Execution contract

Use this page to define how a serious execution bridge should behave before any live integration exists.

01
Keep paper and execution separate
02
Stage order intent, not live orders
03
Require human review
01
Separate signal generation from order creation
02
Use explicit portfolio sleeves
03
Document every operational assumption
Operational boundary

The product is allowed to define intent, sizing, and review checkpoints. It is not allowed to fire orders.

That separation matters because the paper layer answers a research question while a broker layer creates real operational risk. Alphrex should only cross that boundary once order generation, approvals, and fail-safe logic are explicit and auditable.

Signal cadence Daily close freeze
Order output Intent spec only
Human gate Required
01

Freeze research state

Signals, alerts, benchmark context, and regime reads must be captured after the daily refresh finishes. No broker plan should consume a half-updated workspace.

02

Generate order intent

Translate a sleeve or portfolio into a structured intent: target exposure, benchmark context, cadence, order-style preference, and the risk checks that must pass.

03

Run pre-trade checks

Support policy, stale-data warnings, drawdown and regime alerts, rebalance bands, and cash buffers all need to be green or explicitly overridden.

04

Approve and export

Only after review should the system produce a broker-specific export. Approval, timestamps, and any overrides need to be recorded alongside the order spec.

Strategy Sleeve Plans

Each paper strategy becomes an execution candidate only after alerts, cadence, and benchmark context are explicit.

0 tracked
No tracked sleeves yet

Once you have paper strategies, this page will turn them into broker-ready planning records instead of trying to jump straight from backtest to execution.

Portfolio Rebalance Specs

Portfolio execution should consume explicit sleeve targets, rebalance bands, and cash buffers rather than narrative conviction.

0 saved
No portfolio plans yet

Save a paper portfolio first. The execution layer will then expose sleeve targets, rebalance assumptions, and review checks without pretending to be a live OMS.

Broker Handoff Spec

The future adapter should receive a deterministic payload, not scrape the UI.

Minimum payload fields
intent_idstable execution record id
as_of_datefrozen research timestamp
order_styleMOO, limit, staged rebalance
risk_checksalerts, caps, override log
benchmark_contextactive-risk read and policy notes
Example intent payload
{
  "intent_id": "paper-42-2026-04-15",
  "source": "alphrex.execution_playbook",
  "mode": "planning_only",
  "rebalance_window": "next_session_open",
  "sleeves": [
    {"ticker": "SPY", "strategy": "ma_crossover", "target_weight_pct": 35.0}
  ],
  "risk_checks": {
    "alerts_clear": false,
    "human_approval_required": true
  }
}