection rate or performance confirmation. Silhouette only scores cluster geometry of the input features and does not validate the fraud labels.

  2. self definitional [Table III; §IV.D–E; Algorithm 1 Risk & Fraud Labeling; Abstract]
    "TABLE III: Heuristic Rules ... Spoofing <5 >95th pct >95th pct – Same day ... The method highlights 2.02% of trades as suspicious where 51.10% clearly indicate spoofing, 0.10% indicate pump and dump, 0.55% indicate insider trading, 1.43% indicate a fake breakout, and 46.83% are unclassified."

    Fraud-type labels are assigned by re-applying a second table of the same style of price/volume/lookahead heuristics to the already-flagged set. The percentages (51.10% spoofing etc.) are therefore pure counts of how many flagged points match each rule definition; they are not an independent categorization or discovery of market-manipulation prevalence.

full rationale

The paper's central quantitative claims (2.02% suspicious trades; 51.10% spoofing etc.) are obtained simply by counting how many of the ~1M points satisfy the predicates the authors themselves wrote into Algorithm 1 and Table III. There is no external ground-truth, known-event check, synthetic injection, or pure-rule ablation that would make the percentages an independent measurement. Silhouette (0.561) only quantifies separation of the engineered feature clusters and does not speak to whether any flag is actual manipulation. The derivation chain therefore reduces, by construction, to the chosen 95th-percentile cut-offs and the fixed market-practice rules; the reported figures are restatements of those rules applied to the data rather than discoveries or predictions. This is partial circularity of the self-definitional kind; the pipeline itself is a legitimate unsupervised detector, but the evaluation claims treat its definitional outputs as confirmatory evidence. No self-citation load-bearing chain or uniqueness import is present.

Axiom & Free-Parameter Ledger

8 free parameters · 5 axioms · 1 invented entities

The central claim rests almost entirely on hand-chosen clustering and market-practice thresholds plus the domain assumption that those thresholds equal real manipulation. No free parameters are fit to labeled fraud outcomes (there are none); instead many cutoffs are chosen by convention (95th percentiles, 10% price moves, k=5, 30-day windows, 60/40 score weights). The SMMD pipeline is a named packaging of standard steps rather than a new physical or mathematical entity with external evidence.

free parameters (8)
  • k (number of K-Means clusters) = 5
    Set to 5 via elbow method on inertia; directly controls cluster centers and thus distance-based outlier flags.
  • distance quantile threshold t_d = 0.95 quantile
    95th percentile of Euclidean distances to assigned centroids; defines structural outliers.
  • volume/trade spike thresholds t_V, t_T = 0.95 quantile
    95th percentiles of SV and ST; required for the behavioral arm of the suspiciousness rule.
  • price-change threshold Δp_thr = 10%
    Fixed |ΔP%| > 10 used both for anomaly filtering and for several fraud-type rules.
  • rolling window length = 30 days
    30-day rolling means/std for volume, trades, turnover, and volatility features.
  • suspicion score weight α = 0.6 / 0.4 (stated in conclusion)
    Frequency vs severity mix in Eq. 2; conclusion states a 60/40 heuristic that may need retuning.
  • risk-category percentile cuts = 90, 75, 50, 25
    Critical/High/Medium/Low/Minimal bins at 90/75/50/25 percentiles of suspicion score (Table II).
  • lookahead horizon for pump/fake-breakout/rug-pull = 5 days
    5-day future window used to label reversals in Table III and Algorithm 1.
axioms (5)
  • domain assumption Points far from K-Means centroids in standardized feature space are structural anomalies worth treating as candidate fraud.
    Invoked in Algorithm 1 anomaly block and §IV.A; standard anomaly-detection assumption, not proved for DSE manipulation.
  • domain assumption Market-practice heuristic rules (Table III) correctly map observable price/volume patterns onto named manipulation types (spoofing, pump-and-dump, insider trading, fake breakout, rug pull).
    Load-bearing for the fraud-type percentages in the abstract and §IV.D–E; rules are asserted from practice citations, not validated on labeled cases.
  • ad hoc to paper Silhouette score is an adequate performance confirmation for a fraud-detection system when ground truth is absent.
    Stated in Abstract and §IV.G; silhouette measures cluster geometry, not precision/recall of manipulation labels.
  • domain assumption Standard Euclidean K-Means++ on z-scored engineered features is an appropriate model of ‘regular’ vs ‘suspicious’ trading structure.
    §III.D–E; alternative density-based methods are mentioned only via citation, not ablated on this data.
  • domain assumption Dropping incomplete rows and replacing infinities yields a representative 1,019,783-row sample of DSE trading behavior.
    §III.A; no analysis of selection bias from dropped rows or survivorship.
invented entities (1)
  • SMMD (Stock Market Manipulation Detection) framework no independent evidence
    purpose: Name the end-to-end unsupervised pipeline (features + K-Means++ + hybrid filter + heuristic fraud labels + risk scores).
    Packaging of known clustering and rule components; no independent external evidence beyond this paper’s own run.

reviewed 2026-07-11 · how reviews work

0 comments
Cite this review

Pith. "Pith review of A Clustering-Based Framework for Identifying Suspicious Trading Patterns in Capital Market." pith.science (2026). https://pith.science/paper/EVDLNAIH

@misc{pith2026260704184,
  author       = {Pith},
  title        = {Pith review of: A Clustering-Based Framework for Identifying Suspicious Trading Patterns in Capital Market},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EVDLNAIH}},
  note         = {Machine review of arXiv:2607.04184}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Market manipulation is the dubious practice of manipulating stock prices in order to make a quick profit, which truly degrades confidence on trading platforms. We implemented an unsupervised fraud-detection toolkit that begins with K-Means++ clustering to address this issue. A dataset of roughly one million financial transactions from 2012 to 2024 is used. In order to identify fraudulent trades and categorize them using market practice heuristic thresholds, the study suggests a clustering-based pipeline. The method highlights 2.02% of trades as suspicious where 51.10% clearly indicate spoofing, 0.10% indicate pump and dump, 0.55% indicate insider trading, 1.43% indicate a fake breakout, and 46.83% are unclassified. Despite the lack of ground truth, the model's performance is confirmed by a Silhouette Score of 0.561.

Figures

Figures reproduced from arXiv: 2607.04184 by Asif Zaman, Iftekharul Mobin, Romona Magdalene Sarkar, Sabiha Khair Ohi.

Figure 1
Figure 1. Figure 1: Market manipulation detection workflow anomaly filtering system that combines percentile￾based behavioral criteria with distance-based outlier detection. II. LITERATURE REVIEW Traditional supervised learning models have been widely used for fraud detection, but their effectiveness is confined by the need for enormous volumes of labeled data and inability to identify key fraud patterns. Currently, more rese… view at source ↗
Figure 2
Figure 2. Figure 2: Elbow Method for Optimal k F. Model Evaluation Silhouette Score: Silhouette analysis mainly checks how close each item in the cluster is to items in other clusters. The score value ranges from -1 to +1. A score value of +1 indicates that objects are well-clustered, whereas a score value of -1 shows that the objects are not correctly clustered [16]. For a single sample i : s(i) = b(i)−a(i) max{a(i),b(i)} (1… view at source ↗
Figure 3
Figure 3. Figure 3: DSE Stocks closing price with detected suspicious [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Fraud Pattern Breakdown of Top 20 Suspicious Stocks [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

20 extracted references · 2 linked inside Pith

  1. [1]

    N. T. Chowdhury, N. S. Mahdzan, and M. Rahman. Beyond intu- ition: The role of financial knowledge in navigating investments in emerging markets.International Journal of Economics and Financial Issues, 14(4):267–281, 2024

    2024

  2. [2]

    F. A. B. Habib. Evaluating the effectiveness of macroeconomic determinants on the performance of the dhaka stock exchange: A time series approach.Journal of Ekonomi, XX(XX):XX–XX, 2024

    2024

  3. [3]

    Jakaria, H

    M. Jakaria, H. Shikder, N. M. Himel, N. Siddiqui, and M. M. Rahman. Analyzing stock price volatility: A statistical study of three listed companies on the dhaka stock exchange.Asian Business Review, 2025

    2025

  4. [4]

    M. I. Wahab and F. Olakkodu. Insider trading: An overview in indian and global markets.Journal of Informatics Education and Research, 5(XX):XX–XX, 2025

    2025

  5. [5]

    S. K. Vangibhurathachhi. Machine learning for fraud detection in financial transactions.International Journal on Science and Technology, 16(XX):XX–XX, 2025

    2025

  6. [6]

    Huang, H

    Z. Huang, H. Zheng, C. Li, and C. Che. Application of machine learning-based k-means clustering for financial fraud detection. Academic Journal of Science and Technology, 10(1):33–39, 2024

    2024

  7. [7]

    V . N. Mandhala et al. Comparative analysis of clustering algorithms for financial fraud detection.International Journal of Safety & Security Engineering, 15(4), 2025

    2025

  8. [8]

    Y . Peng, X. Sun, Y . Duan, Z. Fang, and T. Tang. Unsupervised de- tection of fraudulent transactions in e-commerce using contrastive learning. arXiv preprint arXiv:2503.18841, 2024. Accessed: 2026-02-20

    Pith/arXiv arXiv 2024

  9. [9]

    W. Yang, R. Wang, and B. Wang. Detection of anomaly stock price based on time series deep learning models. InProc. 2020 Management Science Informatization and Economic Innovation Development Conf. (MSIEID), pages 110–114, December 2020

    2020

  10. [10]

    Y . Li, S. Fan, and H. Wang. Machine learning-based identification of anomalous trading behavior patterns among asia-pacific in- vestors in u.s. securities markets.Spectrum of Research, 5(1):XX– XX, 2025

    2025

  11. [11]

    S. M. Darwish, A. I. Salama, and A. A. Elzoghabi. Intelligent approach to detecting online fraudulent trading with solution for imbalanced data in fintech forensics.XX, 15(1):17983, 2025

    2025

  12. [12]

    Hasan, B

    R. Hasan, B. Biswas, M. Samiun, M. A. Saleh, M. Prabha, J. Akter, F. H. Joya, and M. Abdullah. Enhancing malware detection with feature selection and scaling techniques using machine learning models.Scientific Reports, 15(XX):XX–XX, 2025

    2025

  13. [13]

    Oyewole and G

    G. Oyewole and G. Thopil. Data clustering: Application and trends.Artificial Intelligence Review, 56:6439–6475, 2022

    2022

  14. [14]

    Aliguliyev and S

    R. Aliguliyev and S. F. Tahirzada. Performance comparison of k- means, parallel k-means and k-means++. Unpublished manuscript, 2023

    2023

  15. [15]

    R. H. Khan, D. F. Dofadar, and M. G. R. Alam. Explainable customer segmentation using k-means clustering. InProc. UEMCON 2021, pages 639–643, 2021

    2021

  16. [16]

    J. Meng, A. M. Abed, M. G. Elsehrawy, A. D. Al Agha, N. Ab- dullah, S. Elattar, M. Abbas, H. Al Garalleh, and H. Assilzadeh. Nano-integrating green and low-carbon concepts into ideological and political education in higher education institutions through k-means clustering.Heliyon, 10:e31244, 2024

    2024

  17. [17]

    Fabre and I

    T. Fabre and I. M. Toke. High-frequency market manipulation detection with a markov-modulated hawkes process. arXiv preprint arXiv:2502.04027, 2025. Accessed: 2026-02-20

    Pith/arXiv arXiv 2025

  18. [18]

    Rizvi, A

    B. Rizvi, A. Belatreche, A. Bouridane, and I. Watson. Detection of stock price manipulation using kernel-based principal component analysis and multivariate density estimation.IEEE Access, 8:135989–136003, 2020

    2020

  19. [19]

    Hambuckers, L

    J. Hambuckers, L. Sun, and L. Trapin. Measuring tail risk at high- frequency: An L1-regularized extreme value regression approach with unit-root predictors. arXiv preprint, 2023

    2023

  20. [20]

    Y . Li, N. Yao, Y . Huo, and W. Cai. Trust dynamics and bot-driven responses: An approach to rug pulls in solana meme coin markets. InProc. 17th ACM Web Science Conf. (WebSci), 2025. 6

    2025

This paper was first reviewed by grok-4.5 on July 11, 2026.