All content is for informational purposes only and does not constitute financial advice.
Detection Methods
Six analytical methods. One coherent framework for reading informed order flow.
Our team built each detection method as a standalone analytical discipline and then documented the precise points at which they connect — so that the analyst who masters all six can approach any market episode with a complete, layered evidence base rather than a single signal in isolation.
View curriculum
Single signals are unreliable in noisy market conditions
A volume spike in isolation could reflect institutional accumulation, a mechanical index rebalance, or a liquidity-driven trade with no informational content whatsoever. Without a second or third detection lens applied to the same episode — spread behavior, order imbalance, cross-asset positioning — the analyst has no principled basis for distinguishing between these explanations. Our team designed the six-method framework specifically to address this ambiguity by requiring convergent evidence across multiple independent measures before drawing a detection conclusion.
Convergent signals across methods reduce false-positive interpretation rates
When the Probability of Informed Trading measure is elevated at the same session in which order-flow imbalance accumulates in one direction and the bid-ask spread widens beyond its historical baseline, the convergence of three independent measures is statistically more informative than any one of them alone. The framework trains the analyst to look for that convergence deliberately, and to treat divergence — where one measure fires and the others do not — as a signal of ambiguity rather than evidence of informed activity.
Machine learning methods extend detection into high-frequency data regimes
Traditional structural models like PIN were developed before electronic markets generated tick-level data at millisecond granularity. Supervised classification and anomaly detection techniques allow the analyst to apply detection logic to that higher-resolution data — identifying directional trade clustering, sequence anomalies, and volume deviations that structural models were not designed to capture. Our team documents both the strengths and the documented failure modes of each machine learning approach used in the framework, so the analyst understands precisely where each method is reliable and where it is not.
Anomaly detection identifies deviations from the statistically defined uninformed baseline
Rather than attempting to identify informed participants directly — an attribution problem that public data cannot resolve — the anomaly detection method identifies when observable market behavior deviates from the statistical distribution associated with uninformed trading. Episodes where volume, spread, and order imbalance simultaneously deviate from their historical baselines are documented as anomalous and subjected to the full multi-method framework before any interpretive conclusion is drawn.
Regime-aware calibration preserves detection reliability across shifting market structures
A detection screen calibrated on pre-volatility-regime data will produce systematically miscalibrated outputs when applied to a different market environment. Our team's calibration methodology — documented in Module Eight — addresses this directly by introducing structural break testing, rolling-window recalibration, and explicit regime-identification procedures. The analyst who understands calibration can maintain the reliability of their detection work across different market conditions rather than unknowingly applying a stale model to a changed environment.
Cross-market synthesis identifies informed flow that migrates across asset classes
Sophisticated informed participants frequently distribute their positioning across multiple instruments or venues to manage market impact — entering an options position before the corresponding equity shows any anomalous activity, or accumulating across fragmented venues to avoid triggering single-venue detection thresholds. The multi-market synthesis method trains the analyst to read order-flow evidence across asset classes simultaneously, using documented real-market episodes to illustrate how cross-asset signals develop and how they differ from coincidental correlation.
Community-submitted observations continuously refine method documentation accuracy
Each of the six methods in the framework has been refined by real-market observations submitted through the community feedback loop. When an analyst in the practitioner track identifies an episode where a documented method produces anomalous output, that observation is routed to the research team for review. Observations that meet the team's data quality standards are integrated into method revision notes and, where warranted, into structural updates to the framework documentation.
Every method is documented with stated assumptions and explicit limitations
Our team holds that the most important content in any method documentation is not the worked example where the detection logic succeeds — it is the explicit description of the conditions under which the method tends to fail, produce ambiguous output, or require supplementary evidence before a conclusion can be supported. Every method module includes a dedicated limitations section written with the same analytical precision as the core methodology documentation.
Method Modules
Our team has documented six distinct detection methods — each grounded in market microstructure theory, tested against historical data, and presented with explicit performance boundaries so the analyst knows exactly when to trust the signal and when to seek corroborating evidence from a second method.
Signal Tracker
(FLOW)
Live
Elevated imbalance
Increasing signal
Recent signals
Options Flow Analysis and Interpretation
Unusual options positioning — measured through open interest concentration, put-call ratio deviations, and volume relative to historical baseline — can indicate directional conviction by informed participants before an equity price event. Our team documents how to identify statistically anomalous options flow, how to distinguish genuine directional positioning from hedging activity and mechanical roll transactions, and how to integrate options signals into the broader multi-method detection framework without over-interpreting isolated observations.
View curriculumDark Pool and Block Trade Signal Reading
When informed participants manage position size carefully, they frequently route orders through dark pools or negotiate block transactions away from the lit order book to minimize market impact. These transactions leave indirect traces — post-trade price behavior, shifts in the lit market's order-flow balance, and changes in the bid-ask spread following the transaction. Our team documents how to identify and interpret these secondary signals, and how to distinguish informed dark-pool activity from routine institutional liquidity management that carries no informational content.
Explore modulesUnusual Volume Detection and Baseline Modeling
Identifying genuinely anomalous volume requires a statistically defined baseline against which deviations can be measured. Our team documents the construction of rolling-window volume baselines, the choice of distributional assumptions appropriate to different market liquidity tiers, and the threshold methodology for flagging observations as anomalous rather than simply elevated. The module also addresses the most common false-positive sources — earnings calendars, index rebalancing dates, and expiry-related flow — and how to control for them before applying the detection framework.
Learn the methodInstitutional Accumulation Pattern Recognition
Informed institutional participants accumulating a position over multiple sessions leave a distinctive pattern in the order-flow record: directional imbalance that persists across sessions, volume that clusters at specific price levels, and spread behavior that reflects market-maker adjustment to perceived informational risk. Our team documents how to identify this multi-session accumulation signature, how to distinguish it from price-momentum-driven retail flow, and how the Probability of Informed Trading measure and order-flow imbalance metrics interact when accumulation is occurring across an extended window.
Start learning
Detection methodology versus conventional market reading
The difference between reading market data through a detection framework and reading it without one is a difference of structure, not effort. Our team mapped the specific points at which the gap becomes analytically consequential.
Vaslemrotuem
Conventional Approach
Initial signal identification
Session open
Vaslemrotuem
Order-flow imbalance and spread behavior are measured against statistically defined baselines from the session's first tradeable interval, giving the analyst a structured basis for flagging anomalous activity as soon as the data begins to accumulate.
Structured
•
Early session
Прогресс
85%
Conventional Approach
Volume and price movement are observed visually against recent history, with no statistically defined baseline and no formal threshold for distinguishing anomalous from routine activity in the data.
Unstructured
•
Mid-session
Прогресс
30%
Multi-method signal convergence check
During session
Vaslemrotuem
PIN measure, order imbalance, and spread dynamics are assessed simultaneously across the same time window. Convergent deviation across two or more independent measures increases the analytical confidence that the episode warrants further scrutiny.
Systematic
•
Same session
Прогресс
90%
Conventional Approach
A single metric — typically price or volume — drives the interpretation. Secondary signals are noted informally if they are noticed at all, with no formal methodology for weighting or combining evidence from different data dimensions.
Ad hoc
•
Post-session
Прогресс
25%
Regime calibration and false-positive control
Ongoing
Vaslemrotuem
Structural break testing and rolling-window recalibration are applied periodically to ensure that detection baselines remain appropriate to the current market regime. Model performance is reviewed against recent data and adjusted when documented degradation is observed.
Maintained
•
Ongoing
Прогресс
88%
Conventional Approach
Baselines and thresholds established from prior market conditions are applied indefinitely without a formal recalibration process. Signal degradation in changed market regimes is typically attributed to market noise rather than to model staleness.
Static
•
Indefinite
Прогресс
18%
Vaslemrotuem
88%
Conventional Approach
24%
A structured multi-method framework produces fewer false positives, more consistent signal identification, and a principled basis for distinguishing genuine detection from coincidental pattern observation.
Six mistakes analysts make when reading market signals — and how the detection framework addresses each one
Our team identified these errors repeatedly in community feedback and real-market case study reviews. Each one has a documented structural remedy within the framework.
Methodology
Treating volume spikes as self-evidently informative without a baseline comparison
An elevated volume observation only carries analytical meaning when measured against a statistically defined baseline appropriate to the instrument's liquidity tier, the time of session, and the calendar context. Without that baseline, the analyst cannot distinguish a genuine anomaly from a routine day where participation is simply above the recent rolling average. The framework's unusual volume detection method documents baseline construction as its foundational step — before any threshold or flag is applied to the data.
Treating volume spikes as self-evidently informative without a baseline comparison
An elevated volume observation only carries analytical meaning when measured against a statistically defined baseline appropriate to the instrument's liquidity tier, the time of session, and the calendar context. Without that baseline, the analyst cannot distinguish a genuine anomaly from a routine day where participation is simply above the recent rolling average. The framework's unusual volume detection method documents baseline construction as its foundational step — before any threshold or flag is applied to the data.
Methodology
Signal Quality
Interpreting a single signal as sufficient evidence of informed activity
One elevated measure — whether a volume deviation, a spread widening, or a period of directional order imbalance — does not constitute reliable detection evidence on its own. Each of these observations has multiple plausible non-informational explanations. The multi-method framework exists precisely to address this: the analyst is trained to seek convergent evidence across at least two independent detection measures before drawing any interpretive conclusion about the nature of the observed activity.
Interpreting a single signal as sufficient evidence of informed activity
One elevated measure — whether a volume deviation, a spread widening, or a period of directional order imbalance — does not constitute reliable detection evidence on its own. Each of these observations has multiple plausible non-informational explanations. The multi-method framework exists precisely to address this: the analyst is trained to seek convergent evidence across at least two independent detection measures before drawing any interpretive conclusion about the nature of the observed activity.
Signal Quality
Calibration
Applying a detection model calibrated on historical data to a structurally different market regime without recalibrating it
Supervised classification models and statistical baselines built from data in one market regime — defined by a particular volatility level, liquidity structure, or trading infrastructure — will produce miscalibrated outputs when applied to a regime with different characteristics. The framework's Module Eight documents how to identify regime shifts using structural break tests and how to recalibrate detection screens systematically when a shift is confirmed, rather than continuing to apply a stale model and misattributing its degraded output to genuine market signals.
Applying a detection model calibrated on historical data to a structurally different market regime without recalibrating it
Supervised classification models and statistical baselines built from data in one market regime — defined by a particular volatility level, liquidity structure, or trading infrastructure — will produce miscalibrated outputs when applied to a regime with different characteristics. The framework's Module Eight documents how to identify regime shifts using structural break tests and how to recalibrate detection screens systematically when a shift is confirmed, rather than continuing to apply a stale model and misattributing its degraded output to genuine market signals.
Calibration
Interpretation
Confusing the statistical signature with an attribution to specific participant intent
Detection methodology identifies statistical deviations from the uninformed baseline. It does not identify who is trading or why. Analysts who conflate the statistical observation with a specific attribution — concluding that a particular entity is accumulating ahead of a known event — are making an inferential leap that the data does not support. The framework trains analysts to state detection conclusions as statistical observations and to treat any attribution to participant intent as a separate, explicitly qualified inference.
Confusing the statistical signature with an attribution to specific participant intent
Detection methodology identifies statistical deviations from the uninformed baseline. It does not identify who is trading or why. Analysts who conflate the statistical observation with a specific attribution — concluding that a particular entity is accumulating ahead of a known event — are making an inferential leap that the data does not support. The framework trains analysts to state detection conclusions as statistical observations and to treat any attribution to participant intent as a separate, explicitly qualified inference.
Interpretation
False Positives
Overlooking expiry-related and index rebalancing flows as false-positive sources in volume detection
Options expiry dates, index reconstitution events, and end-of-quarter rebalancing windows all generate volume and order-flow patterns that superficially resemble informed accumulation signals under naive detection models. The framework's unusual volume detection method includes a calendar-control procedure specifically designed to flag these periods before the detection model is applied, ensuring that mechanical institutional flows are not misclassified as informational order flow by an analyst who has not accounted for the structural calendar effect.
Overlooking expiry-related and index rebalancing flows as false-positive sources in volume detection
Options expiry dates, index reconstitution events, and end-of-quarter rebalancing windows all generate volume and order-flow patterns that superficially resemble informed accumulation signals under naive detection models. The framework's unusual volume detection method includes a calendar-control procedure specifically designed to flag these periods before the detection model is applied, ensuring that mechanical institutional flows are not misclassified as informational order flow by an analyst who has not accounted for the structural calendar effect.
False Positives
Dark Pools
Reading dark pool and block trade observations without adjusting for post-trade price impact timing
Block transactions and dark pool executions do not produce the same immediate price impact as equivalent lit-market orders. The informational content of a block trade, if any exists, typically surfaces in the lit market's order flow in the sessions following the transaction rather than immediately. Analysts who look for immediate post-trade price movement as the confirmation criterion for an informed block trade will systematically miss the signal. The framework documents the documented lead-time structure between dark pool activity and subsequent lit-market order-flow behavior across several real-market case study episodes.
Reading dark pool and block trade observations without adjusting for post-trade price impact timing
Block transactions and dark pool executions do not produce the same immediate price impact as equivalent lit-market orders. The informational content of a block trade, if any exists, typically surfaces in the lit market's order flow in the sessions following the transaction rather than immediately. Analysts who look for immediate post-trade price movement as the confirmation criterion for an informed block trade will systematically miss the signal. The framework documents the documented lead-time structure between dark pool activity and subsequent lit-market order-flow behavior across several real-market case study episodes.
Dark Pools
Methods FAQ
Is informed trading detection legal to practice?
Legality
Detecting informed trading activity through statistical analysis of publicly observable market data — order flow records, quote data, volume figures, and trade timestamps — is entirely legal in Australia and in all major jurisdictions where market data is publicly distributed. The framework documented on this website does not involve accessing private information, non-public corporate data, or any communication that would constitute insider trading under the Corporations Act 2001 (Cth) or equivalent legislation. Our team's detection methodology works exclusively with data that is available to any market participant through standard commercial data feeds. The discipline is practised by academic researchers, regulatory bodies, and institutional risk teams internationally. Users should satisfy themselves that their own data access and analytical practices comply with applicable laws and their data licensing agreements.
Start reading order flow the way our team documents it across all six detection methods
The six detection methods are integrated into a single eight-module sequence that takes the analyst from market microstructure fundamentals through to advanced multi-market signal synthesis — each module anchored to a documented real-market case study.
- Progress through all six methods at your own analytical pace
- Every concept traced to a real documented market episode
- Community feedback loop refines the methodology continuously
All content is for informational purposes only and does not constitute financial advice.
Research backed