Platform

The Methodology, Implemented

The thesis describes how disciplined analysis works. This page describes how we make it operational—systematic processes that execute consistently, without the variability that afflicts even disciplined human practice.

Two Modes of Operation

The analytical challenge differs by coverage. A stock with fifteen analysts covering it presents different opportunities than a stock with none. The platform operates in two modes, each designed for its context.

Mode One

Narrative Intelligence

For well-covered securities where consensus exists.

When ten analysts cover a stock, there is no information advantage. Everyone has the same filings. Everyone attends the same calls. The opportunity is not knowing something others don't—it's seeing where the shared interpretation may be flawed.

FunctionOutput
Map the narrativeExtract the consensus story from broker research, management commentary, investor positioning. What does the market believe?
Surface assumptionsWhat must be true for the narrative to hold? Growth rates, margin trajectories, competitive dynamics—made explicit.
Identify coherenceDoes the narrative align with the price? Do the assumptions cohere with each other? Where are the tensions?
Compare to evidenceHow does consensus compare to our cross-domain synthesis? Where do regulatory filings, academic research, or competitor disclosures suggest something different?
Set tripwiresFor each embedded assumption, define the conditions that would indicate it's breaking.
Monitor continuouslyWatch for tripwire conditions across all relevant sources.

The value: Not a better opinion about Woolworths. A systematic map of what the market believes, what must be true for that belief to be correct, what evidence would challenge it, and continuous monitoring for exactly that evidence.

Mode Two

Primary Analysis

For under-researched securities where consensus doesn't exist.

When no analyst covers a stock, there is no narrative to map. The opportunity is different: information asymmetry through effort. Doing the work that hasn't been done.

FunctionOutput
Apply frameworksMining stocks assessed against JORC standards. Biotech against clinical endpoint analysis. Infrastructure against regulatory and engineering constraints.
Synthesise evidenceTechnical literature, regulatory filings, academic research, operational data—integrated into assessment that doesn't exist elsewhere.
Generate analysisNot extracting consensus—creating the fundamental assessment. Management quality, competitive position, technical feasibility, execution risk.
Identify uncertaintiesWhat are the critical unknowns? What would resolve them? Where are the risks that aren't obvious?
Map the caseBull case, bear case, key assumptions, what would change the picture.

The value: For a junior mining company with no coverage, you receive the analysis a specialist would produce—technical assessment, risk mapping, scenario framework—without hiring the specialist or building the expertise over years.

The mode is determined by context. Well-covered stocks get narrative intelligence. Under-researched stocks get primary analysis. Some situations warrant both.

The Methodology Implemented

The thesis describes Analysis of Competing Hypotheses: generate alternatives, assess diagnosticity, seek disconfirmation, identify linchpins. This section describes how that methodology becomes operational infrastructure.

Hypothesis Generation

The problem: Analysts typically satisfice—accepting the first hypothesis that seems adequate without generating alternatives. If the correct explanation isn't among the hypotheses considered, it cannot be discovered.

The implementation: The platform generates hypothesis sets through multiple analytical strategies applied simultaneously:

Situational analysis: What does the specific evidence suggest about this company?
Historical comparison: What do analogous situations imply? How have similar patterns resolved?
Theoretical application: What do base rates indicate? What typically happens to companies with this profile?
Adversarial generation: What is the strongest possible counter-thesis?

The output: A complete hypothesis set—bull case, bear case, variant perceptions—generated before evaluation begins. The analyst cannot satisfice because alternatives are presented alongside any initial impression.

Diagnosticity Assessment

The problem: Most evidence is consistent with multiple hypotheses. Analysts accumulate "supporting" evidence without recognising it equally supports alternatives they never considered.

The implementation: For each piece of evidence, the platform assesses its relationship to each hypothesis:

EVIDENCE ITEM: Q3 gross margin declined 80bps sequentially
Relationship to hypotheses:
├── H1 (Temporary competitive pressure): CONSISTENT
├── H2 (Structural margin impairment): CONSISTENT
├── H3 (Deliberate investment for share): CONSISTENT
└── H4 (Accounting timing issue): INCONSISTENT
Diagnosticity: LOW for H1/H2/H3 discrimination
HIGH for H4 exclusion

The output: Not just evidence, but evidence with explicit diagnostic value. The analyst sees which evidence actually helps choose between alternatives and which is noise—consistent with everything and therefore informative about nothing.

Systematic Disconfirmation

The problem: Analysts naturally seek confirming evidence. They notice what supports their view and rationalise what contradicts it.

The implementation: For each hypothesis, the platform conducts adversarial search—specifically seeking evidence that would undermine it, constructing the strongest possible case against it.

HYPOTHESIS SURVIVAL ASSESSMENT
H1: Quality compounder (structural advantages sustaining)
├── Disconfirming evidence found: 2 items
├── Attempted disconfirmation: 12 searches
└── Survival score: MODERATE
H2: Mature business facing structural decline
├── Disconfirming evidence found: 1 item
├── Attempted disconfirmation: 12 searches
└── Survival score: HIGH
H3: Turnaround in progress
├── Disconfirming evidence found: 4 items
├── Attempted disconfirmation: 12 searches
└── Survival score: LOW
Ranking by survival: H2 > H1 > H3

The output: Hypothesis survival scoring—ranking by what has not been disproved rather than what has been "confirmed."

Linchpin Identification

The problem: A few items of evidence or assumption typically drive any conclusion. Analysts often don't recognise which ones. When analysis proves wrong, it's usually because key assumptions went unchallenged.

The implementation: The platform identifies which evidence items have the highest impact on hypothesis rankings:

LINCHPIN ANALYSIS
Linchpin #1: Management guidance for 15% ROIC on automation investment
Impact: Removing this evidence shifts ranking from H1 to H2
Current interpretation: Automation programme will generate returns above cost of capital
Assumed: Management guidance is credible; technology will perform as claimed
Alternative interpretations:
├── Management is optimistic (historical accuracy: 62%)
├── ROIC calculation excludes implementation costs
└── Technology base rates suggest 8-10% more likely
Sensitivity: HIGH
Recommendation: Seek independent validation; examine peer outcomes

The output: Explicit identification of what drives the conclusion, with stress-testing of each linchpin assumption.

Workflow Consistency

The critical difference: these processes execute identically every time.

Human analysts apply rigorous method inconsistently—when fresh, when time permits, when stakes seem high enough. The platform applies the same hypothesis generation, the same diagnosticity assessment, the same disconfirmation search, the same linchpin identification to every analysis.

No degradation under time pressure. No skipped steps when deadlines are tight. No variation with fatigue or workload.

Cross-Domain Synthesis

The thesis explains that different sources have different relationships to truth. This section describes how that epistemic framework becomes operational.

DomainEpistemic StatusWhat It Reveals
Corporate disclosuresMotivated communicationWhat management wants believed. Useful for understanding narrative, not for validating it.
Regulatory filingsAdversarial scrutinyWhat authorities observe under examination. Often surfaces what companies minimise.
Legal proceedingsTestimony under oathWhat's asserted when accuracy is compelled. Numbers that matter when stakes are high.
Academic researchMethodological rigourWhat evidence-based inquiry concludes. Claims tested and peer-reviewed.
Competitor disclosuresRevealed preferenceWhat rivals show through actions. Market dynamics visible before acknowledged.
Technical literatureEngineering realityWhat's actually achievable. Physical constraints underneath the promises.

Triangulation, not aggregation.

The platform doesn't just collect more information. It triangulates across sources with different biases and different incentives.

When domains agree: Confidence is warranted. Independent sources with different motivations reaching the same conclusion is meaningful.

When domains disagree: The disagreement is informative.

DOMAIN DISAGREEMENT DETECTED
Topic: Capacity utilisation at Newcastle facility
Corporate disclosure (FY24 Annual Report):
"Operating at optimal capacity with room for efficiency gains"
Regulatory filing (EPA compliance report):
Production volume data implies 67% utilisation
Technical assessment (engineering analysis):
Facility design capacity suggests 71% utilisation
Competitor disclosure (peer investor day):
"Industry-wide overcapacity pressuring margins"
Assessment: Corporate narrative inconsistent with regulatory and technical evidence.
Diagnostic value: HIGH - discriminates between H1 and H2

The Tripwire System

The thesis establishes the principle: pre-commitment to revision reduces rationalisation. This section describes the implementation.

How tripwires work.

For each assumption underlying a thesis, the platform defines falsification conditions:

THESIS: [Company] is a quality compounder with sustainable advantages
ASSUMPTION 1: Gross margins will remain above 35%
Tripwire conditions:
├── Two consecutive quarters below 35%
├── OR single quarter below 32%
└── OR management commentary indicating structural pressure
Monitoring: Quarterly filings, earnings transcripts, competitor pricing
Status: NOT TRIGGERED
ASSUMPTION 2: Market share in core segment is stable
Tripwire conditions:
├── Market share decline >200bps in trailing twelve months
├── OR competitor announces capacity expansion >15%
└── OR customer concentration increases above 25%
Monitoring: Industry data, competitor filings, company disclosures
Status: NOT TRIGGERED

What happens when triggered.

A triggered tripwire doesn't mechanically dictate action. It dictates re-examination.

⚠️TRIPWIRE TRIGGERED
Assumption: Market share in core segment is stable
Condition: Competitor announces capacity expansion >15%
Evidence: [Competitor] investor day announced 22% capacity expansion in [region], completion Q3 2026

Required action: Re-examine thesis

The default has shifted. Your view is flagged for reconsideration. The burden is now on finding reasons NOT to update.

The analyst must engage with the triggered condition. They can dismiss it—but they must provide a reason, which is logged. Silent rationalisation becomes visible.

Pre-commitment before pressure.

Tripwires are defined when the thesis is formed—before contradictory evidence appears. This is essential.

When evidence arrives that challenges a view, the pressure to rationalise is immediate. When conditions are defined in advance, the question becomes simpler: "Has my condition been met?" The standard was set when there was no pressure to rationalise.

Catalyst Monitoring

The thesis identifies nine catalyst types beyond new information. This section describes how they're monitored.

Attention shifts

The information was always available but not salient. A journalist writes about it. Suddenly what was known becomes noticed.

Frame changes

The facts haven't changed but the interpretation has. "Investment for growth" becomes "cash burn."

Narrative exhaustion

The story runs out of energy. Good news stops moving the stock. The narrative has been fully priced.

Social proof fracture

When one respected voice breaks ranks, it creates permission for others. The cascade reverses.

Time revelation

Narratives embed implicit predictions. When the deadline passes without the outcome, credibility erodes.

Reflexivity unwinding

Some narratives are self-reinforcing. When the loop runs out of new believers, the mechanism reverses.

Personnel turnover

New analysts arrive without anchoring to the old narrative. They see the situation fresh.

Adjacent events

A competitor's failure forces the question: "Could it happen here?"

Monitoring implementation.

The platform doesn't just watch for new data. It watches for catalyst conditions.

CATALYST MONITORING: Active for [Company]
Narrative exhaustion signals:
├── Price response to positive news (last 6 months)
└── Tracking: Response decaying (was +2.3%, now +0.4%)
├── Positive surprise without follow-through
└── Last 3 beats: Average 1-day fade of -1.8%
└── Assessment: EARLY WARNING - narrative may be tiring
Time revelation signals:
├── Implicit predictions tracked:
├── "Margin recovery by H2 FY25" - 4 months remaining
├── "Cost programme completion Q1 FY25" - EXPIRED, incomplete
└── "New product launch FY25" - on track
└── Assessment: ONE PREDICTION EXPIRED

Most monitoring systems watch for new information—but new information is only one of nine catalyst types. The platform monitors for conditions under which the market might update, not just data that might be new.

What You Receive

The methodology produces specific outputs. This section describes what the analyst actually gets.

For Narrative Intelligence (Well-Covered Stocks)

Narrative Map

Consensus thesis extracted. Key assumptions identified and made explicit. Internal coherence assessment.

Divergence Assessment

Comparison of narrative to cross-domain evidence. Internal divergence (where the market disagrees with itself). External divergence (where our synthesis differs).

Hypothesis Matrix

Alternative interpretations with survival scoring. Evidence mapped to each hypothesis. Diagnosticity assessment. Linchpin identification.

Tripwire Dashboard

Assumptions underlying the thesis. Falsification conditions. Current status. Monitoring coverage.

Catalyst Status

Active catalyst conditions by type. Early warning indicators. Timeline pressures.

For Primary Analysis (Under-Researched Stocks)

Fundamental Assessment

Business model analysis. Competitive positioning. Management quality indicators. Financial assessment.

Technical Evaluation

Domain-specific assessment (JORC, clinical endpoints, engineering feasibility). Evidence from technical literature. Constraint identification.

Risk Map

Key uncertainties identified. Execution risks. Regulatory exposures. Scenario analysis.

Investment Framework

Bull case with required assumptions. Bear case with required assumptions. What would change the picture.

Ongoing monitoring.

Both modes include continuous monitoring with alerts when significant signals are detected:

📊ALERT: Tripwire approaching
Company: [Company]
Assumption: Gross margins remain above 35%
Current status: Q1 margin at 35.2% (approaching threshold)
Risk: Q2 guidance implies potential breach

Recommendation: Review thesis assumptions before Q2 results

Every Claim Traced

For fundamental investors, an insight without attribution is worthless. Every claim the platform makes traces to its source.

"ACCC found concentration levels in grocery retail create conditions where coordinated conduct is more likely." [1]

[1] ACCC Supermarket Inquiry Final Report (2025)
Section: Chapter 4 - Market Structure Analysis
Page: 47, Paragraph: 3
[Open source document] [View in context]

Verification: You can check the source yourself.

Context: You can read surrounding material.

Reliability: You can evaluate the source's epistemic status.

Audit trail: You can trace how conclusions were reached.

The platform does not generate claims from nothing. Every assertion connects to evidence. Every evidence item connects to a source. Every source can be examined.

The discipline, made operational.

The thesis describes what disciplined analysis requires: generating alternatives rather than satisficing, seeking disconfirmation rather than confirmation, assessing what evidence actually discriminates, pre-committing to revision before rationalisation sets in.

The platform implements that discipline as infrastructure—processes that execute consistently, comprehensively, without the variability that afflicts even the most careful human practice.

This is not AI replacing judgment. It is AI implementing the analytical method that human cognition cannot sustain unaided.

The judgment remains yours. The discipline is ours.

Request Access

Continuum is in private access with select institutional investors.