Clustering and Similarity Learning in Financial Markets: A Tutorial for the Practitioners
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Clustering and similarity learning are increasingly indispensable for structuring heterogeneous financial data and supporting real-world decision-making. Traditional heuristics such as industry codes, static style boxes, or return correlations offer only coarse and rigid notions of peer groups. Recent advances in metric learning, graph methods, and large language models now make it possible to build adaptive neighborhoods of securities, funds, companies, and investors that align more closely with actual risk, liquidity, and thematic exposures. This tutorial synthesizes these methodological developments and demonstrates their use across major asset classes. Case studies show how supervised proximities improve bond substitution, how fund similarity systems reconcile category reproducibility with outlier detection, how multimodal pipelines refine company comparables for valuation and strategy, and how investor clustering enhances personalization and “know your client” (KYC) analytics. We emphasize modeling choices that make clustering and similarity auditable and robust under regime shifts. We also outline their evaluation protocols such as neighborhood stability, substitution fidelity, and segment utility, and so on, which align with investment, compliance, and fiduciary objectives. Overall, the central message for practitioners is pragmatic: Similarity systems have moved beyond experimental prototypes and now stand as deployable techniques within real investment workflows.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.030 | 0.010 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it