Designing Data Dependencies for Simulated Trading Strategy Workflows

Summary

This platform guide describes two ways to organize data preparation and simulated trading. In one workflow, a separate task cleans and stores factors, then a trading task consumes its output. In the other, one task first prepares and saves the data and then runs the strategy logic. The guide distinguishes task dependency labels from input and output data references, explaining that data outputs must be mapped to tables or views and that downstream jobs can use outputs only after an upstream task completes successfully.

For the separate-task approach, the factor task publishes its output data and the strategy task declares that data as an input, allowing updates to trigger strategy runs. For the combined approach, data production and trading signals belong to the same daily task, with an option to protect the data-processing code when sharing. These are platform workflow instructions, not evidence that either architecture improves returns or signal quality; successful task completion alone does not establish that the data or strategy is valid.

Key ideas

• Data preparation can run as a separate upstream task or inside the strategy task.
• Downstream jobs must reference outputs from completed upstream tasks to establish dependencies.
• Input and output data references connect tasks to data tables or views.
• The guide describes workflow configuration and sharing controls, not strategy performance evidence.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.