How every Preflop token gets a fair price

Existing venues for human-capital claims fail at one of the three requirements for liquidity. ISAs and athlete-revenue advances (Big League Advance, Purdue) price at a single handshake moment and never reprice. Private cap-table secondaries have no fundamental anchor — they trade on last-round last-comparable logic that falls apart outside venture-backed corridors. Public equities have all three, built over a century on top of a boring denominator: quarterly GAAP earnings.

Preflop’s denominator is Total Economic Benefit. Everything downstream — the anchor, the clearing price, the repricing loop — is mechanical application of standard corporate-finance mathematics with TEB substituted for earnings.

Why it matters —The reference isn’t the price. The price is what the market clears. The reference is the fundamental the market trades around — transparent, derivable, reproducible from public inputs, and revisable on new information.

Stage 1 produces a forward Total Economic Benefit path over a 75-year horizon, with low, mid, and high bands. The path is piecewise exponential: a blended near-term CAGR, then a terminal growth rate beyond year 10.

$g_1$ is the blended near-term CAGR — a function of industry trajectory, age-adjusted runway, trailing TEB CAGR (if available), and pipeline indicators. $g_2$ is the terminal growth rate, pinned at 2% in v1 and planned to become sector-conditional in v2. $\text{TEB}_0$ is a blended mid-band current-year TEB — when real income reporting is unavailable, the forecast falls back on income-bracket priors conditioned on stage, domain, and conviction.

The confidence interval

Every forecast carries three curves. The low and high bands are derived from signal-conditioned dispersion around mean outcomes — issuers with tighter signals (Direct Listings with 5+ years of tax-verified trajectory) get narrow bands; idea-stage founders get wide bands. Formally:

$\sigma(\cdot)$ grows with $t$ and shrinks with conviction. $z_{\alpha/2}$ is the confidence-interval multiplier ($z_{0.025} \approx 1.96$ for the 95% CI band shown on every listing).

Why it matters —The forecast isn’t one number — it’s a band. Pricing uses the mid and carries the CI through to the reference price, so a sharp LP can distinguish high-conviction narrow-band issuers from noisy wide-band issuers at a glance.

The forecast feeds a Gordon-growth present-value calculation. At listing, the reference price is the per-token PV of the issuer’s committed share of future TEB.

$N = 10{,}000$ is the fixed token supply per class. $e$ is the perpetual share-of-TEB committed (e.g. 2% for Direct Listings, varies for covenants). $r$ is the risk-adjusted discount rate. $g$ is the baseline Gordon near-term growth rate (pinned at 5% for the reference calculation).

Gordon collapses the infinite-horizon present-value integral to a single ratio whenever cash flows grow at a constant rate below the discount rate. That requires $r > g$; with$r = 12\%$ baseline and $g = 5\%$ baseline the spread is $700\,\text{bp}$ — comfortable. The derivation, for reference:

The risk-adjusted discount rate · math-foundation Part 3.6

$r$ is not an opaque constant. The math foundation Part 3.6 decomposed it into four named premia, each with a source:

Note —The four-term decomposition below is the math-foundation framing and a pedagogical reference. The live engine v2.1.0 currently collapses it to $r = 12\%$ flat. Discount Rate v1 (/spec/discount-rate Design v1 — In progress) replaces both with three components $r_c = r_f + \beta_c \cdot \text{ERP} + \pi_{\text{illiq}}$, ships in engine v2.2.0.

Top-conviction Direct Listings price around $r = 9{-}11\%$; idea-stage covenants can price at $r = 18{-}22\%$. Each issuer’s rate is displayed on their profile with the breakdown.

Lockup ends. The order book opens. Market makers post two-sided quotes seeded from $P_{\text{ref}}$. Trading rules:

• —Circuit breakers halt trading automatically on moves of more than $\pm 20\%$ from $P_{\text{ref}}$. Halts last 10 minutes; conviction-signal review clears the halt or extends it.
• —Opening cross at T+0 aggregates pre-market orders into a single cross price; the book opens continuously after.
• —Market-maker obligations — the LP that anchors each listing commits to two-sided quotes within the circuit band for the first 30 trading days. Withdrawal triggers automatic re-anchoring to the next available LP.

Why it matters —Every venue for human-capital claims today either anchors on a single handshake-negotiated price and never reprices, or has no anchor at all and no repricing mechanism. A pre-trade anchor plus an open order book is how equities, bonds, and commodities have cleared for a century. Nothing on Preflop is a novel market design — what’s novel is applying the old design to human capital, which has never had a denominator rigorous enough to price off.

Every quarterly TEB report, every material event (single-source income ≥ $100K, notifiable within 7 days), and every conviction-signal update reshapes the forecast. The forecast re-anchors $P_{\text{ref}}$. The market clears the delta.

Between events · GBM drift

Between repricings, secondary prices follow a standard geometric Brownian motion with drift set by conviction and volatility set by phase:

$W_t$ is a Wiener process (standard Brownian motion). $\mu$ is the drift — positive for high-conviction issuers with rising signals, negative during deficiency windows. $\sigma$ is volatility — Phase 2 tokens (post-IPO Person Token) trade at a fraction of the volatility of Phase 1 tokens because the perpetuity discount rate dominates the Phase 1 dividend PV as the conversion date approaches.

Why it matters —Tokens never approach zero. The Phase 2 perpetuity anchors the floor. Even a failed founder retains some TEB (wages at a new job), and the perpetual claim on that residual TEB gives the token a conviction-weighted residual value. Failure modes are bounded; the upside is not.

Information events · what reprices and how

Trace one listing from the Stage 1 forecast through a Year 3 earnings surprise. Issuer is a Direct Listing ($e = 2\%$, no covenant), $\text{TEB}_0 = \$500{,}000$, $r = 12\%$, $g = 5\%$.

Why it matters —Nothing in that sequence is novel. It’s the standard earnings- surprise playbook from public equity markets, substituting TEB for GAAP earnings and a Gordon anchor for a DCF model. The contribution isn’t the mechanism; it’s the denominator.

The token’s fair value is the sum of a bounded dividend stream and a perpetual claim. Both terms see the same TEB forecast, but they respond to forecast perturbations very differently. This section derives the full decomposition and the partial derivatives that explain why the same instrument behaves like venture on a pre-revenue issuer and like a dividend stream on an established one.

Full decomposition

With the piecewise-exponential forecast from Stage 1 — $\text{TEB}(\tau) = \text{TEB}_0\,(1+g_1)^{\min(\tau,10)}\,(1+g_2)^{\max(\tau-10,\,0)}$ — and the covenant ending at $T \le 10$, the Phase 1 integral is a finite-horizon annuity:

The $g_1 = r$ boundary case collapses to the flat annuity $PV_1 = s_1 \cdot \text{TEB}(t) \cdot (T-t) / N$ (a limiting case of the same integral, used only when the forecast is tuned exactly to the discount rate — rare in practice).

Phase 2 is the infinite-horizon tail. With terminal growth $g_2 < r$ this collapses to the Gordon form:

The $r > g_2$ guardrail in the pricing library ($r - g_2 \ge 150\,\text{bp}$) ensures the perpetuity term is numerically stable. The $g_2 \ge r$case is a model failure — the forecast is implying the issuer’s TEB grows faster than the discount rate forever, which isn’t defensible for a human. The library rejects those parameters.

Sensitivity analysis

Which parameters move the value, and by how much? Compute the partial derivatives. For a Maya-style forecast ($s_1 = 3\%$, $s_2 = 1\%$, $T = 10$, $r = 12\%$, $g_2 = 2\%$, $N = 10{,}000$):

Why it matters —The punchline: $\partial V / \partial g_{\text{terminal}}$is overwhelmingly a Phase 2 effect. Every dollar of PV sensitivity to terminal growth lives in the perpetuity term. This is why a re-forecast of Maya’s long-term trajectory (Series A closes; the market re-prices $g_{\text{terminal}}$ from 3% to 4%) substantially lifts $PV_2$ while leaving $PV_1$ more constrained in its response.

Maya numbers · listing → year-4 breakout

Engine cross-reference (v2.1.0) —The Forecast Engine v2.1.0 — running Maya’s exact inputs through cohort priors + bounded adjustments — produces $V_{\text{HC}}^{\text{mid}} \approx \$860\text{K}$, $e_{\text{eff}} \approx 1.78\%$, per-token mid $\$1.53$, and $\kappa \approx 3.92$ at a $\$60\text{K}$ target raise → speculative tier. The engine is more conservative than the math-foundation walkthrough below because the cohort prior says the median pre-seed founder doesn’t reach $\text{TEB}(10) = \$600\text{K}$. Engine output is the operational ground truth; the math-foundation numbers below illustrate the pricing mechanics on the canonical forecast and reconcile to the engine in a future vault session. See forecast-engine §Maya end-to-end.

Reconciled values from the canonical math-foundation piecewise forecast (TEB(0)→ TEB(2)→TEB(5)→TEB(10) = $20K → $60K → $200K → $600K, terminal $g = 3\%$), verified against lib/obligation-ledger. All numbers below are derived, not narrated.

Of the $\$28.45/\text{tok}$ dollar reprice from listing to year 4, 57% came from Phase 2 ($\$16.24$ of $\$28.45$), 43% from Phase 1. The Phase 2 share leverages the terminal-growth re-rating from 3% to 4%; the Phase 1 share leverages the near-term TEB lift from a $20K → $80K speculative path to a $400K → $3M actualized one. Both phases reprice; Phase 2 reprices harder per dollar of forecast change because of the Gordon denominator compression.

Why it matters —The 57/43 split is a feature, not a quirk — the same instrument on Dr. Amara Okafor (established issuer, narrow CI band, modest terminal-growth re-rate scope) would split closer to 20/80 in the opposite direction (Phase 1 dividend stream dominates the realized return). Same math, two coherent regimes, depending on where the forecast actually has room to move.

How this composes with the other primitives

The decomposition is one leg of the three-primitive stack:

• —TEB forecast (Stage 1) provides the time-path $\mathbb{E}[\text{TEB}(\tau)]$ that both integrals consume. Forecast re-rating is what drives $\partial V / \partial g_{\text{terminal}}$ in practice.
• —Cap ledger (cross-listing math spec) enforces $\sum_i s_{\text{rate},i} \le 25\%$so no overlapping obligations make the perpetuity term ill-defined. Phase 2’s numerator is well-bounded only because the ledger keeps the claim stack coherent.
• —Conviction engine (v2 spec) outputs the forward signals that determine $g_1$, the band around $g_{\text{terminal}}$, and the drift $\mu$ between repricings. Every conviction update propagates into the decomposition.

Together, these three primitives compose into a coherent instrument: a pricing decomposition that explains where returns come from, a ledger invariant that keeps the stack solvent, and a conviction engine that refines the forecast feeding both.

Pricing produces a number. The number means a fraction of human capital, and the market arrives at it through an auction. The two specs below close the loop:

The five-stage data flow

Convergence and the 150bp guardrail

Every quantity above relies on the V_HC integral converging. The condition is $\limsup_{\tau \to \infty} g(\tau) < r$. For the piecewise forecast this reduces to $g_{\text{terminal}} < r$. The production library enforces a 150 basis-point guardrail so the perpetuity is numerically stable:

Any forecast that violates this guardrail raises a DivergentPricingError before the pricing engine emits a reference price. The error is structural — a forecast claiming a person’s TEB grows at or above the discount rate indefinitely is asserting infinite human-capital wealth, which is not a defensible claim. The library rejects rather than approximates.

Why it matters —Why this matters at the listing layer: κ-tier eligibility (see human-capital spec) is checked against the same $V_{\text{HC}}^{\text{mid}}$ that flows from this pipeline. Convergence has to hold before eligibility can be evaluated. A listing with a divergent forecast can’t be checked, can’t produce a reserve, can’t run an auction. The pipeline either runs end-to-end or doesn’t run at all.

Negative TEB periods

A loss year (e.g. founder between companies with zero wage income) produces a negative near-term drift contribution but does not make$\widehat{\text{TEB}}(t)$ itself negative — the forecast uses the blended path, not the spot. A single zero year widens the CI but does not drop the mid.

Discount rate approaching growth

If $r - g$ drops below $100\,\text{bp}$ the Gordon formula is numerically fragile. In practice this never triggers — the lowest baseline $r = 9\%$ combined with the highest possible $g = 7\%$ leaves $r - g = 200\,\text{bp}$. A guardrail in the pricing library rejects any parameter combination with $r - g < 150\,\text{bp}$ and falls back to the pricing-integral form:

Numerical integration over a 75-year bounded horizon, discrete monthly steps. Stable under all sane parameter combinations.

Issuer declared non-reporting

A persistent deficiency (missed two or more quarterly reports) moves the issuer into the deficiency-grace window. $\pi_{\text{liquidity}}$ widens by +300 bp; the spot adjusts. Continued non-reporting triggers the delisting framework.

Today, several parameters are seeded by economic judgment. The calibration harness logs every forecast vs. the eventual realized TEB outcome, so as data accrues we can move each seeded parameter onto an empirical footing.

Vision —In 18 months, Preflop owns the only dataset in the world that correlates early-career signals with realized human economic output. That dataset is the moat — not the app, not the UI, not the order book. The dataset recalibrates every parameter on this page every quarter, tightening every CI, and every listing becomes a training point for every subsequent listing. The instrument is the same; the instrument gets more accurate forever.

See also: the Conviction Engine v2 spec — which details the 11 signals that feed $\mu$ and the CI width, and the Cross-Listing Math spec — which proves multi-listing pricing composability.