Each check buys a slice of the company at its pre-money valuation:
Ownership = Investment / (Pre-money + Investment)
Post-money = Pre-money + Investment
You only see proxy signals — team score, MRR growth, runway, and TAM — not the startup's true underlying quality. Each signal nudges a hidden quality score that biases (but never guarantees) the outcome roll toward one of four tiers, modeled loosely on real early-stage return data (Correlation Ventures' study of ~21,000 VC-backed rounds found ~65% returned less than the money invested, while under 5% returning 10x+ produced most of the profit):
Tier Multiple on that check Base odds (quality 0 → 1)
Write-off 0x 80% → 25%
Modest exit 1x – 3x 16% → 21%
Strong exit 3x – 10x 3% → 18%
Home run 10x – 50x 1% → 6%
Portfolio return uses MOIC (multiple on invested capital), the standard VC yardstick, with a rough annualized rate assuming a 5-year hold:
MOIC = Σ(payout) / Σ(invested)
Est. IRR ≈ MOIC^(1/5) − 1
Because most checks return nothing, a healthy fund needs enough independent bets that the rare home run — not the average deal — carries the whole portfolio. That's the power law: pick more startups than the outcome of any single one.