> For the complete documentation index, see [llms.txt](https://docs.openetf.xyz/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.openetf.xyz/how-it-works/risk-controls/managed-execution.md).

# Managed Execution

**Managed Execution is the single path every trade takes: OpenETF checks the request, signs it with the fund's own Trading Agent Wallet and submits it to Hyperliquid. Managers get full trading freedom within the fund's rules, and every trade is held to those rules, whichever client sends it.**

## Why a managed path

The Manager never holds the fund's agent key, and the Vault contract has no Manager withdrawal function ([why](https://docs.openetf.xyz/how-it-works/hypercore-hyperevm.md#why-the-manager-cannot-withdraw)). Neither stops losses through trading: the Manager, or anyone who gains trading authority, can harm the fund by trading.

What an unmanaged agent key would still allow is acting outside the fund's rules: trading markets the fund does not support, spending cash reserved for redemptions, trading beyond a published Mandate, or depositing the fund's margin into a Hyperliquid Legacy Vault. Managed Execution closes that gap. Each ETF's Trading Agent Wallet is generated and kept inside OpenETF's signing infrastructure, never exported and never shown to the Manager, an API client or anyone else. The Manager directs trades; the key signs only requests that pass the checks below.

## Two ways in, one path

| Client                                                        | Who uses it                                              | What it offers                                                                                             |
| ------------------------------------------------------------- | -------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------- |
| **Trading Terminal**                                          | The Manager, by hand                                     | The ETF's positions, orders and available margin in one workspace, next to its NAV and pending redemptions |
| [**OpenETF API**](https://docs.openetf.xyz/developers/api.md) | The Manager's bots, strategies and third-party terminals | Programmatic access with scoped, revocable credentials                                                     |

Both clients submit the same kind of request and pass the same checks. Manual and automated activity draw on the same capacity, so moving a strategy from the Terminal to a bot needs no new signing key or trading wallet. An API credential identifies the ETFs and actions it covers. It is never a signing key and never returns a signature that could be broadcast elsewhere.

## Checked before signing

Before any signature, Managed Execution checks:

* **who** is asking, and whether they may act for this ETF;
* **what** the action is: only known trading actions and defined movements between the fund's own balances are allowed, and anything else is rejected by default;
* **where**: whether the market is one of the fund's [Supported Markets](https://docs.openetf.xyz/how-it-works/risk-controls/supported-markets.md);
* **whether it still applies**: whether the request is still valid and consistent with the fund's current positions, open orders and restrictions.

Transfers to anyone else, withdrawals, staking, delegation and deposits into Hyperliquid Legacy Vaults are rejected, whatever a request calls itself. Warnings and limits in the interface help usability; the authoritative checks run on the backend, immediately before signing.

When an ETF runs under an active [Mandate](https://docs.openetf.xyz/how-it-works/risk-controls/mandate.md) (Coming soon), the Manager's published trading limits are checked here too.

## One request, one outcome, one trail

Retrying a request never creates a second order. Revoking a credential, rotating a key or tightening a restriction takes effect at once, before any earlier approval is signed or sent. An exchange can still reject an order or fill it partially: a submission, a fill and a settled balance are different facts, and each is recorded.

## When something goes wrong

If required data goes stale, a check cannot complete or a key may be compromised, Managed Execution stops accepting risk-increasing actions. A reduction, cancellation or close still goes through when the service can show that it reduces risk; when it cannot, the service is Suspended and ordinary signing stops until it is repaired. Emergency controls can restrict a single market, a single action or the whole service.

Checks run inside the managed signing path. If a managed key were stolen and used outside that path, detection, revoking the compromised authority, key rotation and protocol-level limits would carry the response. [Who controls what](https://docs.openetf.xyz/how-it-works/control.md) sets out that dependency.

## What Managed Execution does not do

It does not choose trades, recommend positions or judge whether a strategy is prudent. It enforces the fund's rules on every request, whoever sends it. Strategy results remain the Manager's.

Next: [Mandate](https://docs.openetf.xyz/how-it-works/risk-controls/mandate.md) · Related: [Trading an ETF](https://docs.openetf.xyz/managing/trading.md)
