Simulation Core

The Heart of the Engine

The Simulation Core is a discrete, step-by-step annual processing pipeline. For each year of a projection, the engine executes a sequence of modules, each responsible for a specific aspect of ESOP administration.

This strict order of operations ensures that legal obligations are met before discretionary actions are taken.

Annual Processing Cycle

Processing Steps Overview

Step 0: Turnover Projection

Optional predictive moduleUses statistical models to forecast which employees will terminate in the current year.Inputs:

Participant age, tenure, compensation
Historical turnover rates
Industry benchmarks

Output:

List of projected termination events with probabilities

Step 1: Initialization

Calculate share price and prepare for annual processingInitialize the annual processing cycle by calculating current share price and preparing state variables.Actions:

Calculate per-share value from company equity and outstanding shares
Apply annual growth rate to company equity value
Initialize security-specific prices in multi-class mode
Set up year-specific state variables

Step 2: Determine Share Pool

Step 3: Allocate Shares

Distribute shares to participant accountsApply allocation formula to credit shares to individual accounts.Formula Options:

Pro-rata by compensation
Pro-rata by hours
Integrated (Social Security-adjusted)

Step 4: Calculate Contribution

Determine annual company contributionBased on contribution_policy in OperatingAssumptions.Policy Types:

Fixed amount
Percentage of payroll
Discretionary formula
Loan payment-based

Step 5: Update Vesting

Calculate vested balances and potential forfeituresApply the plan’s vesting schedule to determine vested vs unvested portions of participant accounts.Actions:

Apply vesting schedule based on years of service
Calculate vested percentages for each participant
Identify non-vested amounts subject to forfeiture
Track vesting per security in multi-class mode
Apply vesting to both shares and cash balances

Step 6: Process Diversification

Handle statutory diversificationProcess elections from eligible participants (age 55+ with 10+ years).Actions:

Identify eligible participants
Process diversification elections
Calculate amounts (25% or 50% of account)
Move funds to diversified investments

Step 7: Process Repurchases

Execute share repurchasesRepurchase shares from terminated participants using the Funding Waterfall.See: Funding Waterfall

Step 8: Year-End Closing

Finalize annual results and prepare for next yearRoll account balances forward, evolve employee data, and capture year-end snapshots.Actions:

Move allocated/diversified amounts to opening balances
Age employees by 1 year and increment service years
Apply compensation growth rates
Remove fully distributed participants
Capture year-end state snapshots and KPIs

Key Logic Modules

For leveraged ESOPs with multiple debt tranches, the engine implements precise loan-by-loan accounting.

Critical: Each ESOPLoan object directly owns the shares that collateralize it. This prevents cross-contamination of suspense accounts.

How It Works

Iterate Through Loans

for loan in esop_loans:
    if loan.principal_balance > 0:
        process_loan_payment(loan)

Calculate Payment

Determine principal and interest for current year based on loan terms

Release Shares

# Release shares proportional to principal paid
shares_to_release = (
    loan.suspense_shares * 
    (principal_payment / original_loan_amount)
)

loan.suspense_shares -= shares_to_release
share_pool += shares_to_release

Update Loan Balance

loan.principal_balance -= principal_payment

Example: Multi-Loan Scenario

# Year 2025 Processing

# Loan 1: Original $3M loan from 2020
loan_1 = ESOPLoan(
    loan_id="LOAN_2020",
    principal_balance=2_000_000,
    suspense_shares=20_000,
    annual_payment=400_000  # $300K principal + $100K interest
)

# Loan 2: New $2M loan from 2023
loan_2 = ESOPLoan(
    loan_id="LOAN_2023",
    principal_balance=1_800_000,
    suspense_shares=15_000,
    annual_payment=300_000  # $200K principal + $100K interest
)

# Process Loan 1
shares_released_loan_1 = 20_000 * (300_000 / 3_000_000) = 2_000 shares
loan_1.suspense_shares = 20_000 - 2_000 = 18_000

# Process Loan 2
shares_released_loan_2 = 15_000 * (200_000 / 2_000_000) = 1_500 shares
loan_2.suspense_shares = 15_000 - 1_500 = 13_500

# Total shares available for allocation
share_pool = 2_000 + 1_500 = 3_500 shares

Why This Matters: Without loan-by-loan tracking, shares from one loan could incorrectly be released when paying down another loan, violating ERISA requirements and creating audit risk.

Funding Waterfall

The repurchase processing module implements a strict, rules-based sequence for drawing funds from the trust’s cash accounts.

The Waterfall Sequence

The engine follows PlanRules.cash_usage_policy to draw funds in the specified order:

cash_usage_policy = [
    "unallocated_company_contributions",  # 1st priority
    "unallocated_forfeiture_cash",        # 2nd priority
    "participant_cash_accounts"           # 3rd priority
]

Processing Algorithm

Calculate Total Repurchase Need

total_repurchase_obligation = sum(
    participant.vested_shares * current_share_price
    for participant in terminated_participants
)

Apply Waterfall

remaining_need = total_repurchase_obligation

for cash_source in cash_usage_policy:
    if remaining_need <= 0:
        break
        
    available = trust_cash_ledger[cash_source]
    amount_to_use = min(available, remaining_need)
    
    trust_cash_ledger[cash_source] -= amount_to_use
    remaining_need -= amount_to_use
    
    log_transaction(cash_source, amount_to_use)

Handle Shortfall

if remaining_need > 0:
    # Unfunded repurchase obligation
    defer_to_next_year(remaining_need)
    # OR trigger company loan/contribution

Example: Waterfall in Action

Repurchase Need: $500,000

Trust Cash Ledger:
├─ unallocated_company_contributions: $200,000
├─ unallocated_forfeiture_cash: $150,000
└─ participant_cash_accounts: $300,000

Waterfall Execution:
1. Draw $200,000 from unallocated_contributions → $300,000 remaining
2. Draw $150,000 from unallocated_forfeitures → $150,000 remaining
3. Draw $150,000 from participant_cash → $0 remaining

Result: ✅ Fully funded

Legal Compliance: The order matters! For example, forfeiture cash often has restrictions on use. The waterfall ensures compliance with plan document rules and ERISA regulations.

State Capture

At the end of each annual cycle, the engine captures complete snapshots:

Company State
Trust State
Participant Snapshot

{
  "simulation_run_id": "run_123",
  "year": 2025,
  "revenue": 10_000_000,
  "ebitda": 2_200_000,
  "total_payroll": 3_500_000,
  "esop_contribution": 500_000,
  "source_type": "simulated"
}

{
  "simulation_run_id": "run_123",
  "year": 2025,
  "total_shares_outstanding": 100_000,
  "allocated_shares": 68_000,
  "suspense_shares": 32_000,
  "unallocated_shares": 0,
  "cash_ledger": {
    "unallocated_contributions": 50_000,
    "unallocated_forfeitures": 25_000,
    "participant_cash": 125_000
  },
  "loans": [...]
}

{
  "simulation_run_id": "run_123",
  "year": 2025,
  "participant_id": "EMP001",
  "allocated_shares": 1_200,
  "vested_percentage": 0.80,
  "vested_shares": 960,
  "account_value": 132_000,
  "cash_balance": 2_500,
  "status": "active"
}

Error Handling & Validation

The engine performs extensive validation at each step:

Input Validation

Schema compliance
Business rule checks
Data completeness
Referential integrity

Processing Checks

Share count reconciliation
Cash balance validation
Loan payment calculations
Legal compliance flags

Output Verification

Total shares consistency
Cash flow balance
Participant account totals
Year-over-year deltas

Audit Logging

Every transaction logged
Decision points captured
Assumption tracking
Error breadcrumbs

Performance Optimizations

Vectorized Calculations

Participant-level calculations use NumPy for efficient batch processing.

Lazy Loading

Historical snapshots loaded on-demand, not preloaded into memory.

Caching

Frequently accessed reference data (e.g., plan rules) cached per simulation run.

Parallel Execution

Independent scenario runs can execute in parallel for sensitivity analysis.

Next Steps

Processing Steps

Detailed breakdown of each simulation step

Data Layer

How state is persisted and retrieved

Data Models

Core objects used in processing

Examples

See the engine in action

Getting Started

Configuration Guide

For Enterprise

Architecture

Data Models

Simulation Pipeline

Examples

API Reference

API Schemas

The Heart of the Engine

Annual Processing Cycle

Processing Steps Overview

Key Logic Modules

How It Works

Example: Multi-Loan Scenario

Funding Waterfall

The Waterfall Sequence

Processing Algorithm

Example: Waterfall in Action

State Capture

Error Handling & Validation

Input Validation

Processing Checks

Output Verification

Audit Logging

Performance Optimizations

Next Steps

Processing Steps

Data Layer

Data Models

Examples

Getting Started

Configuration Guide

For Enterprise

Architecture

Data Models

Simulation Pipeline

Examples

API Reference

API Schemas

​The Heart of the Engine

​Annual Processing Cycle

​Processing Steps Overview

​Key Logic Modules

​Loan-by-Loan Share Release

​How It Works

​Example: Multi-Loan Scenario

​Funding Waterfall

​The Waterfall Sequence

​Processing Algorithm

​Example: Waterfall in Action

​State Capture

​Error Handling & Validation

Input Validation

Processing Checks

Output Verification

Audit Logging

​Performance Optimizations

​Next Steps

Processing Steps

Data Layer

Data Models

Examples

The Heart of the Engine

Annual Processing Cycle

Processing Steps Overview

Key Logic Modules

Loan-by-Loan Share Release

How It Works

Example: Multi-Loan Scenario

Funding Waterfall

The Waterfall Sequence

Processing Algorithm

Example: Waterfall in Action

State Capture

Error Handling & Validation

Performance Optimizations

Next Steps