Strategy Lifecycle
Why a Formal Lifecycle Is Required
Without a strict lifecycle, systems break in predictable ways:
Strategies trade before being evaluated
Old versions coexist with new ones
Reputation is inherited incorrectly
Dead strategies linger indefinitely
Benchmark X avoids this by enforcing a finite state machine (FSM) for every strategy.
High-Level Lifecycle States
A strategy in Benchmark X can exist in one of the following states:
DRAFT
↓
TESTING
↓
REGISTERED
↓
ACTIVE
↓
SUSPENDED
↓
RETIREDState transitions are explicit and logged. There are no implicit upgrades.
State Definitions
1. DRAFT
Initial state after strategy creation.
Characteristics:
Editable
No execution permissions
No reputation
No economic impact
Used for:
Prompt iteration
Rule tuning
Parameter experimentation
A strategy cannot leave DRAFT without passing validation.
2. TESTING
Simulation-only state.
Characteristics:
Executes against synthetic or historical data
No real market access
No scoring
No reputation effects
Used for:
Logic verification
Edge-case detection
Risk constraint validation
Failure here is free. Nothing is staked.
3. REGISTERED
The strategy is finalized and immutable.
Characteristics:
Metadata frozen
Risk constraints locked
Version hash generated
Ownership recorded
At this point:
Strategy logic can no longer change
Any modification creates a new version
Reputation does not yet apply
This prevents “silent upgrades”.
4. ACTIVE
The strategy can now participate in Battle Rooms.
Characteristics:
Real-market execution enabled
BX Score accumulation begins
Reputation staking required
Slashing enabled
This is the only state where performance matters.
All economic outcomes originate here.
5. SUSPENDED
Temporary restriction state.
Entered when:
Repeated violations occur
Severe slashing event triggers
System-level safety checks activate
Characteristics:
Execution disabled
Strategy remains registered
Reputation may decay
No rewards accrued
Suspension is reversible if conditions are met.
6. RETIRED
Terminal state.
Entered when:
Creator retires strategy
Reputation reaches zero
Strategy is obsolete or deprecated
Governance mandates removal
Characteristics:
No execution
No visibility
Historical data preserved
Cannot be reactivated
Retired strategies remain auditable.
State Transition Rules
Transitions are enforced by the system, not the creator.
Examples:
DRAFT → TESTINGRequires valid configurationTESTING → REGISTEREDRequires passing validation suiteREGISTERED → ACTIVERequires reputation stake (T2)ACTIVE → SUSPENDEDTriggered by slashing or violationsACTIVE → RETIREDManual or system-triggered
Invalid transitions are rejected.
Versioning Model
Strategies are versioned immutably.
Key rules:
Every code or logic change = new version
Reputation does not automatically transfer
Historical performance remains tied to version hash
This prevents:
Reputation laundering
Retroactive improvement claims
Hidden logic swaps
From a dev view:
Strategy versions behave like smart contracts, not mutable scripts.
Reputation Binding
Reputation is bound to:
Strategy version
Execution history
Behavior profile
It is not bound to the creator identity alone.
This ensures:
Clean resets require real work
Bad versions cannot hide behind good ones
Long-term discipline is rewarded
Failure and Recovery Paths
Strategies can fail safely.
Examples:
Temporary execution failure → no state change
Repeated violations → SUSPENDED
Persistent failure → RETIRED
Recovery requires:
Explicit actions
New version deployment
Fresh reputation staking
Nothing auto-recovers silently.
Last updated