quantdata.dysanos.grid.gendysanos.static_spread_strategy#
State-dependent traded-spread positive control for dynamic-arbitrage tests.
- class quantdata.dysanos.grid.gendysanos.static_spread_strategy.StateDependentStaticSpreadAccumulator(*, alpha=0.05)[source]#
Bases:
objectReplay predictable traded-grid spread arbitrages without fitting.
At each path state and trading time the policy selects the strike-direction portfolio with the largest strictly positive guaranteed margin after costs and a float32 arithmetic band. The action is zero otherwise. All candidate spreads have L1 norm two before normalization.