Ten lessons connecting fund structure, financing terms, cap tables and investor returns through original cases and explicit assumptions.
College-level conceptual and quantitative introduction. US venture-fund and Delaware-style preferred-equity context; YC US post-money SAFEs only. Not legal, tax or investment advice or a regulatory-compliance course. Historical NVCA-based clauses illustrate mechanisms, not current US enforceability; Canadian adaptations are not US law. All fee, carry and financing numbers are toy terms.
Percentages, algebra, compound growth and basic financial statements; no prior fund-modeling experience.
Course outline
The fund is not the management company
Trace legal entities, commitments and capital calls.
Fees, carry and the waterfall
Model a complete waterfall rather than a fee slogan.
Valuation becomes a price per share
Reconcile pre-money, post-money, price and ownership.
Dilution, pools and cap-table denominators
Solve compounding dilution and a post-round pool target.
Liquidation preferences and exit payoffs
Allocate equity proceeds under complete preferred-stock assumptions.
Control, pro rata and down-round protection
Separate governance, paid participation and conversion-price protection.
SAFEs: cap, discount and conversion order
Use post-money ownership estimates only within their stated regime.
Power-law thinking without a guaranteed power law
Analyze outcome concentration without inventing universal tail statistics.
MOIC, DPI, RVPI and TVPI
Separate gross investment value from net LP flows and marks.
IRR, timing and a complete judgment
Solve money-weighted returns and challenge incomplete rankings.
Sources and curriculum note
Sources fetched October 4, 2026. ILPA 2025 granular definitions and Principles 3.0; current NVCA index plus clearly labeled historical NVCA-based term sheet. Actual legal terms and reporting methods vary.
Read every lesson below. The interactive reader above contains the same explanations, with visual tools and quizzes.
1. The fund is not the management company
Learning goal: Trace legal entities, commitments and capital calls.
A venture fund pools capital to buy securities in portfolio companies. LPs acquire interests in the fund; the fund acquires startup securities. The LP is not automatically a direct shareholder of every startup. In a simplified US limited-partnership structure, a GP controls the partnership under its agreement, while an adviser or management company performs investment-management work. These roles may be related entities, but are not one bank account.
A commitment is a contractual promise, not necessarily cash wired on day one. Paid-in capital records contributions already made. Unfunded commitment records capital still available to call under the documents. A fund can call money for investments, management fees and expenses. The limited partnership agreement, offering documents and side letters define these obligations. A GP commitment is invested capital from the manager side, not a management fee or carried interest.
Fundraising, an investment period, follow-on support, realizations and wind-down can overlap. There is no universal schedule or promised startup exit date. Fund interests and portfolio securities are often illiquid. A later financing mark does not discharge an LP funding obligation. SEC guidance emphasizes understanding private-fund fees, conflicts and withdrawal limitations.
For the simple ledger here, assume no recycling, recallable distributions, borrowing or calls outside commitment. A $12m commitment followed by a $3m call leaves $9m unfunded; another $2m call raises paid-in to $5m and lowers unfunded to $7m. The commitment remains $12m. Real agreements can complicate this relationship, so distinguish this teaching identity from a complete legal capital-account model.
Worked example
An LP commits $10m and funds calls of $2m and $1.5m. Find paid-in and unfunded.
Paid-in = 2 + 1.5 = $3.5m.
Unfunded = 10 - 3.5 = $6.5m.
No recycling or outside-commitment calls are assumed.
Practice problem and solution
An LP commits $18m and funds calls of $3m, $2m and $4m. The GP now calls another $6m. No recycling or outside-commitment calls. What percentage of the commitment remains unfunded after the fourth call? Enter percentage points to two decimals, and explain why the denominator is commitment, not paid-in.
Paid-in after four calls = 3 + 2 + 4 + 6 = 15; unfunded = 18 - 15 = 3; 3/18 = 16.67%. The denominator is commitment, because unfunded capital is the part of the promise not yet called. Dividing by paid-in (3/15 = 20%) would answer a different question.
Mental model: Follow the legal entity and ledger before calculating.
Common trap: Using commitments as the denominator of a paid-in ratio.
2. Fees, carry and the waterfall
Learning goal: Model a complete waterfall rather than a fee slogan.
A management fee pays for services. Carried interest is a contractual share of specified investment profits. A phrase such as "2 and 20" omits the fee base, step-down, offsets, expenses, carry waterfall, preferred return and clawback. Rates alone do not define economics. SEC guidance highlights fee disclosure and conflicts; the actual LPA controls the calculation.
One original fee illustration charges 2% of $100m commitments for five years, then 1.5% of a constant $60m invested-capital base for five years. Fees total $10m + $4.5m = $14.5m before other costs. This is a separate illustration, not the cost budget of the waterfall below. The bases are held constant to isolate arithmetic; no rate is asserted as a universal market standard.
The waterfall toy fund calls $100m, spends $10m on fees and expenses, and invests $90m. It returns all contributed $100m before paying 20% carry on remaining profit. There is no preferred return, catch-up, recycling, GP co-investment or tax withholding. If investment proceeds are $190m, profit above contributions is $90m, carry is $18m and LP distribution is $172m. Charging 20% on all proceeds would charge carry on principal.
A whole-fund waterfall considers aggregate capital-return conditions. Deal-by-deal carry may be paid earlier and later losses can create a clawback obligation under the agreement. ILPA recommends all-contributions-plus-preferred-return-first as best practice, not law. Our no-hurdle example is deliberately simpler. Distinguish carry accrued in NAV from carry actually distributed, and request a worked model of the governing agreement.
Worked example
All-capital-back toy waterfall: $100m contributed, $250m proceeds, 20% carry. What reaches LPs?
Profit = 250 - 100 = $150m.
Carry = 20% × 150 = $30m.
LPs receive 250 - 30 = $220m, or 2.2x paid-in.
Practice problem and solution
A toy fund calls $80m, spends $8m on fees and invests $72m. Investments return $140m. Return all $80m before 25% carry on profit; no other terms. Find LP distribution in millions.
Profit = 140 - 80 = 60. Carry = 0.25 × 60 = 15. LP distribution = 125. Fees are already in contributions; do not subtract them twice.
Mental model: A fee needs a base; carry needs a full waterfall.
Common trap: Charging carry on principal or double-counting fees.
3. Valuation becomes a price per share
Learning goal: Reconcile pre-money, post-money, price and ownership.
A financing valuation is a negotiated equity value used to price securities, not a promise of a future sale price. In a clean primary round, post-money equals pre-money plus new company cash. A $12m pre-money value and $3m check imply $15m post-money and 20% new ownership. A $12m post-money quote with the same check instead implies $9m pre-money and 25%. The label changes the denominator.
Price per share equals pre-money value divided by the agreed pre-financing capitalization used for pricing. Fully diluted requires a definition: common, preferred as converted, outstanding options, reserves and convertibles may receive different treatment. Here assume 6m fully diluted shares, no debt, convertibles or pool change. At $12m pre-money, price is $2. A $3m check buys 1.5m new shares.
Post-round total is 7.5m shares. Existing holders own 6/7.5 = 80%; the new investor owns 1.5/7.5 = 20%. This must agree with 3/15. If share and valuation arithmetic disagree, inspect capitalization before deciding that the terms are inconsistent. Existing holders can keep their share count while losing percentage ownership.
Primary financing issues securities and adds company cash. A secondary purchase transfers an existing holder's shares; it need not add cash or shares to the company. Do not add secondary consideration to the primary round when estimating dilution. Also separate equity value from enterprise value, which incorporates debt-like claims and cash under the selected convention. All examples use equity value. Different security rights mean identical percentages need not imply identical economic payoffs.
Worked example
8m pricing shares, $24m pre-money, $6m primary cash; no other changes. Find price and new ownership.
Price = 24/8 = $3.
New shares = 6/3 = 2m.
Total = 10m; new investor owns 20%. Post-money is $30m.
Practice problem and solution
5m pricing shares; $18m post-money includes $3m primary cash. No other securities or pool change. How many new shares are issued, in millions?
Pre-money = 18 - 3 = 15. Price = 15/5 = $3. New shares = 3/3 = 1m; final new ownership = 1/6.
Mental model: Use one capitalization definition throughout.
Common trap: Treating a post-money quote as pre-money.
4. Dilution, pools and cap-table denominators
Learning goal: Solve compounding dilution and a post-round pool target.
A cap table should label each security, conversion assumption, option reserve and ownership denominator. An outstanding-share percentage excludes unissued reserves; fully diluted may include them. A reserved pool need not issue voting shares immediately, yet it can change a financing price. A 100% reconciliation is necessary but does not show that rights are equal.
If new investors buy 20% post-round, non-participating old holders multiply their percentages by 80%. A founder with 70% moves to 56%. Dilution compounds: retaining 80% in one round and 75% in the next preserves 60% of the original percentage, not a 45-percentage-point deduction. Ownership percentage can fall while the holder keeps the same number of shares.
An option-pool top-up in the pre-money pricing denominator makes old holders absorb the reserve before new capital arrives. Founders own 8m shares with no old pool. A $16m pre/$4m primary round requires an unissued pool of 10% post-round fully diluted shares. Let x be the reserve in millions. Pricing base is 8+x; buyer receives 0.25(8+x); total is 1.25(8+x). Solve x=0.10×1.25(8+x), giving x=8/7m.
Price becomes $1.75, investor shares 16/7m and total 80/7m. Founder owns 70%, reserve 10%, investor 20%. Without a top-up the founder retains 80%. If a 10% pool is instead created after the clean round, both old and new holders are diluted: founder 72%, investor 18%, pool 10%. These are different contractual sequences. Real SAFEs, notes and existing grants require the actual capitalization definitions, not a generic round label.
Worked example
Founder 60%, angels 30%, pool 10%; new investors buy 25%, no pool top-up. Reconcile ownership.
Multiply old stakes by 0.75.
Founder 45%, angels 22.5%, pool 7.5%, new investors 25%.
Sum = 100%.
Practice problem and solution
Founder starts at 75%. One clean round sells 20% post-round and the next sells 30%. No founder purchases or pool changes. Find final founder percentage.
75% × 80% × 70% = 42%. Retention is multiplicative, not subtraction of round percentages.
Mental model: State denominator and ordering before interpreting dilution.
Common trap: Adding dilution percentages or treating reserve as issued votes.
5. Liquidation preferences and exit payoffs
Learning goal: Allocate equity proceeds under complete preferred-stock assumptions.
Ownership is not a complete exit model. Preferred stock may have liquidation preferences and conversion rights that alter payouts on defined events. The charter defines seniority, participation, dividends and conversion. This lesson assumes one preferred class, no debt, no costs or accrued dividends. Equity proceeds are cash available after any senior claims. A 1x preference sets priority, not a guaranteed return.
An investor paid $5m and owns 25% as converted. With 1x non-participating preferred, it chooses the better of available preference and common-equivalent value, not both. With equity proceeds E in millions, payout is max(min(E,5),0.25E). At E=12 it takes 5 rather than 3; common gets 7. At E=40 it converts for 10; common gets 30. At E=3, preferred can get only 3.
Uncapped participating preferred takes its available preference and then shares in residual proceeds. At E=12, the same investor receives 5+25%×7=6.75; common gets 5.25. A participation cap or multiple preferred series changes this model. A holder can also have a conversion choice. Never apply a one-class formula to a senior stacked waterfall without rebuilding the allocation.
Non-participating conversion breaks even at 5/25%=20m. A high financing valuation can coexist with rights that leave common with less at a moderate exit. The historical NVCA-based term sheet published by CVCA supplies alternative preference mechanisms. It illustrates contractual mechanics only, not current US enforceability or the latest NVCA wording. Check current transaction documents and counsel; the original toy numbers here are not contractual advice.
Worked example
$4m paid for 20% with 1x non-participating preferred; equity exit proceeds $15m. Allocate.
Preference 4; conversion 0.20×15=3.
Investor takes 4; common takes 11.
Do not add preference and conversion.
Practice problem and solution
$6m invested for 30% with 1x uncapped participating preferred. Sale leaves $18m equity proceeds after senior claims and costs. No dividends or other classes. What does common receive, in millions?
Residual after preference = 18-6=12. Common gets 70%×12=$8.4m. Preferred gets 6+3.6=$9.6m.
Mental model: Security rights determine exit allocation.
Common trap: Reading 1x as guaranteed cash or combining non-participating paths.
6. Control, pro rata and down-round protection
Learning goal: Separate governance, paid participation and conversion-price protection.
A term sheet combines economics with control. Board composition concerns director selection; protective provisions require specified consent for defined actions. Information rights concern reporting; transfer restrictions concern sales. Minority ownership does not necessarily mean negligible control, and a board seat does not equal majority ownership. Read the holder, threshold and exceptions for each right.
Pro rata is generally an opportunity to buy more securities, not free immunity from dilution. An investor owning 20% needs $2m of a total $10m clean financing to retain 20%, assuming participation is available and no other changes. If outsiders instead contribute a fixed $10m excluding the insider, solve x/(10+x)=20%, giving x=$2.5m. The total becomes $12.5m. The denominator changes because the insider check is additional.
Price-based anti-dilution adjusts preferred conversion terms for defined lower-price issuances. It is not a promise of unchanged ownership. In a toy full-ratchet clause, conversion price resets to the lower issue price, subject to exceptions. A weighted-average model uses CP2=CP1×(A+B)/(A+C), where A is the defined pre-issue common-equivalent base, B is new cash divided by old CP, and C is actual new shares. The instrument defines the precise base and exclusions.
With CP1=$2, A=10m, and $4m raised at $1, B=2m and C=4m. Weighted-average CP2 is 12/7=$1.7143; full ratchet gives $1. A 1m preferred holding initially convertible 1:1 becomes 7/6m common-equivalent shares under weighted average or 2m under full ratchet. Rebuild total capitalization to find ownership. Do not assume every term-sheet provision is nonbinding: confidentiality, exclusivity or expense terms may create obligations, and effect depends on wording and law. Historical NVCA-based clauses ground the illustration, not a legal opinion.
Worked example
CP1=$3, A=12m, $6m raised at $2. Calculate weighted-average CP2.
B=6/3=2m; C=6/2=3m.
CP2=3×(12+2)/(12+3)=$2.80.
Rebuild capitalization to calculate final ownership.
Practice problem and solution
An investor owns 15%. Outside investors will contribute exactly $17m, excluding insider cash. What insider check preserves 15%, in millions? No other capitalization changes.
x/(17+x)=0.15; 0.85x=2.55; x=3. Total is $20m. A $2.55m check would be insufficient because the total grows.
Mental model: Different rights solve different risks.
Common trap: Calling anti-dilution a guarantee of unchanged ownership.
7. SAFEs: cap, discount and conversion order
Learning goal: Use post-money ownership estimates only within their stated regime.
A SAFE gives contractual rights to future equity or specified payments on defined events. YC's US post-money SAFE is not debt and has no interest or maturity date; it is not already stock or riskless. A convertible note is debt with its own interest and maturity terms. Identify form, version, jurisdiction and amendments before modeling conversion. This course uses YC US post-money forms, not international forms or every security called a SAFE.
In the cap-driven regime, investment divided by post-money cap estimates ownership after SAFE financing but before later priced-round new money. $1m at a $10m post-money cap gives 10%. YC's guide warns that a priced valuation below or too close to the cap can create more than the estimate. The cap is a conversion term, not a company appraisal or maximum future value. Priced-round pool increases and actual capitalization definitions require separate treatment.
Two cap-driven post-money SAFEs of $0.6m/$6m and $0.8m/$10m sell an estimated 10%+8%=18% before the priced round. A clean round selling 25%, without a pool increase or pro rata, reduces them to 7.5% and 6%; founders retain 61.5%. Post-money SAFE is post-SAFE money, not post-every-future-financing. Old pre-money SAFEs do not use the same simple arithmetic.
A discount-only example with a 20% discount on a $4 share price converts at $3.20; $0.64m buys 0.2m shares. Ownership still requires final shares. A form may call the retained-price percentage an 80% Discount Rate rather than a 20% discount; read the definition. YC offers cap-only, discount-only and uncapped MFN alternatives. Do not combine cap and discount without actual terms. Post-money pro rata rights use an optional side letter rather than appearing automatically in the standard SAFE.
Worked example
$0.9m cap-driven post-money SAFE at $9m cap, then 30% clean new-money round. No pool or pro rata. Estimate final SAFE ownership.
Before round: 0.9/9=10%.
After round: 10%×70%=7%.
Check actual cap regime before applying this estimate.
Practice problem and solution
Cap-driven SAFEs: $0.4m at $8m post cap and $0.9m at $10m post cap. Later round sells 20%, no pool top-up or pro rata. Find combined SAFE percentage after the round.
Before round: 5%+9%=14%. After round: 14%×80%=11.2%. Founder estimate is 68.8%, new money 20%. Actual conversion must satisfy the stated regime.
Mental model: Post-money SAFEs can still suffer later dilution.
Common trap: Treating a cap as guaranteed value or mixing SAFE versions.
8. Power-law thinking without a guaranteed power law
Learning goal: Analyze outcome concentration without inventing universal tail statistics.
VC performance can be highly skewed: a small number of winners may dominate portfolio proceeds. Kaplan and Lerner emphasize that omissions matter in highly skewed VC data. Cochrane analyzes volatile outcomes and selection bias when successful firms are more likely to produce observable exit values. These support caution about averages and sample construction, not a claim that every fund follows one universal mathematical power law.
A power-law tail is a specific statistical model, roughly P(X>x) proportional to x raised to a negative exponent over a specified range. A fitted exponent requires data, a threshold and a defensible sampling model. Ten invented companies cannot identify it. Power-law thinking here is a heuristic: large outliers may dominate dollars and removing one winner can change a portfolio radically. It does not prove that large markets imply high success probabilities or that a particular fund has infinite expected returns.
Ten equal $1m investments: six return zero, three return $2m each and one returns $30m. Proceeds are $36m on $10m invested, or 3.6x gross MOIC. Median company multiple is zero. Winner share of proceeds is 30/36=83.3333%. Replace the winner with a $3m outcome and the portfolio returns $9m, or 0.9x. These payoff counts are original constructed cases, not measured failure rates.
Fund size and ownership matter. A 10% stake in a $1bn equity exit yields $100m before preferences, dilution and costs. It is 10x on a $10m check, but only 0.2x of a $500m fund commitment. Reserve allocations trade further exposure to a company against other checks and expenses. One famous exit does not establish a forecast for the next portfolio. Model outcome scenarios, dilution and financing needs together rather than confusing exit headlines with fund returns.
Worked example
Eight $2m checks: five fail, two return $4m each, one returns $40m. Compute gross MOIC and concentration.
Invested=8×2=$16m; proceeds=8+40=$48m.
Gross MOIC=48/16=3x.
Winner share=40/48=83.3333%; no tail exponent is estimated.
Practice problem and solution
Twelve $1m checks: eight return zero, three return $2m each, one would return $30m. The winner stake is diluted by 20% before an otherwise unchanged exit. Find revised gross portfolio MOIC.
Winner=30×80%=24. Other proceeds=6. Total=30 on invested 12, so 2.5x. Original was 3x. Dilution reduces the aggregate by 0.5x.
Mental model: Count dollar contribution and ownership, not just winners.
Common trap: Presenting invented skew as proof of an empirical power law.
9. MOIC, DPI, RVPI and TVPI
Learning goal: Separate gross investment value from net LP flows and marks.
MOIC relates realized proceeds plus residual investment value to invested capital under a specified methodology. A fully exited $5m investment returning $15m has 3x gross MOIC, without saying how long it took. Gross portfolio MOIC is not automatically a net LP multiple: fees, expenses, carry, idle cash and cash-flow boundaries differ. ILPA's 2025 granular definitions distinguish investment and fund-level calculations, and gross from net.
At the LP/fund boundary, define P as cumulative contributions, D as distributions and R as LP-attributable residual NAV, all on the same net basis and date. In this course's cash-only, no-recycling examples, DPI=D/P, RVPI=R/P and TVPI=(D+R)/P=DPI+RVPI. Contributions, not commitments, form the denominator. Do not combine gross portfolio NAV with net LP distributions without adjustments.
P=50, D=30, R=45 gives DPI 0.6x, RVPI 0.9x and TVPI 1.5x. P=50, D=65, R=10 also gives TVPI 1.5x but DPI 1.3x and RVPI 0.2x. Equal total value does not mean equal liquidity or valuation uncertainty. NAV is a mark on residual assets, not itself a distribution. Real funds may distribute securities rather than cash, so even DPI is not automatically cash realized into an LP bank account.
A markdown from R=45 to R=25, without flows, reduces TVPI to 1.1x and RVPI to 0.5x but leaves DPI at 0.6x. Recallable distributions, reinvested capital and subscription facilities need consistent specified treatment. ILPA has multiple reporting methods, so MOIC does not always have precisely the same denominator as net TVPI. Report net/gross status, ledger boundary, valuation date and methodology with the multiple. Academic work warns that missing data and valuation subjectivity can distort fund comparisons.
Mental model: DPI records distributions; TVPI includes residual marks.
Common trap: Confusing gross portfolio MOIC with net LP TVPI.
10. IRR, timing and a complete judgment
Learning goal: Solve money-weighted returns and challenge incomplete rankings.
IRR is the discount rate making the present value of all cash flows zero. LP contributions are negative, distributions positive. For ongoing funds, terminal net NAV can enter as a positive measurement-date value. This is a valuation convention, not cash received. ILPA defines net IRR using the applicable LP flows and residual value net of fees, expenses and performance compensation.
For -100 at time zero and +200 at year T, solve -100+200/(1+r)^T=0, giving r=2^(1/T)-1. A 2x multiple in three years implies 25.9921% annual IRR; in eight years it implies 9.0508%. A high short-period IRR can still produce fewer dollars than a lower-IRR larger multiple. IRR is not profit divided by years, and identical multiples need not have identical IRRs.
For multiple irregular dated flows, solve the dated present-value equation with an explicit day-count convention, such as an XIRR-style annual exponent. Here all intervals are exact integer years. Conventional outflow-then-inflow cases are tractable; streams with later negative flows may have multiple IRRs or no useful unique solution. Do not average company IRRs to get a fund IRR. Aggregate actual dated flows and solve again.
A subscription line can delay LP contributions after investments already exist and raise with-facility IRR without an equivalent improvement in investment outcomes. Borrowing also costs money. ILPA calls for consistent with/without-facility reporting. Final judgment should inspect net multiples, DPI, residual valuation, timing, fees, risk, vintage and mandate. Public-market comparison needs a defined benchmark and method; subtracting a stock-index return from IRR is not a complete risk-adjusted analysis. Academic findings depend on samples and periods, not a universal current VC return forecast.
Worked example
Calls $100m at 0; distributions $60m at year 1 and $60m at year 2; no residual. Solve IRR.
-100+60/(1+r)+60/(1+r)^2=0.
Let y=1+r: 100y²-60y-60=0.
Positive y=(60+√27600)/200=1.130662386.
IRR=13.0662%; final DPI and TVPI are 1.2x.
Practice problem and solution
A liquidated toy fund calls $50m at year 0 and distributes $150m at year 4; no other flows. Find annual IRR as percentage points rounded to two decimals, and explain how it differs from DPI.
DPI=TVPI=150/50=3x. IRR=3^(1/4)-1=31.6074%, rounded 31.61%. DPI records distributed value relative to paid-in; IRR incorporates four-year timing.
Mental model: Compare cash realization, marks, timing and net economics together.
Common trap: Ranking funds by one IRR without inspecting flows.