StochasticRoundingStrategy Class

Definition

Namespace: Bodu.Financial

Assembly: Bodu.Financial.dll

Package: Bodu.Financial 1.0.0

Source: StochasticRoundingStrategy.cs

An IRoundingStrategy that rounds a value up or down probabilistically, with the probability of rounding up equal to the fractional part discarded at the target scale. Over many roundings the expected value equals the raw amount, so the convention is statistically unbiased and does not accumulate the directional drift that a fixed midpoint rule can introduce across a long series of operations.

public sealed class StochasticRoundingStrategy : IRoundingStrategy

Inheritance: object

StochasticRoundingStrategy

Implements: IRoundingStrategy

Inherited Members: object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.ReferenceEquals(object, object)

object.ToString()

Extension Methods: ComparableExtensions.AtLeast<T>(T, T, IComparer<T>)

ComparableExtensions.AtMost<T>(T, T, IComparer<T>)

ComparableExtensions.IsEqualTo<T>(T, T, IComparer<T>)

ComparableExtensions.Max<T>(T, T, IComparer<T>)

ComparableExtensions.Min<T>(T, T, IComparer<T>)

Remarks

For a value whose magnitude beyond scale digits has fractional part f (in the half-open interval [0, 1)), this strategy rounds up with probability f and down with probability 1 - f. A value already exact at the target scale is returned unchanged. The rule is applied on the number line (toward positive infinity for "up"), so negative amounts are unbiased in the same way as positive amounts.

Randomness is drawn from an injected sampler returning a value in [0, 1); the parameterless constructor uses Shared. Because the result depends on the sampler, this strategy is inherently non-deterministic and two roundings of the same input may differ - supply a fixed-sequence sampler to make a test deterministic. Unlike MidpointRoundingStrategy, this convention cannot be expressed as a MidpointRounding mode, which is why it is a distinct implementation of the IRoundingStrategy seam rather than another mode of the midpoint strategy.

// Unbiased rounding for a stochastic allocation policy.
var context = new MonetaryContext { Rounding = StochasticRoundingStrategy.Shared };

// Deterministic in a test: a sampler that always draws 0 rounds every non-exact value up.
var up = new StochasticRoundingStrategy(() => 0.0);
decimal r = up.Round(1.001m, 2); // 1.01

Constructors