Brainstormed candidates — each grounded in a structural reason, not a backtest curve-fit. Low turnover by design (the project's hard drawdown constraint). These are hypotheses to test, not signals to trade.
Every idea here pairs a structural driver (carry, storage, seasonality, curve factor, scheduled catalyst) with a risk overlay (the carry floor, inventory regime, hard stops). Spreads reward patience: multi-day holds, tight-but-honest sizing from the OU half-life, and humility about the days the model has no signal — April 2020 is the permanent reminder that a "−2σ" can keep going.
"Edge is structure you can name. Everything else is a curve fit waiting to fail."