A Forensic Guide: Using Extreme Value Theory (EVT) to Find Weaknesses in Pharma and Biotech Pitch Decks

When a venture capital (VC) firm or a legal team evaluates an early-stage biotech or pharma startup, due diligence is critical. The stakes are high: investments hinge on scientific credibility, commercial viability, and the ability to navigate regulatory hurdles. This guide details a forensic, red-team approach to uncovering weaknesses in pitch decks using Extreme Value Theory (EVT) and related statistical frameworks. The aim is not only to identify flaws but also to stress-test assumptions and ensure that the presented narrative withstands scrutiny.

This detailed guide outlines how to apply EVT in a "red team" evaluation, where the goal is to stress-test a pitch deck, expose weaknesses, and identify opportunities for improvement.

1. Framework for the Forensic Analysis

1.1 Adopt a Critical and Forensic Mindset

• Assume Bias: Begin with the assumption that the pitch deck is designed to highlight strengths and obscure weaknesses. Scrutinize every claim, particularly those that seem too optimistic or lack supporting data.
• Focus on Extremes: Use EVT to identify tail risks—those rare but severe events that startups may neglect or intentionally downplay.
• Prioritize Transparency: Insist on granular data and explanations for omissions, thresholds, or unusual reporting practices.

1.2 Define Key Areas of Investigation

• Clinical Risks: Adverse events, dose-limiting toxicities (DLTs), and variability in efficacy.
• Financial Models: Pricing assumptions, funding requirements, and cost projections.
• Market Access: Payer acceptance, regulatory timelines, and geographic variability in health technology assessments (HTAs).
• Operational Vulnerabilities: Manufacturing, supply chain, and commercialization readiness.

2. EVT Applications in Clinical Data Analysis

2.1 Safety Data: Adverse Events and Toxicities

Safety profiles are critical for regulatory approval, payer acceptance, and public perception. EVT is particularly valuable in modeling rare but severe adverse events.

What to Analyze

• Threshold Selection:
  • Are safety thresholds (e.g., for toxicity) clinically validated, or arbitrarily set to exclude problematic data?
  • Example: A trial might define SAEs as liver enzyme elevations above 5x the normal limit, omitting cases just below the threshold that still carry significant risks.
• Dose Cohort Stratification:
  • Are data aggregated across dose groups? This can mask toxicity trends in higher doses.
• Tail Risks:
  • Are extreme adverse events (e.g., anaphylaxis or cytokine release syndrome) modeled, even if they are rare?

How EVT Helps

• Peaks Over Threshold (POT):
  • Identify and model events that exceed specific safety thresholds.
• Fréchet Distribution:
  • Analyze heavy-tailed risks, such as catastrophic toxicities, to estimate their frequency and potential impact.

Questions to Ask

• Why were specific safety thresholds chosen? Are they based on regulatory guidelines or designed to minimize reported risks?
• How does safety data differ between cohorts or dose groups? Are extreme events confined to higher doses?

Red Flags

• Absence of stratified safety data by dose.
• Over-reliance on pooled safety data.
• Lack of acknowledgment or modeling of rare severe events.

2.2 Efficacy Data: Variability and Super-Responders

Startups often focus on average efficacy metrics, which can obscure variability, outliers, and non-responders.

What to Analyze

• Super-Responders:
  • Are claims of exceptional efficacy supported by data? Are super-responders linked to identifiable biomarkers or random outliers?
• Response Variability:
  • Does the company report variability (e.g., standard deviation or full distribution) or only mean/median values?
• Subpopulation Analysis:
  • Are there identifiable subgroups with markedly different responses (e.g., biomarker-positive patients)?

How EVT Helps

• Gumbel Distribution:
  • Model moderately extreme efficacy outcomes to evaluate their consistency and significance.
• Weibull Distribution:
  • Capture bounded efficacy measures (e.g., biomarker expression) to assess upper limits of treatment effectiveness.

Questions to Ask

• Are super-responders reproducible across cohorts or sites?
• How is efficacy variability presented? Is it obscured by focusing solely on averages?

Red Flags

• No discussion of non-responders or variability.
• Highlighting super-responders without linking them to identifiable markers.
• Lack of subgroup analyses.

3. Financial Analysis: Stress-Testing Assumptions

3.1 Pricing Models

Pricing is a critical determinant of revenue potential but often relies on optimistic assumptions.

What to Analyze

• Payer Dynamics:
  • Are pricing assumptions aligned with payer expectations for cost-effectiveness and budget impact?
• Price Caps:
  • Has the company modeled extreme scenarios, such as payer-imposed caps or significant price reductions under frameworks like the Inflation Reduction Act (IRA)?
• Regional Variability:
  • Are pricing models adjusted for geographic differences in HTA criteria?

How EVT Helps

• Model extreme pricing outcomes, such as significant discounts or reimbursement rejections.
• Stress-test revenue projections under adverse pricing scenarios.

Questions to Ask

• What pricing assumptions underpin the revenue forecast? Are they realistic given payer constraints?
• What happens to revenue if key payers impose a 50% price reduction?

Red Flags

• Assumptions of uniform global pricing.
• No modeling of payer-imposed caps or discounts.
• Lack of sensitivity analysis for pricing variability.

3.2 Funding and Burn Rate

Startups often underestimate the impact of delays or cost overruns on cash flow.

What to Analyze

• Timelines:
  • Are development timelines realistic, with buffers for delays in recruitment or regulatory approvals?
• Funding Requirements:
  • Are funding projections stress-tested for increased costs or extended timelines?
• Commercialization Costs:
  • Are post-approval expenses, such as manufacturing and salesforce development, included?

How EVT Helps

• Use EVT to model financial risks from delayed milestones or expanded costs.
• Evaluate funding needs under pessimistic scenarios.

Questions to Ask

• What are the implications of a six-month trial delay on cash flow?
• How does the company plan to bridge funding gaps if commercialization costs exceed expectations?

Red Flags

• No contingency for trial delays or cost overruns.
• Optimistic cash flow projections without buffers.

4. Market Access and Operational Risks

4.1 Regional Market Entry

Market access strategies must account for fragmented HTA systems, payer variability, and regulatory complexities.

What to Analyze

• HTA Variability:
  • How does the company address differences in payer criteria across geographies (e.g., NICE vs. IQWiG)?
• Early Payer Engagement:
  • Are Phase 2 trials designed with payer expectations in mind?

How EVT Helps

• Model delays or restrictions in key markets due to payer rejections or HTA variability.
• Evaluate the likelihood of staggered or partial market entry.

Questions to Ask

• What are the fallback plans if reimbursement is delayed in major markets?
• How does Phase 2 data align with payer requirements for pricing and reimbursement?

Red Flags

• Assumptions of uniform market access.
• Lack of payer engagement during Phase 2.

4.2 Black Swan Events

Rare but catastrophic events—such as supply chain failures or regulatory rejections—can derail development.

What to Analyze

• Operational Resilience:
  • Does the company have contingency plans for manufacturing or supply chain disruptions?
• Regulatory Challenges:
  • Are there plans to address potential regulatory pushback?

How EVT Helps

• Quantify the impact of Black Swan events on timelines and costs.
• Stress-test operational plans for rare but severe disruptions.

Questions to Ask

• What happens if the primary manufacturer experiences a major disruption?
• How does the company plan to address regulatory delays or rejections?

Red Flags

• No contingency plans for operational disruptions.
• Overconfidence in regulatory timelines.

5. Presentation and Data Transparency

5.1 Aggregation and Reporting

Aggregated data often masks variability or tail risks.

What to Analyze

• Are safety and efficacy data stratified, or aggregated to obscure subgroup-specific risks?
• Are outliers included, or selectively excluded?

Questions to Ask

• Are data distributions provided, or only summary statistics?
• How are outliers explained or excluded?

Red Flags

• Lack of stratified analyses.
• Absence of raw data visualizations.

Conclusion

Using Extreme Value Theory in a red team analysis allows for a forensic examination of pharma and biotech pitch decks. This method ensures that no assumption goes unchallenged, from clinical risks to financial projections and market strategies. By focusing on tail risks, rare opportunities, and operational resilience, the evaluator can uncover weaknesses, ensure transparency, and provide actionable recommendations for improvement. This rigorous approach not only protects investors but also strengthens the company’s ability to withstand scrutiny and succeed.