How Preflop anchors cohort priors against survivorship bias

Cohort definition. US person aged 22-35 who self-identifies as a founder of a pre-seed-stage B2B SaaS startup, regardless of funding status. Includes the unfunded majority. Funded subset is captured via Adjustment #13 (Notable Prior Achievement) per the existing 15-adjustment framework.

Matching data sources:

• —Primary. US Census ABS Nonemployer Statistics (NES-D) and ABS owner characteristics (Table AB2100CSA01), 2022.
• —Secondary. Kauffman Indicators of Early-Stage Entrepreneurship 2023.
• —Tail-shape only. Crunchbase + YC outcome data — informs upper-tail heaviness and Adjustment #13 calibration. Not the median anchor.
• —Triangulation. Pierre Azoulay et al. (2020), “Age and High-Growth Entrepreneurship,” AER:I — median age of top-0.1% startup founders is ~45, not 22. Implication: high-TEB outcomes concentrated in older founders.

Source bias direction and magnitude:

• —ABS NES-D: under-counts pre-revenue ventures with no EIN; over-states median TEB by including only registered. Bias: anchor too high. Magnitude: ~10-20%.
• —Kauffman: CPS supplement self-report; over-counts gig-work-as-entrepreneurship. Bias: deflates median TEB. Magnitude: ~10%.
• —Crunchbase/YC: catastrophic survivorship/selection bias if used as median. Magnitude: 5-20× too high.

Anchor TEB at listing age 22 (entry-pool):

Reduction at p50: ~37%. At p90: 56%. This is the dominant correction in the spec.

Age curve (conditional on still-in-cohort, i.e., still founding):

Survival curve $P_{\text{stay}}(t)$ — P(still self-identifies as founder at age 22+t):

Exit-cohort fallback: map to the _other-fallback baseline (BLS economy-wide median for college-educated). Exited founders mostly take corporate jobs / consulting / join larger startups as employees.

Growth segments: kept from current founder-pre-seed-bb-saas.ts because they describe conditional-on-still-founding trajectory. The naive integration would over-weight these segments because it assumes everyone stays; survival mixture corrects this.

Terminal growth (carry forward, in-cohort): $\{p_{10}: -0.01,\, p_{50}: 0.025,\, p_{90}: 0.05\}$ per Discount-Rate-v1.

Residual error bound:±25% on anchor at age 22 (entry-pool median bounded by 3 corroborating sources); ±40% on age-curve at 35-45 (Azoulay’s findings imply funded-tail dominates in ways entry-pool data poorly captures); survival curve ±15% per decade.

Data pull date: 2026-04-26.

Cohort definition. US board-certified surgical specialist (orthopedic, general, cardiothoracic, neurosurgical, vascular) actively in private practice, age 35-65 at listing.

Already heavily survival-conditioned at the cohort definition (med school → residency → fellowship → board-cert → years of clinical hours → established private practice). The primary fix here is source diversification and reconciling the current $1.8M anchor with what its cited sources actually report.

Matching data sources:

• —Primary. Medscape Physician Compensation Report 2024. Orthopedic mean $558K, general $412K, blended cross-specialty mean ~$525K.
• —Primary. MGMA DataDive Provider Compensation 2024. Orthopedic median $668K (private practice), general $476K. Skews higher (samples group-practice administrators).
• —Secondary. AAMC Physician Specialty Data Report 2022.
• —Floor anchor. BLS OES 29-1242 (orthopedic surgeons). Top-codes at $239K — useful as floor reference only.
• —Tail. Surgeon-owned ASC ownership data — top decile in ortho/spine includes ASC equity income, pushing p90 to ~$1.4M.

Source bias direction and magnitude:

• —Medscape: voluntary survey, response bias toward complaint. Direction: probably under-states top decile by 5-10%. Median is accurate.
• —MGMA: samples group-practice administrators, skews larger groups. Direction: over-states median by 10-15%.
• —BLS OES: top-coded; under-estimates everything. Floor only.
• —AAMC: census-quality for headcount; weak on compensation.

Anchor TEB at listing age 45:

Reduction at p50: ~68%. This is the largest single correction in the spec. The current $1.8M anchor was unjustified — none of its cited sources support it. Likely originated from a misread of MGMA (which reports $668K for orthopedic specifically, not blended cross-specialty).

Age curve (conditional on still-in-private-practice):

Survival curve $P_{\text{stay}}(t)$ — P(still in active surgical private practice at age 45+t):

Exit-cohort fallback: non-clinical physician income (~$320K median per AAMC for industry/admin/consulting roles). Implemented as a cohort parameter, not a fallback to BLS-economy-wide median — surgeons who leave clinical practice still earn meaningfully more than the population median because of credentialing.

Growth segments (kept; describe conditional-on-still-practicing):

• —Years 0-5: $g \, p_{50} = 0.03$ — practice building
• —Years 5-12: $g \, p_{50} = 0.025$ — peak earning
• —Years 12-20: $g \, p_{50} = 0.01$ — plateau / taper start
• —Years 20-30: $g \, p_{50} = -0.05$ — part-time / wind-down

Terminal growth (carry forward): $\{p_{10}: 0.005,\, p_{50}: 0.02,\, p_{90}: 0.035\}$.

Residual error bound: ±20% on anchor (Medscape and MGMA agree within range); ±25% on age-curve at 56+ (less data on private-practice late-career); survival curve ±5% per decade.

Data pull date: 2026-04-26.

Cohort definition. US lawyer who is an equity or non-equity partner at a US law firm. Current code defines as “AmLaw 100/200 equity partner” — narrow. New spec broadens to all US partner-level lawyers (sub-AmLaw-200 firms, non-equity income partners). Justification: the AmLaw-only definition excludes ~60% of US partners per ABA, and the listing application can’t reliably verify firm tier.

Matching data sources:

• —Primary. NALP Associate-to-Partner Compensation Survey 2024. Median equity-partner comp at AmLaw 100 = $1.28M; non-equity median = $485K; cross-firm-size blended median = $520K.
• —Secondary. AmLaw 200 Profitability Report (ALM Media). Profits Per Equity Partner (PPEP) by firm — for upper-tail shape only, NOT median anchor.
• —Triangulation. ABA Lawyer Demographics 2023 — ~40% of US partners are at firms <20 attorneys; this subset is invisible in NALP/AmLaw data.

Source bias direction and magnitude:

• —NALP: over-samples larger firms with HR infrastructure. Bias: over-states partner median by 20-30% relative to true national partner population.
• —AmLaw: only top-200; PPEP excludes non-equity. Massively over-states median if used as anchor.

Anchor TEB at listing age 38 (US partner across firm-size distribution):

Note: 38 is early for equity partner. A 38-yo “partner” is more likely non-equity / income partner, or equity at a mid-market firm. NALP equity-partner-median anchored to age ~45-50.

Reduction at p50: ~76% if cohort broadens. If the cohort definition stays AmLaw-100/200 only, anchor at p50 should be $1.28M (NALP-cited) not $2.20M — a ~42% reduction. Recommend: broaden the cohort definitionbecause the listing flow can’t verify firm tier reliably and the broader cohort is more defensible.

Age curve (conditional on still-being-partner, broadened cohort):

Survival curve $P_{\text{stay}}(t)$ — P(still a US partner at age 38+t):

Exit-cohort fallback: split between (a) in-house GC / corporate counsel (~$320K median per ABA), (b) government / public-sector law (~$180K), (c) boutique / solo practice (~$280K). Weighted average ~$280K as exit fallback.

Growth segments (kept; conditional-on-still-partnering):

• —Years 0-5: $g \, p_{50} = 0.04$
• —Years 5-12: $g \, p_{50} = 0.025$
• —Years 12-20: $g \, p_{50} = 0.005$
• —Years 20-30: $g \, p_{50} = -0.06$ (taper)

Terminal growth (carry forward): $\{p_{10}: 0.005,\, p_{50}: 0.025,\, p_{90}: 0.04\}$.

Residual error bound: ±25% on anchor (firm-size mix uncertainty dominates); ±20% on age-curve mid-career; survival curve ±10% per decade.

Data pull date: 2026-04-26.

Cohort definition. US active major-league athlete (NFL, NBA, MLB, NHL) with ≥4 years of league service, age 25-35 at listing.

Already survival-conditioned (drafted, signed, survived rookie cliff). Primary issues: (a) league-blended numbers hide huge cross-league variance, (b) current $5M p50 anchor is plausible only for one-league subsets (NBA), (c) need explicit position-conditioning hint though full position-conditioning is v3.

Matching data sources:

• —Primary. Spotrac aggregated league salary data 2024. NFL median $895K, NBA $5.0M, MLB $1.5M, NHL $1.95M.
• —Secondary. League CBAs (NFL 2020, NBA 2023, MLB 2022, NHL 2020).
• —Survival source. League career-length actuarial reports — NFL avg 3.3 yr, NBA 4.5 yr, MLB 5.6 yr, NHL 4.4 yr. Already-veteran subset (4+ yr) has much better forward survival.
• —Tail. Forbes Highest-Paid Athletes 2024.

Anchor TEB at listing age 28 (4+ year vet, blended across four leagues weighted by roster size ~3,600 total):

Per-league breakdown (use when issuer’s league is known):

Survival curve $P_{\text{stay}}(t)$ — P(still active in major league at age 28+t):

Exit-cohort fallback: bifurcate by tail. Top 10% transition careers (broadcasting, coaching, business) at $200K-2M; median exited athlete drops to ~$60K-100K. Weighted average ~$180K as exit fallback. This is the cohort where exit-fallback matters most — short careers, large post-career income drops.

Orthogonality alert — growth segments must be revised —The current growth segment “years 8-15: $g \, p_{50} = -0.40$” encodes the career cliff. Adding survival mixture (which models P(still in-cohort) dropping post-year-8) would double-count. Resolution: growth segments describe conditional-on-still-playingtrajectory, NOT unconditional. The “-0.40” segment is revised to “years 5-8: $g \, p_{50} = -0.05$(in-cohort taper)”; the unconditional decline is captured by the survival mixture × exit-fallback. Real change to current values; implementation must apply carefully.

Terminal growth (carry forward per DR-v1): $\{p_{10}: -0.05,\, p_{50}: 0.005,\, p_{90}: 0.025\}$.

Residual error bound: ±15% on per-league anchors; ±30% on roster-weighted blended anchor (sport-mix uncertainty); survival curve ±5% per league per year (well-measured).

Data pull date: 2026-04-26.

Cohort definition. US creator with monetized presence on at least one major platform (YouTube YPP-eligible, Substack paid tier, Patreon active, Twitch Affiliate+, etc.), age 22-40 at listing. $25K creator-income floor added — anyone earning <$25K from creator work is unlikely to be a viable issuer. Refinement of cohort definition, not a sub-cohort split.

Matching data sources:

• —Primary. ConvertKit (Kit) State of the Creator Economy 2024 — median creator income $9.5K/yr; mean $36K; top decile $108K.
• —Primary. Patreon “What Creators Earn” + Graphtreon — median monthly $60 (~$720/yr); p90 $1,800/mo (~$22K/yr); p99 $25K/mo.
• —Triangulation. IRS Schedule SE filings, NAICS 711510 / 711410 (IRS SOI Tax Stats Table 1.4) — median net SE earnings $8.2K (2021).
• —Tail. Forbes Top Creators 2024.

Anchor TEB at listing age 28 (monetized creator earning >$25K/yr):

Without the floor, p50 would be ~$3.5-6K/yr, which makes the engine produce nonsensical V_HCfor issuers who shouldn’t realistically be listing.

Age curve (conditional on still-monetized-creator):

Survival curve $P_{\text{stay}}(t)$— P(still a monetized creator earning >$10K/yr at age 28+t):

Exit-cohort fallback: map to BLS economy-wide median (~$48K for age 28). Exited creators take corporate jobs in marketing, content roles, freelance writing.

Growth segments: verify against double-counting with survival; current values likely encode creator volatility — kept pending implementation review.

Terminal growth (per DR-v1): $\{p_{10}: -0.02,\, p_{50}: 0.01,\, p_{90}: 0.04\}$.

Residual error bound: ±30% on anchor (platform-mix uncertainty); ±35% on age-curve (no public dataset cleanly age-conditions creator earnings); survival curve ±15%.

Data pull date: 2026-04-26.

Cohort definition.US adult age 22-65 with bachelor’s degree or higher, not in any of cohorts 1-5.

Matching data sources:

• —Primary. US Census ACS 1-year 2023 — Table B20004 + PUMS for percentiles.
• —Secondary. Federal Reserve Survey of Consumer Finances 2022.
• —Floor. BLS OES economy-wide median.

Anchor TEB at listing age 28 (US adult, bachelor’s+):

Selection adjustment for PreFlop self-selection signal

PreFlop listers self-select for motivation, financial sophistication, willingness to sign a personal covenant. They are not the median worker.

• —Lower bound: ACS p50 unadjusted ($68K)
• —Upper bound: ACS p75-p90 ($92K-$135K, consistent with fintech-onboarding studies showing new investing-app signups at ~1.4× age-cohort median)
• —Defensible midpoint: ACS p65-p70 (~$80K) at age 28, BA+. Use as the adjusted p50 for the fallback cohort.

Age curve (ACS-derived, age + education conditioned):

Survival curve: not really applicable for the fallback. Use $P_{\text{stay}}(t) \approx 0.92$ flat (10-yr labor force participation per BLS LAUS) and exit fallback maps to retirement income (~$45K median for 65+).

Growth segments:

• —Years 0-7: $g \, p_{50} = 0.04$ (early-career promotion)
• —Years 7-15: $g \, p_{50} = 0.025$ (mid-career)
• —Years 15-25: $g \, p_{50} = 0.015$ (late-career)
• —Years 25+: $g \, p_{50} = 0.005$ (plateau)

Terminal growth (per DR-v1): $\{p_{10}: -0.01,\, p_{50}: 0.02,\, p_{90}: 0.04\}$.

Residual error bound: ±20% on anchor (selection-signal uncertainty); ±15% on age-curve (ACS is well-measured); survival ~±5%.

Data pull date: 2026-04-26.