Understand how a model portfolio progresses from creation to live production within BIASafe. This guide focuses specifically on the lifecycle of a model, how versions evolve, how backtests are managed, and how a strategy becomes a live production model. It does not cover detailed strategy configuration.
Lifecycle Overview
Every model portfolio within BIASafe follows a structured lifecycle designed to separate research from production while preserving version control, governance discipline, and operational continuity.A model begins as a configurable draft, progresses through backtesting and versioned research, and can ultimately be promoted into a live production model with daily data updates.This framework allows teams to iterate freely during research while maintaining strict controls once a model becomes operational.
A model begins as a draft.During creation, users define the strategy identity and establish the initial framework of the portfolio. Once created, the model exists as a configurable draft that has not yet been validated through historical testing.Draft models are fully editable and are not eligible for deployment.At this stage, the model is simply an idea translated into a working structure.
Save Draft: Preserve Your Progress
Saves the current model configuration without launching a backtest. This allows you to pause your work and return later to continue refining the strategy.
When a model is created and saved as a draft, it becomes visible in the R&D Hub as a model card under the Configuration state.
This card acts as the central entry point to continue working on the model. Draft models remain fully editable at this stage and are not yet validated or eligible for deployment.
Delete Draft: Remove the Saved Configuration
Deletes the current draft model and exits the setup process. All saved configurations and unsaved changes will be permanently removed and cannot be recovered.
Important: Deleting a draft is permanent and cannot be undone.
A draft can also be deleted directly from its configuration card in the Strategy R&D Hub by opening the card actions menu and selecting Delete.
2
Launch a Backtest
Once the draft is created, users configure the strategy through the model creation workflow. This may include universe selection, security eligibility, ranking methodology, weighting logic, and rebalance frequency.When configuration is complete, the final action in the workflow is to launch a backtest.
3
Backtest in Progress
Once submitted, the model enters the Backtest in Progress stage.The selected configuration is temporarily locked while the platform processes historical calculations and generates the validation report. This ensures the version being tested remains unchanged during computation.While one backtest is running, users may still continue research by creating a new draft or duplicating the strategy into a separate model.At this stage, the strategy is transitioning from concept to measurable research output.
Interrupt Backtest: Stop the Running Simulation
Stops the currently running backtest before completion. Any in-progress calculations will be canceled, and no validated backtest version or report will be generated.
4
Versioned Backtest
When processing is complete, the model receives a validated backtest version.The report becomes available for review and typically includes performance history, drawdowns, holdings evolution, turnover, benchmark comparison, and other research metrics.Each completed run is stored as an official version of the strategy.Examples include:
v1 Initial release
v2 Revised filters
v3 Updated weighting logic
v4 Lower turnover methodology
This allows users to iterate while preserving a complete research history.
Restore Previous Backtest Versions
Previous versions remain accessible and may be restored if needed.
Notice: Restoring a historical version does not overwrite prior records. Instead, the platform creates a new latest version based on the selected historical configuration.
For example, restoring Version 2 after Version 4 exists would create a new Version 5.This approach preserves full auditability while allowing prior research frameworks to become current again.
5
Production Activation
Only the latest version of a strategy may be promoted to production.During activation, users finalize the transition from research to live operations. This includes confirming governance approvals, refreshing missing dates since the last validated backtest, and optionally extending the full historical record.
Important: Once confirmed, the selected version is locked as the official production basis and can no longer be edited directly. To continue research or generate new backtests, a new model must be created through duplication.
6
Live Portfolio Model
After deployment, the strategy becomes a Live Model.The model begins generating ongoing daily outputs based on the approved methodology and schedule. It is now available for portfolio implementation, mandate replication, and account-level workflows.Typical live outputs include:
Model Constituents File
Model Corporate Actions File
Model Levels File
At this stage, the strategy operates as a production engine rather than a research model.
7
Archived Model
Models that are no longer active can be archived.Archiving removes the model from active production use while preserving its full history, including versions, backtest reports, deployment records, and live calculation history.Archived models remain accessible for reference, audit, governance review, and potential future reuse through duplication.
Important: Archiving is a permanent lifecycle action and cannot be undone.An archived model cannot be reactivated directly. To use the strategy again, a new model must be created through duplication.
A portfolio model inside BIASafe is built to move through a disciplined progression:
Draft models become tested strategies.
Completed backtests become versioned research records.
Approved versions become live production models.
Live models can be duplicated for new research paths or archived when no longer active.
This lifecycle gives teams the flexibility to iterate during research while preserving the discipline, traceability, and stability required in production.
