Reading the
Tape

Exaggerated Claims, Speculative Promises

The retail trading content space is dominated by claims that would not survive the most basic tests of academic peer review, regulatory scrutiny, or forensic discipline. Unconditional performance promises — language guaranteeing that a subscriber will never miss a bull run or will always make money — appear regularly in marketing copy from products that, on the evidence of their own websites, are not registered investment advisers and have no documented mechanism by which such guarantees could be produced.¹ Recommendation systems publish lists of tickers without methodology disclosure. Performance numbers are reported without compliance with the SEC Marketing Rule's requirement of net-of-fees, balanced presentation, or clear distinction between hypothetical and actual results.² Track records are cherry-picked. Failures are not published. Audit logs do not exist.

The pattern is not new. The behavioral finance literature has documented for decades how retail traders are systematically targeted by content that exploits well-characterized cognitive biases — overconfidence, recency, loss aversion, and the disposition effect — without offering any analytical apparatus that would help the reader resist them.³ The result is an industry in which subscribers churn from product to product, persistently disappointed, persistently susceptible to the next promise of guaranteed returns.

This paper argues that an alternative is possible, and demonstrates one. The alternative is the application of forensic science methodology — the same evidentiary discipline used in laboratory criminalistics, court-admissible expert testimony, and peer-reviewed academic research — to the analysis of public market data. The argument of this paper is procedural, not promotional. Forensic methodology produces fewer claims than promotional content. The claims it does produce are documented, timestamped, and reconstructable. They are bounded by the same regulatory frame that governs publishers under §202(a)(11)(D) of the Investment Advisers Act and the disclosure requirements of CFTC Rule 4.41. They are, in the language of the source disciplines, falsifiable.

The remainder of this paper presents the methodology in operation. A single ticker — DGXX — is reconstructed from the Forensics.us audit log entry of 2026-05-07, with the seven signal layers that the methodology evaluates documented in turn, each with its academic or regulatory pedigree, and the trade outcome reported as it was recorded. The reader is invited to verify any specific claim against the published audit record.

Forensic Science as a Methodological Tradition

Forensic science is, in its mature form, a discipline organized around four commitments. The first is the principle of contact and exchange, formulated by Edmond Locard in the 1920s, which holds that no actor can engage with a scene without leaving and taking residue.⁴ The second is chain of custody — the documented, unbroken record of how each piece of evidence was collected, handled, and analyzed, without which findings are inadmissible.⁵ The third is the Daubert standard, a 1993 ruling by the United States Supreme Court that requires expert scientific testimony to rest on testable hypotheses, peer-reviewed methodology, known error rates, and general acceptance within the relevant scientific community.⁶ The fourth is the requirement of independent corroboration — a single piece of evidence, however suggestive, is not the basis for a finding; multiple independent lines of evidence must converge before a conclusion is recorded.⁷

These four commitments are not unique to criminal forensics. They are the operational standards of any analytical practice that claims to produce findings reliable enough to support consequential action. They are met, in different vocabularies, by laboratory medicine, civil engineering, intelligence analysis, and academic research. They are not met, in any meaningful sense, by the dominant content patterns of the retail trading industry.

The translation of these commitments to financial markets is not a metaphor. Markets, like crime scenes, are domains in which actors leave residue. Institutional traders cannot accumulate or distribute large positions without moving the public tape in characterizable ways. Insiders cannot transact in their own securities without filing Form 4 disclosures with the Securities and Exchange Commission within two business days.⁸ Pharmaceutical companies cannot receive Complete Response Letters from the Food and Drug Administration without those letters becoming part of the public record. Each of these residues is observable. Each is timestamped. Each can be subjected to evidentiary discipline.

Forensics.us was founded to apply this discipline systematically to the public residue of market activity. The methodology that follows is the operational form of that commitment.

From Crime Scene to Order Book

The translation from forensic science to market analysis is more direct than it first appears. The criminalist confronting a scene asks four questions: What happened here? Who left this residue? Is the evidence consistent with multiple lines of inquiry? Would another examiner, given the same evidence, reach the same finding? These four questions are the operational core of any disciplined analytical practice.

Applied to a candidate ticker, the questions become: What pattern is the public tape recording? Is the pattern consistent with institutional activity, retail noise, or active manipulation? Do independent evidentiary domains corroborate one another? Would the methodology, applied by a different operator at a different time, surface the same candidate from the same data?

The methodology described in the remainder of this paper is the documented procedure by which Forensics.us answers these questions. It evaluates seven independent signal layers — each with its own academic or regulatory pedigree, each producing a discrete finding, and each subject to the convergence requirement that prohibits any single layer from generating an actionable verdict on its own. The seven layers are not equally weighted, and not all of them fire on every candidate. The methodology requires corroboration among those that do fire, and explicit disclosure of those that did not. This procedural structure is what distinguishes a forensic finding from a forecast.

DGXX, 2026-05-07

At 08:15 ET on 2026-05-07, the Forensics.us pre-market intelligence run executed against the firm's pre-market candidate set. The run is one of five scheduled SMILE pipeline executions that span the trading day. The 08:15 ET run targets pre-market activity captured in the preceding hours through public market data and the firm's internal candidate-sourcing pipeline. The candidate set on this morning numbered forty-eight tickers. The methodology applied a series of independent filters and produced ten finalists. One of those finalists was DGXX.

DGXX entered the pre-market session showing a +22.02% gain on volume of 29.5 million shares, against a typical pre-market baseline volume an order of magnitude lower. The pre-market activity was sufficient to generate a FireFly confidence reading of 74%. From the FireFly stage, DGXX advanced through the methodology's six subsequent evaluation layers.

At 08:15:47 ET — forty-seven seconds after the run began — the system produced its NORTHSTAR card for DGXX. The verdict was LONG, confidence 81, signal strength HIGH, NORTHSTAR score 3. The signal price was recorded as $6.24. The T1 target was $6.51, the T2 target was $6.68, and the methodological stop was placed at $6.15. The card was persisted to the audit log within the same second. The trade was now part of the public record of the methodology's findings.

The remainder of this paper documents how the system arrived at that finding, layer by layer.

FireFly — Volume + Momentum

What it looks for. FireFly evaluates the relationship between current pre-market price action and the ticker's recent volume baseline. The layer does not generate verdicts. It produces a candidate set — the population of tickers for which subsequent layers will conduct fuller analysis. A FireFly confidence score is the quantified output of this stage.

Pedigree. The academic foundation for momentum-with-volume scanning traces to Jegadeesh and Titman's 1993 demonstration of price-momentum predictability in cross-sectional equity returns,⁹ later extended by Lee and Swaminathan to incorporate volume as a conditioning variable that distinguishes durable momentum from short-lived noise.¹⁰ The empirical finding is robust across decades, markets, and asset classes: tickers exhibiting elevated volume during directional price moves persist in those moves at rates above chance.

What it saw on DGXX. The FireFly scan recorded DGXX at +22.02% pre-market on volume of 29.5 million shares. The volume reading was sufficient to place DGXX in the top decile of the candidate set. The FireFly confidence value of 74% indicated the layer's strength signal but did not, on its own, establish a verdict.

What would have suppressed it. A pre-market move of similar magnitude on materially lower volume would have failed FireFly's volume threshold and the ticker would not have advanced to subsequent layers. A pre-market move at the open of a session classified by the regime gate as RED would have been similarly de-prioritized.

HBuy — Contrarian Accumulation

What it looks for. HBuy is a proprietary contrarian-accumulation detector that integrates four sub-components: trend direction, drawdown depth from recent highs, channel position within the recent trading range, and volume multiplier against the ticker's own historical baseline. The layer scores tickers exhibiting the geometric pattern of an established uptrend that has experienced a meaningful pullback into a buyable position with elevated participation.

Pedigree. The methodology draws on the mean-reversion-within-trend literature, including Lo and MacKinlay's foundational work on serial correlation in equity returns¹¹ and the broader academic treatment of accumulation patterns under volume confirmation as predictors of continuation moves.¹² The combination of pullback geometry with volume corroboration reflects the well-documented institutional preference for accumulating during declines rather than at extended levels — a preference visible in the public tape as a recognizable signature.

What it saw on DGXX. The HBuy layer recorded the following components for DGXX: trend = UP, drawdown = 21.5% from recent highs, channel position = 75% (upper quadrant of the recent range), volume multiplier = 3.5× baseline. The composite score was 79 on a 0–100 scale. The HBuy boolean signal flag did not fire — that flag requires a higher composite threshold. The score-without-signal pattern is itself informative: HBuy contributed evidentiary weight to the convergence calculation but did not, by itself, produce a verdict. This is the methodology's discipline operating as designed.

What would have suppressed it. A trend reading of DOWN would have nullified the contrarian-accumulation thesis entirely. A volume multiplier below baseline would have indicated absence of the institutional participation that the layer is designed to detect. A channel position above 90% would have placed the ticker in the chase zone and the score would have been correspondingly degraded.

NS-LF — Liquidity Forensics (VPIN)

What it looks for. The Liquidity Forensics layer monitors order-flow toxicity through Volume-synchronized Probability of Informed Trading — a microstructural metric designed to detect periods in which the public tape reflects predominantly informed (asymmetrically positioned) trading rather than random retail flow. High VPIN readings flag conditions in which a directional move is more likely to be driven by adverse-selection forces against retail participants and less likely to be a sustainable institutional accumulation.

Pedigree. VPIN was developed by Easley, López de Prado, and O'Hara, and published in The Review of Financial Studies in 2012.¹³ The metric was subsequently validated in regulatory analysis of the May 2010 Flash Crash and adopted in academic and institutional liquidity research.¹⁴ The methodology applied at Forensics.us uses a 16-snapshot exponentially weighted moving window, a refinement to the original specification that improves responsiveness to intraday regime shifts while preserving the underlying statistical foundation.

What it saw on DGXX. The Liquidity Forensics layer evaluated DGXX during the candidate qualification stage and produced no KILL signal. By contrast, in the same run, two other candidates (ERNA at VPIN 0.577 and AVTX at VPIN 0.725) produced KILL conflicts that demoted them from LONG candidates to WATCH. The discriminatory power of the layer is visible in this run: the methodology distinguishes between superficially similar pre-market moves on the basis of microstructural toxicity, and DGXX passed the discrimination cleanly.

What would have suppressed it. A VPIN reading above the configured threshold (currently anchored to a 16-snapshot exponentially weighted standard deviation band) would have triggered a KILL conflict and demoted the LONG candidacy to WATCH, regardless of the strength of corroborating signals from other layers.

Earnings Surprise — Post-Earnings Drift

What it looks for. The Earnings Surprise layer enriches candidates with their recent quarterly earnings history — beat rate over the most recent four reporting periods and average surprise magnitude — and applies a binary gate when a candidate is within zero days of an upcoming earnings event.

Pedigree. Post-earnings announcement drift is among the most thoroughly documented anomalies in the empirical asset-pricing literature. Ball and Brown established the foundational evidence in 1968,¹⁵ and Bernard and Thomas extended it in 1989 to demonstrate persistence over longer horizons than market efficiency would predict.¹⁶ The phenomenon has been replicated across markets and decades. The 0-day binary gate reflects the methodology's recognition that pre-announcement positions face informational asymmetries that retail traders are systematically disadvantaged in evaluating.

What it saw on DGXX. The earnings surprise enrichment did not produce a record for DGXX in this run. The ticker was not within the layer's lookback window for an upcoming earnings event, and the absence of an enrichment event indicates that the binary gate had no occasion to fire. The silence of this layer is itself a finding. The methodology requires explicit disclosure of which layers fired, which were silent, and why — and the silence in this case was the correct outcome of the layer's gating logic.

What would have suppressed it. Had DGXX been within the same-day earnings window, the binary gate would have demoted the verdict from LONG to WATCH regardless of corroboration from other layers. This is the same gate that, in the same run, demoted SMR — a ticker with elevated FireFly confidence and HBuy score — from LONG 65% to WATCH 55% because of a 0-day earnings flag.

Insider & Regulatory — SEC and FDA

What it looks for. The Insider/Regulatory layer queries two public-record streams: SEC Form 4 filings of insider transactions, and openFDA records of recent application activity (Approval, Complete Response Letter, and related events). Both streams are subject to mandatory disclosure timelines under federal regulation, and both produce timestamped records that can be cross-referenced against candidate tickers.

Pedigree. The predictive content of insider transactions has been documented since Lakonishok and Lee's 2001 study of insider trading patterns,¹⁷ with subsequent literature confirming that systematic insider buying — particularly cluster activity across multiple insiders within a short window — exhibits return predictability above the index baseline. The regulatory side draws on the well-characterized abnormal returns surrounding FDA application events for pharmaceutical and biotechnology issuers.¹⁸

What it saw on DGXX. The 08:15 ET run logged zero insider transactions and zero regulatory events for the candidate set on this morning. The openFDA scan completed normally, returning two applications with recent activity — neither matched DGXX. The absence is again a methodological finding rather than a gap. DGXX did not derive evidentiary weight from this layer in this run; the convergence calculation reflected this absence rather than substituting for it.

What would have contributed. A Form 4 filing within the lookback window indicating cluster insider buying would have added evidentiary weight to the convergence calculation. An FDA event matching the candidate would have similarly contributed. The absence of these is data, not noise.

Regime Gate — Market State

What it looks for. The Regime Gate evaluates the broad-market state at run time and applies a confidence floor to the methodology's verdicts on the basis of that state. The gate uses the SPDR S&P 500 ETF as its referent and classifies the regime into one of three states: NEUTRAL (favorable to long-side methodology output), CAUTION (degrades long-side confidence), and RED (suppresses long-side verdicts entirely).

Pedigree. The conditioning of cross-sectional return predictability on broad-market state is a well-established finding in the empirical asset-pricing literature.¹⁹ Strategies that perform reliably during favorable regimes deteriorate or invert during unfavorable ones, and the discipline of incorporating regime state into signal generation is a defense against the false-positive contamination that arises when methodology calibrated to one market state is applied unchanged to another.

What it saw on DGXX. At 08:15 ET on 2026-05-07, the regime gate evaluated SPY at $733.83, +0.12% on the session, classified the state as NEUTRAL, and raised the confidence floor for the run to 65. DGXX's eventual confidence of 81 cleared the floor with margin. The verdict was permitted to pass.

What would have suppressed it. A CAUTION classification would have raised the floor and degraded the methodology's tolerance for marginal candidates. A RED classification would have suppressed long-side verdicts altogether for the run.

NORTHSTAR Convergence

What it looks for. The seventh and integrative layer is the convergence requirement itself — the procedural rule that no single signal, regardless of its strength, produces a verdict on its own. The methodology requires evidentiary corroboration across multiple independent layers before a NORTHSTAR card is issued. This is the direct application of Locard's principle of independent corroboration to the analysis of public market data.

Pedigree. The convergence requirement is the procedural core of forensic methodology in the source disciplines. It is the rule that separates an examiner's defensible finding from a guess, and the rule that survives Daubert challenge in a courtroom. Its application to market analysis is documented in detail in Forensics.us SCIF Paper 003, "Stock Market Trace Evidence."

What it saw on DGXX. The convergence calculation for DGXX integrated five active evidentiary streams: a FireFly confidence of 74, an HBuy composite of 79 with confirmed trend and channel position, a clean Liquidity Forensics reading with no KILL conflict, a NEUTRAL regime classification, and the secondary HBuy watchlist tier (a separate confirmation pathway that recorded DGXX at $6.24 on the same morning). Two streams were silent — Earnings Surprise (no event window) and Insider/Regulatory (no Form 4 or FDA match). The methodology weighted the active streams against the silent ones, applied the regime floor, and produced a final NORTHSTAR confidence of 81.

The signal strength was classified as HIGH. The integer NORTHSTAR score recorded in the audit log was 3. The verdict was LONG.

What Happened

The DGXX position was tracked from the signal time of 08:15:47 ET through the trading session. The methodology's T1 target was $6.51, representing a 4.33% advance from the signal price. The T2 target was $6.68, representing a 7.05% advance.

By the time of the post-open evaluator pass — the 08:20 ET tracking run that updates outcome status for the morning's calls — DGXX had recorded a maximum gain of +6.25% from the signal price. The T1 target was hit. The T2 target was not hit. The position resolved as a documented T1_HIT outcome.

This is a modest result by the standards of the speculative trading content space. It is not a doubling. It is not a triple-digit-percentage move. It would not appear in a marketing screenshot designed to inflame the reader's regret about the move they did not catch. It is a documented, methodologically grounded outcome that fell within the methodology's stated targets and that survives the verification any sophisticated reader could apply to it. The audit log row that recorded this finding is reproducible from the published methodology. The signal price, the targets, and the maximum gain are not assertions; they are records.

The methodology will produce outcomes of this character — modest, bounded, documented — across a substantial fraction of its calls. It will also produce losses. It will produce calls that touch threshold but do not sustain. It will produce calls that resolve as MISSED when entries chase the move past the optimal entry zone. The discipline of the methodology is to record each of these outcomes with the same forensic preservation, regardless of whether the outcome flatters the platform or does not.

Chain of Custody

The Forensics.us audit log row for DGXX, 2026-05-07, run slot 1, contains the following fields. This is presented verbatim from the system's persisted record.

Each field in this record was computed by the methodology described in the preceding sections. Each was persisted at the moment of card generation. None has been altered since. The chain of custody from the underlying market data, through the seven evaluation layers, to this row, is reconstructable from the published methodology and the operational logs that the methodology produces. This is the difference between a published claim and a published finding.

The Procedural Argument

The argument of this paper has been procedural rather than promotional. The methodology described here does not promise that subscribers will become wealthy. It does not guarantee that any specific trade will produce a specific outcome. It does not, in the language of the sources of speculative trading content, tell anyone what to buy. What it does is produce documented findings under documented procedures, preserved with chain of custody and bounded by the regulatory frame that governs publishers under §202(a)(11)(D) of the Investment Advisers Act and the disclosure requirements of CFTC Rule 4.41.

This is a different product than the dominant pattern in retail trading content. It will not appeal to readers who arrive seeking unconditional performance promises, prepackaged buy lists, or guaranteed outcomes. It is built for readers who recognize that the discipline of forensic science — falsifiable methodology, independent corroboration, chain of custody, and explicit disclosure of failure modes — is the appropriate standard for any analytical practice that claims to inform consequential decisions.

The DGXX case study presented here is one row of an audit log that contains many rows. Some are wins. Some are losses. Some are touches that did not sustain. Each was generated by the same methodology, persisted at the moment of generation, and preserved without alteration. The reader who has reached this paragraph by way of seven signal layers, twenty-three audit fields, and nineteen academic citations has, in the language of the source disciplines, encountered evidence — and is in a position to evaluate it.

The methodology is the product. The audit log is the proof. The reader is the verifier.