Model notes & sources
Seven structures, one business. The business is three numbers — revenue, growth, free-cash-flow margin —
a volatility (the spread of growth rates across a small Monte Carlo), and an estimated, industry-specific
survival curve calibrated from BLS Business Employment Dynamics
establishment survival by sector: ~50% of new establishments survive five years overall, with retail /
e-commerce among the weakest (~58% gone by year 5) and skilled trades / home-services and established
recurring SaaS the most durable. Restaurants run only a little below the baseline on survival (the
"90% fail" figure is a debunked myth; the real number is ~50–55% over five years) — their financing risk is
the thin 5–9% margin, not closure. How well you pick scales that baseline failure rate up or down
(the midpoint is the industry baseline; better screening means fewer failures). Every simulated business
draws a growth rate and a failure time from that curve, then the chosen instrument's payment rules run
against the trajectory until the cap is repaid, the business dies, or the horizon ends. Deals in the
book repeats this for N independent financings of the same deal (the cheque split across them) and solves
one book-level IRR — diversification collapses the spread (a single deal's coin flip becomes predictable),
and for the power-law equity instrument it flips a negative single-stake median into a solid book return.
The investor's result is a distribution, shown with its median, mean and P10–P90. The financing feeds back on survival: pledging a share d of free cash flow to repayment
thins the buffer against shocks, so the failure hazard is scaled by (1−d)−0.6 — a
debt-service-coverage view in which heavier terms fail more businesses (a thin-margin restaurant loses far
more survival than a fat-margin SaaS), while equity, with no cash-flow drag, leaves survival untouched. The
exponent is an illustrative sensitivity, not a fitted constant. Straight-equity stakes draw both their
multiple and their holding period from the Outcomes distribution (losses ripen in ~3 years, the biggest
winners ~6). The curves (and the revenue / growth / margin figures) are illustrative, industry-typical estimates
for already-operating businesses, not a forecast for any specific company. The owner's cost of capital is
the deterministic rate a surviving business pays, quoted as a nominal APR so it sits honestly beside a bank
loan. All seven run on the same businesses each pass, so the selector is an apples-to-apples comparison.
The instruments: a secured loan (the recourse benchmark — the only one that is a loan);
capped revenue share (RBF); a revenue royalty (a fixed-term share with no cap on the return);
a profit share paid only from cash flow;
redeemable preferred (a dividend plus a buy-back); a shared-earnings agreement (SEAL — a slice of
founder earnings to a cap, then a little equity); and straight equity (no payments, return only on a
sale). The legal hinge that keeps the other six out of lending law is contingency: tie repayment to revenue,
profit or equity rather than a fixed sum on a fixed date and usury caps and lending licences don't apply.
Rates are stylised — a real SBA/bank term loan runs ~11–15% APR, online revenue lenders ~20–40%, and merchant
cash advances frequently north of 50% effective. A teaching model — not legal, tax or investment advice.