P34 turns a menu of possible business actions into a sized decision plan. It is the product of HyperC’s research into learning economic decisions from incomplete information, selective outcome histories and changing markets. It evaluates what to take, how much to commit and when to pass, using your history and operating economics.
The stack connects research to practice: economic models + simulations + P34 + agents + market tools + operating feedback. Your agent gathers information and prepares actions; P34 evaluates the economic choices; the workspace supports the workflow under your controls.
Everything P34 does is one loop, run with discipline.
| Concept | What it means |
|---|---|
| Business menu | The set of actions your business could take right now — deals to buy, loans to approve, loads to accept, offers to make — each with quantities, timing, counterparties, prices and constraints. |
| Context | What the market and your business look like around that menu: history, inventory, capital, seasonality, fees. |
| Decision model | P34 evaluates the menu as an economic decision problem, not only as a prediction task. The output is a decision, with predicted economics attached. |
| No-trade | Doing nothing is a valid output. When no candidate meets the model's criteria, P34 declines — a first-class, rewarded decision, not a failure mode. |
| Portfolio output | P34 selects actions and sizes them as one coordinated portfolio or workflow, so positions don't fight each other for capital or capacity. |
| Partial observability | The true market state is never fully visible. Operational telemetry — your realized outcomes — carries the signal that missing price discovery doesn't. |
| Controlled autonomy | The system can move from recommendation to execution — but only within user-defined limits and rollout stages. |
Language models interpret business descriptions, investigate sources and coordinate tools. They help turn an operating workflow into structured inputs whose assumptions can be inspected and tested.
P34/PARML uses profit as the learning target and evaluates candidate actions in their market context. Simulation and historical outcomes help test which selections and quantities remain attractive after the business’s costs and constraints. The output is a plan with predicted economics to compare against realized results.
This is the working loop behind our research into generalized arbitrage: investigate differences in value, evaluate the whole opportunity menu, act within approved limits and learn from what happens. Read the economic foundations and evidence →
Business history is biased: you only observe outcomes for the deals a previous policy already accepted. Models trained naively on that history become systematically optimistic — they inherit yesterday's enthusiasm and none of its regret. P34's model class treats declining to act as a rewarded output and controls false positives explicitly. In live operation, most of what P34 sees, it refuses.
Every candidate action is embedded with its context — including candidates historical policies never touched.
The model is penalized for confident acceptance of deals that would have lost money — not just rewarded for winners.
Survivors are sized jointly under constraints. If nothing survives, the answer is nothing — and that answer is respected.
Four stages from decision support to self-driving business. Toggle the view — the loop doesn't change; the stakes and the sign-off do.
You: read ranked menus, decide, act.
P34: scores and explains.
Controls: none needed — nothing executes.
You: approve each prepared action.
Your agent: assembles orders via the API.
Controls: per-action approval, budget cap.
You: review daily digests.
P34 + agent: execute approved action classes.
Controls: caps, drawdown limits, instant pause.
You: set the envelope, watch the P&L.
System: full loop within limits.
Controls: audit logs, drift monitoring, kill switch.
Stages labeled “Rolling out” / “Planned” are roadmap: planned, not guaranteed, and gated by market, controls readiness and legal review.