Where detection methodology meets live market data
Our team publishes research articles, documented case studies, and analytical notes
drawn from real market episodes — not constructed scenarios.
Every article in the Insights hub begins with observed data. Our team identifies an
episode — an anomalous volume cluster, an unusual spread widening, a cross-asset signal
that preceded a price event — and then traces it back through the detection framework to
document exactly what the order-flow evidence showed, when it appeared, and what the
statistical models produced. This is applied market microstructure research, presented
at a level of precision that a serious analyst can replicate and challenge.
Latest research articles and market analysis from our team
New case studies, methodology updates, and analytical notes are published regularly — each grounded in documented real-market episodes and reviewed by the full research team before release.
Case Study
Order flow imbalance ahead of ASX earnings announcements
Our team examined tick-level data from a sample of ASX-listed mid-cap equities in the sessions preceding quarterly earnings releases, identifying statistically significant order-flow imbalance patterns that preceded price adjustments in a majority of documented cases.
Methodology
How bid-ask spread dynamics signal informed accumulation
Spread widening patterns in the minutes before a significant price move carry a measurable statistical signature. This article traces the mechanism through three documented episodes and explains what the microstructure literature says about why it occurs.
Research Note
VPIN in low-liquidity markets: where the model degrades
The volume-synchronized probability of informed trading measure performs differently in thin markets than in high-liquidity venues. Our team documents the conditions under which the model's assumptions break down and what analysts should use instead.
Case Study
Cross-asset signals before a documented corporate event
In one of the more instructive episodes in our case study library, informed flow appeared in the options market several sessions before the corresponding equity venue showed any statistically anomalous activity — a multi-market synthesis problem worth examining carefully.
Analysis
Supervised classification accuracy across different market regimes
Back-testing classification models built on pre-2024 data against post-2024 market conditions reveals measurable performance degradation in two of the four model families tested. Our team documents the findings and discusses regime-aware recalibration approaches.
Most financial commentary is written backwards — from a known outcome, with an explanation constructed after the fact. Our team takes a different approach. Before any article is published, the research team identifies the episode, retrieves the original order-flow and quote data, runs the detection methodology against that data, and documents the results — including cases where the signal was ambiguous, late, or absent. That sequencing matters. It is the difference between a case study that illuminates how detection logic works under real conditions and one that simply confirms what the author already believed. We hold every article to the same standard: the methodology must be traceable, the data must be identified, and the limitations must be stated before the conclusions are drawn.
1
Every article begins with observed market data, not with a conclusion that the data is then selected to support. The episode comes first; the analytical interpretation follows from the evidence the order-flow record actually contains.
2
Where a detection model produces ambiguous output — where the statistical evidence is present but insufficient to draw a confident conclusion — our team documents that ambiguity explicitly rather than rounding toward the more compelling narrative.
3
Each published case study identifies the specific detection measures applied, the data source and time period examined, and the model outputs produced. The analytical steps are documented at a level of detail that allows a reader to reproduce the core findings independently.
4
Our team distinguishes between what the order-flow data shows and what it implies about participant intent. Statistical patterns are documented as patterns. Attribution to specific participant types is presented as inference, not fact, and is qualified accordingly.
5
When new research in market microstructure or applied machine learning affects the reliability of a methodology documented in a prior article, our team publishes a revision note and updates the relevant content. The publication record is treated as a living document, not a fixed archive.
6
The research team reviews all submissions from community members who contribute real-market observations. Observations that meet the team's data quality and methodological consistency standards are integrated into subsequent articles or framework revisions, with appropriate attribution.
7
No article published in the Insights hub constitutes financial advice, a market forecast, or a recommendation to act on any signal described therein. All content is retrospective, analytical, and presented for informational and research reference purposes only.
Thirty-eight basis points. That was the average lead in the price-forming signal we identified across the case studies that went into Module Seven. It sounds small. In a market where institutional participants are moving size, it represents a structurally significant interval — enough time for the statistical footprint to accumulate in the order book before the price begins to adjust. The point is not that this number is actionable in any simple sense. The point is that it is measurable. And once something is measurable, it can be studied, modeled, and understood. That is the entire premise of the detection framework we have built. Markets are not perfectly legible — but they are not opaque either. The data is there. The question is whether the analyst has the methodology to read it honestly, including the discipline to report when the signal is absent, degraded, or insufficient to support a conclusion. That discipline is what separates rigorous detection work from pattern-matching dressed up in quantitative language. Our team has built this entire research program around that distinction, and it is the standard we apply to every article published in this hub.
Dr. Marcus Ellery, Lead Research Analyst
Head of Market Microstructure Research, Vaslemrotuem — Researcher in informed
trading detection methodology and order-flow statistical modeling
The spread-widening signal is one of the most consistently documented patterns in the informed trading literature — yet most retail analysts have never been shown how to measure it from standard quote data. The three case studies in that article were chosen specifically because they represent different liquidity regimes. The mechanism is the same in all three; the magnitude and timing vary in instructive ways. Reading those variations is where the analytical skill actually develops.Dr. Marcus Ellery Lead Research Analyst, Head of Microstructure Research
The VPIN degradation article came directly from a community member observation.
An analyst in the practitioner track noticed that the model was producing
systematically elevated readings in a thinly traded instrument and submitted the
data. Our team replicated the finding, identified the structural reason — trade
clustering violating the model's arrival-rate assumptions — and built the
article around the documented episode. That feedback loop is what keeps the
framework grounded in how markets actually behave.
Amelia Tan
Head of Community Research Insights, Vaslemrotuem
The supervised classification back-testing article was the most difficult to
write because the results were unflattering to two of the four model families we
tested. There was an internal discussion about whether to narrow the scope to
the models that performed well. The team decided unanimously to publish the full
picture. If we document only the cases where detection works cleanly, we produce
a misleading account of what the methodology can and cannot do.
Priya Nair
Senior Quantitative Analyst, Machine Learning Applications
What distinguishes the cross-asset case study from most informed trading analyses is the multi-session lead time. The options market signal preceded the equity venue signal by three full sessions. That is not a microstructure anomaly — it is a structural feature of how informed participants allocate across markets when they are managing position size carefully. Documenting the detection logic across both venues in sequence is the only way to understand the episode accurately.James Whitfield Director of Research Documentation, Vaslemrotuem
Landmark events
Milestones that shaped informed trading detection research
1985
PIN Model Theoretical Foundations
Glosten and Milgrom publish the sequential trade model that establishes the theoretical relationship between informed order flow, bid-ask spread dynamics, and market maker pricing behavior. This paper becomes the conceptual foundation on which the Probability of Informed Trading measure is later constructed, and it remains the primary reference for the mechanical explanation of why informed trading leaves a measurable statistical footprint in public market data. Our team draws on this framework in Modules One and Three of the detection curriculum.
1996
Easley, Kiefer, O'Hara PIN Estimation
Easley, Kiefer, O'Hara, and Paperman publish the structural model that operationalizes the Probability of Informed Trading measure from observable trade and quote data. The model's maximum-likelihood estimation procedure allows researchers and practitioners to calculate an informed trading probability from standard market records without access to order attribution data. This paper represents the transition from theoretical microstructure to applied detection methodology and anchors the statistical content of Module Three in the Vaslemrotuem framework.
2008
Regulation NMS and Electronic Market Fragmentation
The full implementation of Regulation NMS in United States equity markets accelerates the fragmentation of trading across multiple venues and the rise of high-frequency market making. This structural shift creates significant challenges for traditional PIN estimation, which assumes a single-venue trade arrival process. The resulting need to adapt detection methodology to fragmented, high-frequency environments motivates the development of volume-synchronized approaches to informed trading probability and directly shapes the content of Modules Four and Seven of the Vaslemrotuem detection framework.
2012
VPIN and Flash Crash Analysis Published
Easley, López de Prado, and O'Hara publish the volume-synchronized probability of informed trading measure and apply it retrospectively to the 2010 Flash Crash episode, demonstrating that the measure showed elevated informed trading probability in the period preceding the event. The paper generates significant practitioner and regulatory interest in order-flow toxicity as a real-time market stress indicator. It also spurs a productive academic debate about the measure's assumptions and limitations — a debate our team documents in detail in Module Four, presenting both the original findings and the peer responses that followed.
Weekly Insights
Informed trading signals and analytical notes delivered every Monday morning to
your inbox.
One documented signal observation per week
Methodology notes and framework updates
Priority access to new case study publications