Profile catalog¶
Generated from INDUSTRY_PROFILES — run examples/profile_catalog/render_catalog.py after changing the registry; the page never drifts from code.
| key | industry | fit | base default | checks | benchmarks |
|---|---|---|---|---|---|
| consumer_electronics | Consumer electronics / hardware | strong | trend | asp_rising | 2 bands |
| semiconductor | Semiconductors (mature cycles) | strong | trend | hypergrowth_base | 2 bands |
| saas_subscription | SaaS / subscription | adapt | net_growth | arpu_accelerating, net_churn_positive | 2 bands |
| advertising | Advertising | adapt | trend | adload_high | 2 bands |
| retail_store | Retail (store network) | adapt | trend | shrinking_base | 2 bands |
| telecom_subscriber | Telecom / subscribers | adapt | net_growth | base_saturated, net_churn_positive | 2 bands |
| industrial_capacity | Industrial / capacity-driven | adapt | trend | utilization_cap | 2 bands |
| financial_interest | Financials (interest income) | weak | growth | balance_growth_hot | 2 bands |
| commodity_cyclical | Commodities / cyclical | weak | hold | cycle_top | 2 bands |
| regime_shift_tech | Regime-shift tech (AI inflection) | weak | trend | regime_always | — |
consumer_electronics¶
Consumer electronics / hardware · 消费电子 / 硬件 · fit: strong — trend-driven: shipments, attach rates and ASPs continue their history (Luxun consumer-electronics hold-out −0.1% sMAPE)
| driver | default method | params |
|---|---|---|
| base | trend |
— |
| penetration | incremental |
delta_pp=0.02 |
| share | hold |
— |
| price | erosion |
rate=0.05 |
checks: asp_rising
benchmarks (Damodaran US clusters, 2026-01, grade B):
| metric | P25 | P50 | P75 | cluster |
|---|---|---|---|---|
| revenue_cagr_5y | 4.2% | 7.0% | 7.5% | Electronics (Consumer & Office), Electronics (General), Computers/Peripherals, Office Equipment & Services |
| revenue_exp_growth_2y | 1.2% | 10.2% | 19.8% | Electronics (Consumer & Office), Electronics (General), Computers/Peripherals, Office Equipment & Services |
semiconductor¶
Semiconductors (mature cycles) · 半导体（成熟周期） · fit: strong — mature-semi factors trend until they don't (NVDA Gaming 1.0% sMAPE); ASPs hold rather than erode (pricing power per node), attach rates are structural product-cycle choices (held)
| driver | default method | params |
|---|---|---|
| base | trend |
— |
| penetration | hold |
— |
| share | hold |
— |
| price | hold |
— |
checks: hypergrowth_base
benchmarks (Damodaran US clusters, 2026-01, grade B):
| metric | P25 | P50 | P75 | cluster |
|---|---|---|---|---|
| revenue_cagr_5y | 9.8% | 10.3% | 10.7% | Semiconductor, Semiconductor Equip |
| revenue_exp_growth_2y | 19.2% | 26.4% | 33.7% | Semiconductor, Semiconductor Equip |
saas_subscription¶
SaaS / subscription · SaaS / 订阅 · fit: adapt — MAU/customers × ARPU: adoption follows an S-curve, ARPU grows by escalator — the tree works with swapped factors
| driver | default method | params |
|---|---|---|
| base | net_growth |
gross_rate=0.3, churn=0.12 |
| penetration | logistic |
L=0.6, k=0.35, t0=anchor_last |
| share | hold |
— |
| price | growth |
rate=0.02 |
checks: arpu_accelerating, net_churn_positive
benchmarks (Damodaran US clusters, 2026-01, grade B):
| metric | P25 | P50 | P75 | cluster |
|---|---|---|---|---|
| revenue_cagr_5y | 16.4% | 23.3% | 27.6% | Software (System & Application), Software (Internet), Computer Services, Information Services |
| revenue_exp_growth_2y | 11.9% | 18.7% | 26.4% | Software (System & Application), Software (Internet), Computer Services, Information Services |
advertising¶
Advertising · 广告 · fit: adapt — traffic × ad-load × eCPM: ad load is sticky (product choice), eCPM is cyclical and mean-reverts with the ad market
| driver | default method | params |
|---|---|---|
| base | trend |
— |
| penetration | hold |
— |
| share | hold |
— |
| price | mean_revert |
target=None, speed=0.3 |
checks: adload_high
benchmarks (Damodaran US clusters, 2026-01, grade B):
| metric | P25 | P50 | P75 | cluster |
|---|---|---|---|---|
| revenue_cagr_5y | 6.3% | 12.4% | 16.7% | Advertising, Entertainment, Publishing & Newspapers, Broadcasting |
| revenue_exp_growth_2y | 0.5% | 4.6% | 8.3% | Advertising, Entertainment, Publishing & Newspapers, Broadcasting |
retail_store¶
Retail (store network) · 零售（门店网络） · fit: adapt — stores × sales-per-store: openings are company-controlled (trend), same-store sales mean-revert to CPI + low single digits
| driver | default method | params |
|---|---|---|
| base | trend |
— |
| penetration | hold |
— |
| share | hold |
— |
| price | mean_revert |
target=None, speed=0.5 |
checks: shrinking_base
benchmarks (Damodaran US clusters, 2026-01, grade B):
| metric | P25 | P50 | P75 | cluster |
|---|---|---|---|---|
| revenue_cagr_5y | 2.3% | 8.4% | 10.1% | Retail (General), Retail (Special Lines), Retail (Automotive), Retail (Building Supply), Retail (Grocery and Food), Retail (Distributors) |
| revenue_exp_growth_2y | 4.2% | 4.9% | 7.0% | Retail (General), Retail (Special Lines), Retail (Automotive), Retail (Building Supply), Retail (Grocery and Food), Retail (Distributors) |
telecom_subscriber¶
Telecom / subscribers · 电信 / 用户数 · fit: adapt — subscribers × ARPU: subscriber growth saturates (logistic), ARPU drifts slowly and is policy-capped
| driver | default method | params |
|---|---|---|
| base | net_growth |
gross_rate=0.05, churn=0.035 |
| penetration | hold |
— |
| share | hold |
— |
| price | hold |
— |
checks: base_saturated, net_churn_positive
benchmarks (Damodaran US clusters, 2026-01, grade B):
| metric | P25 | P50 | P75 | cluster |
|---|---|---|---|---|
| revenue_cagr_5y | 8.6% | 13.6% | 20.3% | Telecom (Wireless), Telecom. Services, Cable TV |
| revenue_exp_growth_2y | -4.2% | -2.9% | 25.9% | Telecom (Wireless), Telecom. Services, Cable TV |
industrial_capacity¶
Industrial / capacity-driven · 工业 / 产能驱动 · fit: adapt — capacity × utilization × price: utilization is bounded and mean-reverts (70–90%); capacity is a capex decision (trend)
| driver | default method | params |
|---|---|---|
| base | trend |
— |
| penetration | mean_revert |
target=0.8, speed=0.4 |
| share | hold |
— |
| price | trend |
— |
checks: utilization_cap
benchmarks (Damodaran US clusters, 2026-01, grade B):
| metric | P25 | P50 | P75 | cluster |
|---|---|---|---|---|
| revenue_cagr_5y | 8.8% | 11.0% | 13.6% | Machinery, Electrical Equipment, Engineering/Construction, Building Materials |
| revenue_exp_growth_2y | 8.6% | 11.4% | 28.4% | Machinery, Electrical Equipment, Engineering/Construction, Building Materials |
financial_interest¶
Financials (interest income) · 金融（利息收入） · fit: weak — interest-earning assets × yield: the rate cycle is not in the historical window — point forecasts structurally unreliable
| driver | default method | params |
|---|---|---|
| base | growth |
rate=0.08 |
| penetration | hold |
— |
| share | hold |
— |
| price | mean_revert |
target=0.04, speed=0.5 |
checks: balance_growth_hot
benchmarks (Damodaran US clusters, 2026-01, grade B):
| metric | P25 | P50 | P75 | cluster |
|---|---|---|---|---|
| revenue_cagr_5y | 8.3% | 8.6% | 8.8% | Bank (Money Center), Banks (Regional) |
| revenue_exp_growth_2y | 9.8% | 11.0% | 12.2% | Bank (Money Center), Banks (Regional) |
When fit is weak: anchor the yield to the forward policy-rate curve and run rate scenarios; treat the balance-sheet growth as a credit-cycle variable, not a trend
commodity_cyclical¶
Commodities / cyclical · 大宗商品 / 周期 · fit: weak — price is cyclical around marginal cost — trend extrapolation peaks the forecast at exactly the wrong moment of the cycle
| driver | default method | params |
|---|---|---|
| base | hold |
— |
| penetration | hold |
— |
| share | hold |
— |
| price | mean_revert |
target=None, speed=0.3 |
checks: cycle_top
benchmarks (Damodaran US clusters, 2026-01, grade B):
| metric | P25 | P50 | P75 | cluster |
|---|---|---|---|---|
| revenue_cagr_5y | 9.3% | 14.5% | 18.8% | Metals & Mining, Coal & Related Energy, Oil/Gas (Integrated), Oil/Gas (Production and Exploration), Steel, Precious Metals |
| revenue_exp_growth_2y | 6.8% | 23.4% | 46.8% | Metals & Mining, Coal & Related Energy, Oil/Gas (Integrated), Oil/Gas (Production and Exploration), Steel, Precious Metals |
When fit is weak: mean-revert price to marginal cost and run demand/capacity-shock scenarios; watch leading indicators (inventories, spreads) instead of fitting the trend
regime_shift_tech¶
Regime-shift tech (AI inflection) · 范式跳变科技（AI 拐点） · fit: weak — NVDA Data Center: same formula as Gaming, 60% hold-out sMAPE (actual $115B vs trend $18B) — the breakout is not in the training data, and no trend fit can recover it
| driver | default method | params |
|---|---|---|
| base | trend |
— |
| penetration | trend |
— |
| share | trend |
— |
| price | trend |
— |
checks: regime_always
No industry benchmarks by design — the profile's heuristic checks are the only layer.
When fit is weak: the point forecast below is a baseline, not a forecast: report Monte Carlo scenarios (simulate_segment) with wide honest ranges plus explicit trigger conditions; update fast as the regime reveals itself