strategy-pivot-designer

To break backtest stagnation, you can analyze your iteration history JSON and source strategy YAML to generate structurally different strategy pivots. This process detects performance plateaus and outputs new candidate drafts to escape local optima.

Strategy pivots are triggered by deterministic stagnation signals in your iteration history, specifically improvement plateaus, overfitting proxies, cost defeat, and tail risk. These flags indicate that further parameter tweaking is unlikely to improve real-world performance.

Yes, you can feed iteration history JSON and source strategy YAML into stagnation diagnostics to detect overfitting proxies. The analysis evaluates your tuning loop history and flags when parameter adjustments become brittle or overfitted.

You apply these pivot generation techniques by processing your stalled strategy YAML through diagnostic routines. The system automatically applies assumption inversion, archetype switch, and objective reframe approaches to produce and score new candidate proposals.

The strategy pivot generator outputs pivot_drafts in YAML, pivot_report markdowns, pivot_manifest JSON, and pivot_diagnosis JSON. These files are structured for direct integration into edge-candidate-agent and backtest-expert pipelines for further validation.

You should avoid automated pivot generation if your iteration history JSON lacks sufficient tuning rounds to establish a baseline plateau, or if your strategy YAML does not contain structured parameters suitable for objective reframe and inversion techniques.