This breaks you out of local optima when backtesting iterations stall. It watches your iteration history for plateau signals, overfitting, transaction cost defeat, or tail risk blowouts, then generates structurally different strategy proposals instead of nudging parameters again. The pivot techniques are assumption inversion, archetype switching, and objective reframing. You feed it your backtest evaluation JSONs and current strategy YAML, and it spits out ranked alternatives ready for the candidate agent pipeline or further research. It's the escape hatch for when you've tuned a strategy to death but the Sharpe ratio won't budge. Works best when you've actually accumulated enough iteration history to detect the stagnation pattern.
npx skills add https://github.com/tradermonty/claude-trading-skills --skill strategy-pivot-designer