We Built the Fund Screening Tool Family Offices Have Been Doing in Spreadsheets (With 5,000+ Funds Inside)

Here's how most family offices screen VC funds today: a PitchBook export, a 47-tab Excel workbook that someone built in 2022, and a CIO who keeps the real scoring criteria in their head.

The spreadsheet has 200 funds. Maybe 50 have complete data. The vintage-adjusted benchmarks are from last year. Nobody remembers which columns are formulas and which were manually overridden. And when a new fund deck arrives, someone spends 2 hours entering it into the sheet — if they get to it at all.

This is how institutional capital allocates billions to emerging managers. It's absurd.

NUVC is an AI intelligence platform for private markets — pitch deck scoring, investor matching, and fund screening using 8 specialised AI agents and 13 intelligence layers. We built the Fund Library to give family offices the same structured screening process that institutional LPs use, without the headcount.

What Is a Fund Library and How Does AI Fund Screening Work?

NUVC's Fund Library is a structured database of over 5,000 venture fund records — emerging managers, established firms, and everything in between. Each record includes fund size, vintage, strategy, geography, GP team data, track record metrics (IRR, TVPI, DPI where available), and fund terms.

You can search by natural language: "Show me sub-$50M pre-seed funds in Australia with female GPs and ESVCLP structure." The AI search understands fund terminology and maps your query to structured filters — it's not keyword matching.

You can also import your own data. The import system detects column formats automatically — whether you're coming from PitchBook, AngelList, Airtable, or a custom CSV. It maps 30+ common field formats and handles edge cases like Roman numeral fund numbers, SPV boolean fields, and multi-tag status columns.

How Does LP-Grade Fund Scoring Work? (6 Dimensions Explained)

Every fund in the library can be scored across 6 LP-grade dimensions:

1. GP Team Quality — track record, team stability, operational experience, network depth
2. Deal Flow Access — sourcing channels, geographic advantage, sector relationships, proprietary pipeline
3. Track Record — vintage-adjusted IRR/TVPI/DPI blended with age-appropriate weights. A Fund I showing 1.3x TVPI at year 2 is evaluated differently than a Fund III showing 1.3x at year 7.
4. Fund Terms — management fees, carry, hurdle rate, LP-friendliness, key person provisions
5. Fund Status & Timing — deployment pace, fundraising momentum, close timeline
6. LP Network — quality of existing LP base, co-investment track record, institutional backing signals

The scoring engine uses Bayesian priors computed from structured fields — informed starting scores clamped between 3.0 and 7.5, then refined by AI analysis of qualitative data. This means even a fund with minimal data gets a reasonable baseline, not a zero.

Scores are personalised to your investment program. If you've shortlisted 20 funds and passed on 15, the system learns your preferences and nudges dimension weights accordingly. Not enough to override your criteria — maximum adjustment is ±15% per dimension — but enough to surface funds that match your revealed preferences, not just your stated ones.

What Is a Mandate Builder and How Does Monte Carlo Validation Work?

Before you screen funds, you need to define what you're looking for. The Mandate Builder is a 10-step wizard that captures your investment program in structured, computable form:

• Program setup — name, type, currency, team size
• Thesis approach — NUVC-recommended templates, simple presets, or full custom configuration
• Deployment schedule — vintage window, funds per year, cheque size range
• Return targets — target TVPI, investment horizon, co-investment parameters
• Binary filters — hard requirements that eliminate non-qualifying funds (with a live count showing how many funds pass)
• Sector focus — primary sectors with semantic expansion (selecting "climate" also surfaces "cleantech", "carbon", "energy transition")
• GP quality signals — minimum thresholds for unicorn exits, DPI track record, exit count
• Geographic hubs — 12 global investment hubs with quality ratings and concentration limits

Step 8 is where it gets interesting. The system runs a Monte Carlo simulation against your configured mandate — 1,000 iterations of portfolio outcomes — and shows you the probability distribution of your target TVPI.

You see your P10 (pessimistic), P50 (median), and P90 (optimistic) TVPI outcomes, a feasibility score (0-10), the J-curve trough year, and a portfolio variance assessment. If your target of 3.0x TVPI is only achievable in the P90 scenario with your current filters, you'll know before you screen a single fund — not after you've committed capital to 8 managers who can't collectively deliver it.

The validation step also shows warnings: "Your geographic concentration exceeds 60% in one region", "Your deployment window allows only 6 months of sourcing per vintage", "No funds in the library match your binary filters for Fund Size AND ESVCLP." These aren't suggestions. They're structural issues that will limit your program's effectiveness.

What Does an AI-Generated Fund Memo Include?

Select any fund in the library and generate an LP-grade investment committee memo. The memo has 8 sections:

1. Executive Summary — the 3-sentence case for or against
2. GP Team Assessment — experience, stability, track record, gaps
3. Track Record Analysis — vintage-adjusted performance with benchmark context
4. Fund Terms Review — fee structure, carry mechanics, LP protections
5. Thesis Alignment — how the fund maps to your stated mandate
6. Risk Flags — concentration risk, key person dependency, fundraising concerns
7. Recommendation — Invest (score ≥7.5), Watch (≥5.5), or Pass (<5.5)
8. Next Steps — specific follow-up actions based on the recommendation

The deterministic sections (terms, track record numbers, thesis mapping) work without any AI — they're computed from structured data. The qualitative sections (executive summary, GP assessment, recommendation narrative) use AI to synthesise signals into prose. You get a usable memo even if the LLM is down.

How Should Family Offices Measure Portfolio Diversification?

As you build a shortlist, the portfolio fit overlay scores diversification across 5 dimensions:

• Sector diversity (25%) — are you concentrated in one sector or properly diversified?
• Vintage diversity (30%) — are all your commitments in the same year, or spread across the J-curve?
• Geography diversity (20%) — regional concentration risk
• Strategy overlap (15%) — are you backing 3 funds that all do pre-seed AI in SF?
• Risk curve balance (10%) — mix of established vs emerging managers

The scoring is pure math — no LLM, sub-10ms computation. It enforces deployment window constraints: funds closing outside your vintage window score 2.0, regardless of other merits. This prevents the common FO mistake of committing to a great fund that doesn't fit the program timeline.

What Does a Family Office Allocation Dashboard Look Like?

The FO dashboard brings everything together in a 3-column layout:

• Left: Portfolio Timing — where your existing commitments sit on the J-curve, cash flow projections, upcoming capital calls
• Centre: Recommendation Queue — funds requiring decisions, prioritised by urgency (IC review or approaching close date first, stale follow-ups next, new sourced funds last)
• Right: Portfolio Simulation — live Monte Carlo projection based on your current portfolio plus shortlisted candidates

Below: GP tracker strip, insights on drift from your mandate, and mandate health metrics.

Who This Is For

This is built for the family office CIO or principal who:

• Allocates $5M-$50M annually to venture as an LP
• Screens 50-200 funds per year with a team of 1-3 people
• Needs institutional-grade process without institutional-grade headcount
• Wants to explain their fund selection methodology to the family principal or investment committee with data, not gut feel

The LP Fund Screening Framework (Use This Even Without NUVC)

If you're a family office allocating to venture and you don't yet have a structured screening process, here's the framework we recommend — regardless of what tool you use.

1. Define binary filters first. Before you look at a single fund, write down your hard constraints: fund size range, geography, vintage window, minimum fund number, tax structure requirements (ESVCLP in Australia). These are non-negotiable — don't waste time scoring funds that fail binary filters.
2. Score on 6 dimensions, not 1. "Good fund" is not a dimension. Break it down: GP team quality, deal flow access, track record (vintage-adjusted), fund terms, fund status/timing, and LP network quality. Weight them based on your investment program — a first-time allocator should weight track record lower and GP team higher, since emerging managers have less data.
3. Vintage-adjust everything. A Fund I showing 1.8x TVPI at year 3 is outperforming. A Fund III showing 1.8x at year 8 is underperforming. Raw IRR/TVPI/DPI numbers without vintage context are misleading. Use Cambridge Associates or Burgiss benchmarks by vintage and strategy.
4. Run a portfolio simulation before committing. If you're targeting 2.5x TVPI across your venture allocation, model it: How many funds, at what average TVPI, with what vintage spread, produces that outcome? Monte Carlo gives you P10/P50/P90 scenarios. If your target only works in P90, either lower the target or change the fund selection criteria.
5. Track mandate drift. Every quarter, compare your actual shortlist and pass decisions against your stated criteria. If you've been passing on 60% of funds that meet your binary filters because of unstated preferences, your mandate is drifting — update it to match reality, or discipline yourself to follow it.

Frequently Asked Questions

How do family offices screen VC funds as LPs?

Family offices screening VC funds as LPs typically evaluate funds across 6 dimensions: GP team quality, deal flow access, track record (vintage-adjusted IRR/TVPI/DPI), fund terms (fees, carry, hurdle), fund status and timing, and LP network quality. The most sophisticated FOs define an investment mandate with binary filters (fund size, geography, vintage, tax structure) and score funds against that mandate. NUVC's Fund Library automates this process across 5,000+ fund records with AI-powered scoring and natural language search.

What is a Monte Carlo simulation for fund allocation?

A Monte Carlo simulation for fund allocation runs thousands of random iterations of portfolio outcomes based on your configured parameters (number of funds, target TVPI per fund, vintage spread, sector allocation). It produces a probability distribution showing P10 (pessimistic), P50 (median), and P90 (optimistic) outcomes for your total portfolio TVPI. This tells you whether your target return is realistic given your constraints — before you commit capital.

What are the 6 dimensions of LP-grade fund scoring?

LP-grade fund scoring evaluates venture funds across 6 dimensions: (1) GP Team Quality — experience, stability, operational depth; (2) Deal Flow Access — sourcing channels, proprietary pipeline, geographic advantage; (3) Track Record — vintage-adjusted IRR, TVPI, and DPI with age-appropriate weighting; (4) Fund Terms — management fees, carry, hurdle rate, LP protections; (5) Fund Status & Timing — deployment pace, fundraising momentum; (6) LP Network — quality of existing LP base and institutional backing signals. NUVC uses Bayesian priors to produce informed baseline scores even with incomplete data.

What is ESVCLP and why does it matter for Australian family offices?

ESVCLP (Early Stage Venture Capital Limited Partnership) is an Australian tax structure that provides capital gains tax exemptions on investments held for 12+ months. For family offices allocating to Australian venture, ESVCLP-structured funds offer significant tax efficiency. NUVC's Fund Library includes ESVCLP as a searchable filter, so FOs can surface only tax-advantaged funds that match their mandate. See also: The Family Office's Guide to Startup Deal Screening.

If you're still running your fund screening program in Excel, see what the Fund Library can do.