Markets & use cases

Computable markets

A market is computable when feasible actions can be enumerated, outcomes can be attributed back to decisions, constraints can be encoded, and repeated feedback exists. Not every market qualifies — and we won't claim it does. Market support is published deliberately, with states.

Interactive check

Is your market computable?

Six criteria from the published market-fit framework. Answer honestly — P34's whole discipline is refusing what doesn't clear the bar.

Menu representationCan available actions be represented as rows with quantities, timing, counterparties, prices and constraints?
Measurable outcomeCan the business measure profit, contribution margin, recovery, yield or another economic target?
Historical choice signalDid the business historically have more options than it selected?
Executable constraintsCan capital, inventory, compliance, risk, liquidity, capacity and platform rules be encoded?
Market impactDoes taking one action preserve enough validity in the remaining menu?
Testable deploymentCan P34 be backtested, forward tested, shadowed and piloted before scale?
Answer all 6 criteria to see the computability read.

This check reflects structural fit only. Supported status additionally requires data validation, model coverage, execution understanding and risk controls under a defined methodology.

Coverage map

Market states

Four states: Supported Pilot-ready Research Not currently supported

Ordinary commerce — the launch perimeter

Supported

Amazon wholesale — US & EU

Purchase and shipping portfolio decisions from live wholesale menus. In production since 2023 at ~$100M/yr reseller scale.

Menu · objective · maturity

Menu: SKU × quantity × supplier offers with lead times and fees.

Objective: contribution margin after fees, returns, holding costs.

Controls: budget caps, per-SKU limits, human PO approval.

Maturity: live production; member workflows available today.

Pilot-ready

Amazon discounter / online arbitrage

Retail and online arbitrage on public telemetry (Keepa-class data), pointed at small-ticket buyers.

Menu · objective · maturity

Menu: discounted listings × quantity with price history.

Objective: resale margin net of fees and returns risk.

Maturity: pipeline case; pilot by application.

Research

Collectibles & niche resale

Domain drops, vinyl (Discogs), retro games (PriceCharting), LEGO, TCG, watches — the finest public comp datasets in commerce.

Menu · objective · maturity

Menu: live listings/auctions × condition tiers.

Objective: realized flip margin; abstention discipline is decisive.

Maturity: research workflows; member experiments welcome.

Regulated commerce — gated behind market-specific review

Pilot-ready

Micro-lending (individual)

Lend / no-lend calls where credit history is sparse. ~3,000 loans issued in a live model test (2025). Enterprise pilots only, after compliance review.

Pilot-ready

Card overdraft / micro-loan & onboarding

Limit increases and calibrated onboarding offers scored on lifetime economics. Active discovery cases with banks.

Research

Manager underwriting, MCA, insurance-like

CLO manager selection, B2B advances, plan payments, customs and fraud insurance — menu-shaped decisioning under partial observability.

Financial markets — separate perimeter

Research

Crypto transaction routing

Routing and execution in adversarial order flow. Deployed in research; a separate product and regulatory perimeter — not part of membership.

Not currently supported

Securities, derivatives, prediction markets

Regulated-market uses are excluded from profit-share pricing and gated under the API Terms. We do not infer support from ordinary-commerce positioning.

Not currently supported

Restricted categories

Event-ticket resale, reservation arbitrage, aged accounts and similar categories with legal or platform-rule exposure are declined by policy.

Why we publish states instead of promises

Each supported market ships with a capability matrix: supported data, action types, automation level, known failure modes, capital assumptions, execution dependencies and model version. If we can't publish that matrix, the market isn't supported — it's research.

The long list

Customer interviews and internal research track 45+ candidate markets — from freight and GPU capacity to tax credits and equipment rental. The working paper's application templates show how each maps to the MSD data contract.

See the research →