Financial Health Scoring Skill

Purpose

Score a biotech company's financial survivability to produce a normalized 0-100 financial health score. This skill encodes the exact rules from Wake Robin's Module 2 pipeline (v1 + v2), the Dilution Risk Engine, Liquidity Scoring, and Short Interest Engine so that scoring is reproducible and auditable.

Preconditions

• •All arithmetic MUST use Decimal (never float). Initialize from strings: Decimal("500000000").
• •All dates MUST be ISO 8601 (YYYY-MM-DD). Never call datetime.now().
• •PIT safety: only use data where source_date <= as_of_date - 1.
• •Rounding: ROUND_HALF_UP. Scores to 2 dp (0.01), rates to 4 dp (0.0001).

Step 1: Determine Cash Burn Rate

Use a hierarchical priority to select the best available burn rate proxy. Stop at the first available source.

Burn Rate Source Priority

YTD Period Detection (by filing date month)

Step 2: Compute Cash Runway

runway_months = current_cash / abs(quarterly_burn) * 3

If burn rate is zero or positive (cash-generating), runway_months = 1200 (100 years, effectively infinite).

Runway Severity Classification

Runway Score (v1, tier-based)

Runway Score (v2, piecewise linear)

Linear interpolation between breakpoints.

Step 3: Burn Acceleration Analysis (v2 only)

Detect quarter-over-quarter changes in burn rate.

Acceleration Penalty Formula

penalty_pct = min(0.30, avg_qoq_change / 100 * 0.5)
adjusted_runway_score = runway_score * (1.0 - penalty_pct)

Step 4: Dilution Risk Scoring

Cash-to-Market-Cap Sigmoid (v1)

sigmoid = 100 / (1 + exp(-15 * (cash_to_mcap - 0.15)))

• •Inflection point (midpoint): 15% cash/mcap
• •Steepness parameter (k): 15.0
• •Clamp exp input to [-50, 50] to avoid overflow
• •Final score clamped to [0, 100]

Runway-Based Penalty (v1)

If runway_months < 12:

penalty_factor = clamp(0.5 + (runway_months / 24), 0.5, 1.0)
dilution_score = dilution_score * penalty_factor

Dilution Risk Buckets (v2)

Financing Pressure Score (v2, 0-100)

Average of three components:

Runway Component:

Cash/Mcap Component:

Share Dilution Component (if available):

Step 5: Dilution Risk Engine (Forced-Raise Probability)

Used for catalyst-aware dilution analysis. Computes probability that a company will need to raise capital before its next catalyst.

Core Parameters

• •RISK_SCORE_MIN: 0.0
• •RISK_SCORE_MAX: 1.0
• •USABLE_CAPACITY_FACTOR: 0.70 (only 70% of shelf/ATM capacity is realistically accessible)

Cash Gap Calculation

monthly_burn = abs(quarterly_burn) / 3
months_to_catalyst = days_to_catalyst / 30.44
cash_needed = monthly_burn * months_to_catalyst
usable_capacity = (shelf_capacity + atm_remaining) * 0.70
total_available = current_cash + usable_capacity
cash_gap = cash_needed - total_available

Raise Feasibility

dilution_pct_mcap = cash_gap / market_cap
days_to_raise = cash_gap / (avg_daily_volume * share_price * 0.10)

cap_penalty = min(1.0, dilution_pct_mcap / 0.20)
volume_penalty = min(1.0, days_to_raise / 30)
raise_feasibility = 1.0 - ((cap_penalty + volume_penalty) / 2)
raise_feasibility = clamp(raise_feasibility, 0.0, 1.0)

Hard Limits

• •DILUTION_PCT_MCAP_HARD_THRESHOLD: 0.20 (>20% of mcap is very difficult)
• •DAYS_TO_RAISE_HARD_THRESHOLD: 30 days
• •DAILY_VOLUME_UTILIZATION: 0.10 (10% of daily volume)

Risk Bucketing

Confidence Factors (sum, clamped to [0, 1])

Step 6: Liquidity Assessment

Market Cap Tier Boundaries

ADV (Average Daily Volume) Thresholds by Tier

ADV Scoring (0-70 pts)

ratio = adv / (2 * tier_threshold)
adv_score = clamp(int(ratio * 70), 0, 70)

Spread Scoring (0-30 pts)

Penny Stock Penalty

• •Threshold: Price < $2.00
• •Effect: Max liquidity score capped at 10

Dollar ADV Tiers (v1 scoring)

Liquidity Hard Gates

Liquidity Risk Flags

• •FLAG_WIDE_SPREAD: spread >= 400 bps
• •FLAG_LOW_LIQUIDITY: ADV < tier threshold
• •FLAG_PENNY_STOCK: price < $2.00

Step 7: Revenue Scoring (v1)

Pre-Revenue Baseline

Companies with no revenue start at 50 pts (neutral, no penalty).

Three Components

Presence Bonus (binary):

• •Revenue >= $10M: +40 pts
• •Revenue < $10M: 0 pts

Scale Bonus (log-bucketed):

Coverage Penalty (if burning despite revenue):

Maximum revenue score: 80 pts (40 + 40 + 0).

Step 8: Short Interest Signal

Squeeze Potential Classification

Signal Components (base score = 50)

Trend Contributions (from SI % change)

Covering (bullish):

• •<= -20%: +15, <= -10%: +8, <= -5%: +4

Building (bearish):

• •

  = +20%: -12, >= +10%: -6, >= +5%: -3

Signal Direction

Crowding Risk

Step 9: Compute Composite Financial Score

V1 Weights

V2 Weights

financial_score = sum(component_score * weight for each component)
financial_score = clamp(financial_score, 0, 100)

Step 10: Apply Severity Penalties

final_score = financial_score * severity_multiplier

Data Quality Requirements

Minimum Field Coverage

• •Financial Field Coverage >= 50%: proceed with scoring
• •Financial Field Coverage < 50%: issue warning, score with available data
• •Market Field Coverage >= 80%: required for inclusion
• •Market Field Coverage < 80%: exclude ticker

Data Quality States

Composite Integration

The financial score enters Module 5 composite with these weights:

Pre-Flight Checks

Before scoring, verify:

1. •Financial records are PIT-admissible (source_date <= as_of_date - 1)
2. •Cash and market cap are present and positive (Decimal type)
3. •At least one burn rate source is available (otherwise SEV1)
4. •Runway is computed before dilution scoring (dependency)
5. •Liquidity data (ADV) is available for gating check
6. •All score components use Decimal arithmetic
7. •Output includes both financial_score and legacy financial_normalized fields (same value)
8. •Output includes _governance block with score_version, schema_version, pit_cutoff

Source Files