Expand description
Additive, report-independent provider evaluation.
The module accepts normalized PositionOutcome rows instead of depending on
crate::report::TradeResult or crate::report::PositionSummary. A current
integration can map one PositionSummary to one outcome (net_pnl, symbol,
side, group, close reasons, and final-close ordering); future report versions
can additionally supply R, excursions, lifecycle counters, and execution
diagnostics without changing this evaluator.
Filtering is exact and case-sensitive. Values within a filter dimension are ORed, while different dimensions (including each tag key) are ANDed. Breakdowns and rolling order are deterministic. No overall score, rank, or rating is produced.
Structs§
- Bootstrap
Config - Configuration for deterministic bootstrap confidence intervals.
- Breakdown
Bucket - Breakdown
RowSummary - Visibility into minimum-count filtering and global breakdown row truncation.
- Confidence
Interval - Cost
Metrics Section - Commission and swap actually charged across the selected positions.
- Coverage
Section - CumulativeR
Point - One chronologically ordered point on the cumulative realized-R curve.
- Evaluation
Breakdown - Evaluation
Context - Provider and source identifiers attached to an evaluation without changing the normalized position rows.
- Evaluation
Options - Typed report configuration, deliberately separate from normalized position and lifecycle inputs.
- Evaluation
Position Rows - Filtered normalized position rows included for metric reconciliation.
- Evaluation
Report - Provider-evaluation result. It intentionally has no aggregate score, rank, or rating; consumers decide which individual sections matter for their use case.
- Evaluation
Request - Complete input to
super::evaluate.optionsis flattened so payloads produced beforeEvaluationOptionswas introduced retain the same serde shape. - Excursion
Input - R-normalized maximum favorable and adverse excursion values.
- Excursion
Metrics Section - Execution
Diagnostics Input - Optional per-position execution observations.
- Execution
Diagnostics Section - Intrinsic
Robustness Section - Lifecycle
Counts - Aggregate lifecycle counters supplied by a provider integration.
- Metric
Value - A metric accompanied by an explicit availability status.
- PnlConcentration
Section - Shares of gross positive completed-position P&L contributed by the top N winners.
- Position
Costs Input - Per-position commission and swap already applied to
outcome. - Position
Dimensions - Dimensions used for filtering and deterministic breakdowns.
- Position
Filter - Typed position filter.
- Position
Outcome - Generic completed-position input for provider evaluation.
- Position
Performance Section - RMetrics
Section - RQuantiles
- Deterministic type-7 quantiles of finite realized-R observations.
- Removal
Impact - Rolling
Outcome - Rolling
Outcomes - Source
Coverage Counts - Parser/source coverage supplied by an integration that can observe raw input.
Enums§
- Breakdown
Dimension - A requested categorical breakdown. Duplicate requests are evaluated once.
- Breakdown
Value - Typed and sortable breakdown key.
- Evaluation
Section - Independently selectable provider-report sections.
- Group
Filter - Selects grouped positions, including positions that have no group.
- Metric
Status - Describes whether a metric can be interpreted by a consumer.
- Outcome
Classification - Provider-supplied classification of a completed-position outcome.
- Position
Side - Normalized position direction, independent of the execution engine’s side type.
Functions§
- bootstrap_
mean_ confidence - Computes a deterministic percentile-bootstrap confidence interval for a mean.
- evaluate
- Evaluates normalized provider outcomes without coupling to a backtest report.
- wilson_
interval - Computes a two-sided Wilson score interval for a binomial proportion.