quantdata.dysanos.grid.empirical_arbitrage#

Empirical arbitrage diagnostics for finite simulated path sets.

Array convention#

dC[omega, t, ell] = dC_t^ell(omega), with shape (num_paths, num_steps, num_instruments).

The basic LP uses deterministic positions a[t, ell], common to every path. It therefore does not use pathwise foresight.

class quantdata.dysanos.grid.empirical_arbitrage.ArbitrageConfig(delta=0.0, prop_cost=0.0, max_abs_position=1.0, payoff_tolerance=1e-10, active_tolerance=1e-08, method='highs', presolve=True, threads=0)[source]#

Bases: object

delta: float = 0.0#
prop_cost: float = 0.0#
max_abs_position: float = 1.0#
payoff_tolerance: float = 1e-10#
active_tolerance: float = 1e-08#
method: str = 'highs'#
presolve: bool = True#
threads: int = 0#
class quantdata.dysanos.grid.empirical_arbitrage.ArbitrageResult(solver_success, found, message, config, positions=None, payoff=None, gross_payoff=None, total_cost=None, minimum_payoff=None, mean_payoff=None, maximum_payoff=None, l1_position=None, max_abs_position=None, num_active=None, fraction_negative=None, mc_error=None, solver_status=None, solver_nit=None)[source]#

Bases: object

positions: NDArray[float64] | None = None#
payoff: NDArray[float64] | None = None#
gross_payoff: NDArray[float64] | None = None#
total_cost: NDArray[float64] | None = None#
minimum_payoff: float | None = None#
mean_payoff: float | None = None#
maximum_payoff: float | None = None#
l1_position: float | None = None#
max_abs_position: float | None = None#
num_active: int | None = None#
fraction_negative: float | None = None#
mc_error: NDArray[float64] | None = None#
solver_status: int | None = None#
solver_nit: int | None = None#
summary()[source]#
class quantdata.dysanos.grid.empirical_arbitrage.SparseResult(solver_success, message, required_mean_payoff, positions=None, payoff=None, gross_payoff=None, total_cost=None, minimum_payoff=None, mean_payoff=None, maximum_payoff=None, l1_position=None, num_active=None, fraction_negative=None, solver_status=None, solver_nit=None)[source]#

Bases: object

positions: NDArray[float64] | None = None#
payoff: NDArray[float64] | None = None#
gross_payoff: NDArray[float64] | None = None#
total_cost: NDArray[float64] | None = None#
minimum_payoff: float | None = None#
mean_payoff: float | None = None#
maximum_payoff: float | None = None#
l1_position: float | None = None#
num_active: int | None = None#
fraction_negative: float | None = None#
solver_status: int | None = None#
solver_nit: int | None = None#
summary()[source]#
quantdata.dysanos.grid.empirical_arbitrage.mc_error_by_instrument(dC)[source]#

Standard error of the sample mean for each (time, instrument).

quantdata.dysanos.grid.empirical_arbitrage.build_cost_tensor(dC, X=None, *, delta=0.0, prop_cost=0.0)[source]#

Return pathwise absolute-position cost coefficients and MC errors.

quantdata.dysanos.grid.empirical_arbitrage.evaluate_positions(dC, positions, X=None, *, delta=0.0, prop_cost=0.0)[source]#
quantdata.dysanos.grid.empirical_arbitrage.find_empirical_arbitrage(dC, X=None, *, delta=0.0, prop_cost=0.0, max_abs_position=1.0, payoff_tolerance=1e-10, active_tolerance=1e-08, method='highs', presolve=True, threads=0, verbose=False, solver_output=False)[source]#

Maximise mean net payoff subject to non-negative payoff on every path.

quantdata.dysanos.grid.empirical_arbitrage.sparsify_empirical_arbitrage(dC, X=None, *, required_mean_payoff, delta=0.0, prop_cost=0.0, payoff_tolerance=1e-10, active_tolerance=1e-08, weights=None, method='highs', presolve=True, threads=0, verbose=False, solver_output=False)[source]#

Minimise weighted L1 exposure subject to an arbitrage certificate.

quantdata.dysanos.grid.empirical_arbitrage.iterative_reweighted_l1(dC, X=None, *, required_mean_payoff, delta=0.0, prop_cost=0.0, num_iterations=5, reweight_epsilon=1e-06, payoff_tolerance=1e-10, active_tolerance=1e-08, threads=0, verbose=False, solver_output=False)[source]#

Iteratively reweighted L1 heuristic for fewer active positions.

quantdata.dysanos.grid.empirical_arbitrage.position_contributions(dC, positions, X=None, *, delta=0.0, prop_cost=0.0)[source]#
quantdata.dysanos.grid.empirical_arbitrage.rank_positions(dC, positions, X=None, *, delta=0.0, prop_cost=0.0, top_n=100, active_tolerance=1e-08)[source]#
quantdata.dysanos.grid.empirical_arbitrage.deletion_importance(dC, positions, X=None, *, delta=0.0, prop_cost=0.0, active_tolerance=1e-08)[source]#
quantdata.dysanos.grid.empirical_arbitrage.greedy_prune(dC, positions, X=None, *, minimum_mean_payoff, delta=0.0, prop_cost=0.0, payoff_tolerance=1e-10, active_tolerance=1e-08)[source]#
quantdata.dysanos.grid.empirical_arbitrage.step_payoff_summary(dC, positions, X=None, *, delta=0.0, prop_cost=0.0)[source]#
quantdata.dysanos.grid.empirical_arbitrage.binding_paths(payoff, count=100)[source]#
quantdata.dysanos.grid.empirical_arbitrage.compare_rays(results)[source]#
quantdata.dysanos.grid.empirical_arbitrage.save_experiment(output_dir, name, dC, X, result, sparse, *, delta, prop_cost)[source]#
quantdata.dysanos.grid.empirical_arbitrage.run_standard_experiments(dC, X, *, output_dir='empirical_arbitrage_results', sparse_fractions=(0.5, 0.25, 0.1), payoff_tolerance=1e-10, active_tolerance=1e-08, max_abs_position=1.0, threads=0, verbose=True, solver_output=True)[source]#

Run the three requested experiments and save their diagnostics.

quantdata.dysanos.grid.empirical_arbitrage.main(argv=None)[source]#