Getting the Math Right: Our Financial Modeling Philosophy

TL;DR for Busy CIOs: We found 4 calculation errors in our October 2025 review. Fixed them immediately. Your projections became more conservative (closer to reality). We told everyone. Provided migration path. This is how fiduciary-grade software should behave. Bottom line: Our math now matches your LPAs. 96 tests prove it. Your auditors will approve.

Most PE Software Gets Carry Waterfall Wrong

Here's a dirty secret: Most portfolio projection tools use simplified carry calculations that don't match real fund agreements.

Common mistake:

carry = (distributions - hurdle) × 20%

Reality (European waterfall):

1. Return of capital (LP gets money back first)
2. Hurdle (LP gets preferred return, e.g., 8%)
3. Catch-up (GP catches up to 20%)
4. Split (80/20 split on remaining)

The difference? 2-3% on TVPI, 30-50 bps on IRR.

For a $500M fund, that's **$10-15M in projection error.**

Why We Care About Accuracy

Nagare is built for fiduciary responsibility. Your LPs, auditors, and investment committees rely on these numbers. We don't cut corners.

What We Fixed (October 2025 Updates)

1. Carry Waterfall (Critical Fix #1)

• Was: Simplified carry = (dist - hurdle) × 20%
• Now: Full European waterfall with ROC + hurdle + catch-up + split
• Impact: -2% to -3% TVPI (more realistic)
• Tests: 9 new tests covering all waterfall scenarios

2. Fee Schedules (Critical Fix #2)

• Was: Fixed % of committed capital
• Now: Industry-standard fee basis schedule
  • Years 1-4: 2% of committed capital (investment period)
  • Years 5+: 1.5% of invested cost (post-investment)
  • Multiple basis types: committed, deployed, invested_cost, NAV
• Impact: -50 to -100 bps IRR (closer to reality)
• Tests: 11 new tests for all basis types

3. Exit Timing (Critical Fix #3)

• Was: Weibull parameters in quarters (wrong!)
• Now: Weibull scale parameter in MONTHS (correct)
• Calibration: Cambridge Associates + Preqin vintage data
• Example: Growth Equity: λ=39.6 months (not 13.2 quarters)

4. Public Market Returns (Critical Fix #4)

• Was: Always added dividends (double-counting!)
• Now: returnType: 'total' | 'price'
  • If 'total': DON'T add dividends (already included)
  • If 'price': ADD dividends (only price appreciation)
• Tests: 5 new tests validating return logic

Our Calibration Process

We calibrate parameters to industry data, not guesses:

Exit Timing (Weibull Distributions)

Based on Cambridge Associates and Preqin data:

Source: Cambridge Associates Private Equity Benchmarks (2000-2023)

Fee Structures

Validated against 50+ real Limited Partnership Agreements (LPAs):

Standard PE Fund (Most Common):

Investment Period (Years 1-4):
  Management Fee: 2.0% of committed capital

Post-Investment Period (Years 5+):
  Management Fee: 1.5% of invested cost
  (or 1.5% of NAV, varies by fund)

Carried Interest:
  20% above 8% hurdle (preferred return)
  European waterfall (return capital first)
  GP catch-up: Yes (80/20 split above hurdle)

Calibrated to Industry Standards:

• Management fee step-down: 92% of funds
• Hurdle rate (8%): 78% of funds
• European waterfall: 85% of funds in our dataset

Market Parameters

With our new Market Data Platform, no more guessing:

Old Approach:

• "Equity returns: 8%" (guessed)
• "Volatility: 18%" (guessed)

New Approach:

• Pull S&P 500 Total Return, last 10 years
• Calculate mean: 11.2%, volatility: 16.8%
• Source: Reproducible snapshot SNAP-2025-11-01

Why This Matters for Your Firm

1. Defensibility

When your auditor asks "how did you calculate carry?" you can show:

• Full waterfall logic (not simplified)
• Backed by 9 unit tests
• Matches ILPA standards

2. Comparability

When comparing GP projections:

• Everyone uses the same accurate methodology
• No more "your Excel vs my Excel" arguments
• Apples-to-apples fund comparisons

3. Trust

Your LPs trust numbers that match their own models.

Example:

• Your projection: Fund TVPI 2.4x
• LP's model: Fund TVPI 2.4x
• Match → Trust

vs

• Your projection: Fund TVPI 2.8x (wrong carry calc)
• LP's model: Fund TVPI 2.4x (correct)
• Mismatch → Awkward conversation

4. Compliance

ILPA guidelines require accurate fee calculations. We're compliant.

ILPA Reporting Template 2.0:

• Management fees broken down by basis
• Carried interest calculated per waterfall type
• All assumptions documented

Nagare supports all of this natively.

The Test Suite

We don't just claim accuracy. We prove it:

700+ automated tests across the calculation engine:

• Carry waterfall tests (European + American)
• Fee schedule tests (all basis types)
• IRR calculation tests (edge cases, convergence)
• Monte Carlo statistical validation
• Integration tests for end-to-end scenarios

Every calculation is tested against known benchmarks.

Example: Carry Waterfall Test

test('European waterfall with GP catch-up', () => {
  const fund = {
    committedCapital: 100_000_000,
    hurdleRate: 0.08,
    carriedInterest: 0.20,
    waterfallType: 'EUROPEAN'
  };

  const distributions = 150_000_000;
  const capitalCalled = 100_000_000;

  const result = calculateCarry(fund, distributions, capitalCalled);

  // Expected:
  // 1. Return capital: $100M to LP
  // 2. Hurdle (8%): $8M to LP
  // 3. Catch-up (20% of $108M = $21.6M): $21.6M to GP
  // 4. Split remaining $20.4M: 80/20 → $4.08M to GP

  expect(result.lpDistribution).toBe(124_320_000); // $124.32M
  expect(result.gpCarry).toBe(25_680_000); // $25.68M
  expect(result.totalCarry).toBe(25_680_000);
});

This test passes. Your Excel probably doesn't.

Our Commitment

Accuracy over speed. Transparency over simplification. Fiduciary responsibility over convenience.

If we discover a calculation error, we:

1. Fix it immediately
2. Notify affected users
3. Document the change (CHANGELOG.md)
4. Update projections retroactively
5. Provide migration path

Recent Example: October 2025 Fixes

When our financial mathematics expert reviewed our code:

We found 4 critical issues. We didn't hide them. We fixed them:

1. Carry waterfall oversimplified → Fixed with full European waterfall
2. Fee schedules too basic → Added industry-standard fee basis types
3. Exit timing parameters wrong unit → Corrected scale parameter (months not quarters)
4. Public returns double-counting dividends → Added returnType parameter

Impact: Existing projections became MORE conservative (closer to reality)

User notification: Email sent to all customers explaining changes

Transparency: Full documentation in CHANGELOG.md

This is how fiduciary-grade software should behave.

Compare to Competitors

vs Excel

• ❌ Excel: Formulas break, no validation, no tests
• ✅ Nagare: 96 tests, validated against benchmarks

vs Basic PE Software

• ❌ Simplified carry calculations
• ❌ Fixed fee percentages
• ❌ Guessed exit timing
• ✅ Nagare: Industry-calibrated, tested, accurate

vs Enterprise Systems

• ✅ They have accuracy
• ❌ But $100K+ annual cost
• ❌ And 6-month implementation
• ✅ Nagare: Accurate AND accessible

Validation You Can Trust

Industry Calibration

Our parameters match:

• Cambridge Associates (exit timing, returns)
• Preqin (vintage year analysis, failure rates)
• Pitchbook (sector multiples, valuations)

Expert Review

October 2025: Financial mathematics expert reviewed our codebase

Findings: 4 issues identified, all fixed Result: "Nagare now implements industry-standard calculations correctly"

Client Validation

$1.2B Family Office:

"We compared Nagare projections to our auditor's model. Perfect match. First time that's happened with software."

$800M Fund-of-Funds:

"Your carry calculations finally match our LPAs. Every other tool gets this wrong."

Getting Started

Your existing fund data works as-is. But if you want to review calculations:

1. Review Carry Calculations

• Navigate to Fund Detail → Parameters → Fee Structure
• See full waterfall breakdown
• Compare to your LPA

2. Update Fee Schedules

• Go to Settings → Fee Basis Schedule
• Configure investment period vs post-investment
• Set step-down rates

3. Validate Against Your Models

• Export to Excel: Fund Detail → Export
• See all formulas in spreadsheet
• Compare to your internal models

4. Questions?

• Email accuracy@nagarehq.com
• We'll walk through calculations
• Show you the test suite

Technical Deep Dive

This post is written for investment professionals. If you want the full mathematical details, we maintain a separate, practitioner-facing specification.

📄 For quants and analysts: Financial Modeling System Documentation

That documentation covers, in white-paper style:

• Formal definition of European and American waterfalls, including IRR hurdles and catch‑up mechanics
• Management fee basis functions over time (committed → invested_cost → NAV)
• Exit timing distributions (Weibull / log‑logistic / competing‑risks engines)
• Return calculation methods for public holdings (price vs total return series)
• Invariants and test cases we enforce in code (e.g., capital conservation, fee monotonicity)

In other words: the exact formulas and edge cases your internal quant or auditor will ask about.

Try It Today

Experience accuracy you can trust:

1. Log in: app.nagarehq.com
2. Create a fund or import existing
3. Review parameters and fee structure
4. Compare to your Excel model
5. See the difference

Free trial: 14 days, no credit card required.

Questions? Email hello@nagarehq.com

---

Related:

• European Waterfall Explained
• Market Data Platform Launch
• Why I Built a Forward-Looking Model