Our Methodology

From Signal to Thesis: A Three-Phase Process

Pelican is an AI research platform for hedge funds. We turn market signals into defensible, cited investment theses through a three-phase workflow: (1) programmatic idea generation across six analyzers scored on a transparent 10-point scale, (2) a 25-stage, archetype-aware research gauntlet tailored to each company’s business model, and (3) an AI Agent Layer—a mixture-of-experts interface that makes every artifact in our database instantly queryable with citations. Below are the three phases.

Phase 1: High-Conviction Signal Generation

The process begins by filtering the market for actionable signals. We use six independent programmatic modules to scan millions of data points, each designed to capture a different source of potential alpha. Every signal is graded on a transparent, 10-point conviction scale.

Behavioral & Conviction Signals

• Insider Buys: We screen for direct, open-market purchases by corporate insiders (CEOs, CFOs, Directors), viewing these actions as the strongest signal of conviction in a company's value.
• 13F Institutional Analysis: We track the quarterly filings of over 50 of the world's most successful activist and value-oriented funds to detect "smart money" movements. By focusing on a curated list of elite managers—such as Elliott Management, Starboard Value, Pershing Square, and Trian Partners—we can identify high-conviction signals based on convergence (many top funds buying) and concentration (single funds making huge bets).
• Congressional Trades: We monitor stock purchases by members of Congress, with a heavy emphasis on bipartisan cluster buys—where multiple politicians from both parties acquire the same stock.

Fundamental & Market Signals

• Quantitative Screens: We apply rigorous Growth at a Reasonable Price (GARP) and Deep Value screens using TTM fundamental data to find companies that are statistically cheap or exhibit high financial quality.
• Price Target Analysis: We screen for significant upgrades from reputable sell-side analysts where new price targets imply substantial upside from the current price.
• Press Release Catalysts: An AI-powered module filters out routine announcements to score only definitive, positive catalysts (e.g., major contracts, spin-offs) with a 3 to 24-month horizon.

Phase 2: The 360-Degree Research Gauntlet

Every high-conviction signal automatically triggers a comprehensive due diligence report. This is where we apply the archetype-specific lens to analyze the company from every critical angle.

Pillar 1: The Quantitative Foundation

We conduct a forensic audit of the company's financial health to establish an objective baseline.

• Historical Performance & Volatility: We analyze trends and volatility in Revenue, Profit Margins, and Cash Flow, culminating in an Overall Financial Predictability score.
• Liquidity & Solvency Stress-Testing: We perform a full credit risk analysis, including a review of covenants and a calculation of the company's Liquidity Runway under severe stress.
• Asset Quality & Downside Protection: We perform a Net Liquidation Value (NLV) Analysis, applying conservative haircuts to estimate the absolute floor of value.

Pillar 2: Management & Accountability

We assess the competence and credibility of the leadership team, focusing on their alignment with shareholder interests.

• Forensic Capital Allocation: We quantitatively score management’s past capital allocation decisions (M&A, Buybacks, CapEx) against value-accretive criteria, resulting in an Overall Capital Allocation Skill Score.
• Execution Track Record: We systematically track management's historical guidance against actual results to quantify their forecast reliability and execution hit rate.
• Communication Credibility: We quantitatively score earnings call transcripts for Candor, Realism, and Clarity of Strategy to provide an objective benchmark for management's trustworthiness.

Pillar 3: Intrinsic & Relative Valuation

We triangulate valuation using multiple, archetype-appropriate methodologies to determine an intrinsic value range.

• Intrinsic Valuation: We build an archetype-specific Discounted Cash Flow (DCF) model to estimate fair value.
• Relative & Historical Valuation: We conduct a Comparable Company Analysis (Comps) and analyze current multiples against the company's own historical ranges.
• Scenario Modeling: The final valuation uses a Probability-Weighted Expected Return (PWER) framework, with Bull, Base, and Bear case probabilities informed by our analysis.

Pillar 4: The Strategic "Outside View"

We pressure-test the internal view against external market reality and strategic forces.

• External Reality Check: We use Open Source Intelligence (OSINT) to uncover unfiled risks, customer "ground truth," and contrarian bull/bear theses.
• Capital Cycle Analysis: We determine the industry's position in its capital cycle (from "Bust" to "Peak") to identify structural tailwinds or headwinds.
• Variant Perception: We systematically challenge consensus to articulate a non-consensus Variant Perception—what we believe the market is missing.

Phase 3: The AI Analyst Team & Conversational Interface

The entire data ecosystem—from the initial signals to the final research outputs—is made fully accessible through our AI Agent Layer. This is not a single chatbot; it is a "mixture of experts" system where a central orchestrator intelligently routes your questions to a team of specialist AI agents, each an expert in a specific domain.

• The Orchestrator: Acts as the central "brain," understanding your intent and directing your query to the specialist best equipped to provide a fast, accurate, and cited answer.
• Document Experts: A team of agents, each a specialist in a specific document type. Whether you're asking about an annual report (10-K), a proxy statement, or an earnings call transcript, your question is handled by an agent with deep expertise in that source.
• Live Data Agent: Connects to real-time APIs to answer questions about current market data, from stock prices and analyst targets to the full text of the latest 8-K filings.
• Idea Generation Agent: Allows you to conversationally explore the entire database of historical signals, filtering by conviction score, source, or theme.
• Synthesis Agent: The capstone agent that has read all of our completed research. It can synthesize findings across multiple reports and answer high-level questions about our final investment theses, catalysts, and risks.

This architecture ensures that every answer is not just a summary, but a piece of evidence, with every key fact cited directly back to its source document.

The Capstone: An Actionable Investment Thesis

The final output of this three-phase process is not just data; it's a decisive, actionable investment plan. It includes a clear Buy/Hold/Sell Recommendation, a probability-weighted price target, a specific position sizing rationale, and a detailed monitoring plan focused on tracking the primary catalyst and mitigating the primary risk ("Thesis Killer").