Methodology page flow vs Open Core engine

Source path in repo: docs/examples/10-methodology-flow-and-engine.md. [View on GitHub]

Mirrored from the engine repository. Prefer reading here for a consistent experience; use GitHub for file history and blame.

The hosted Kiploks methodology page (product UI) shows how analysis blocks connect in a fixed order. That order matches how a full report is assembled from payload data. The Open Core repository ships deterministic math for those blocks in @kiploks/engine-core, but the OSS examples (0107, CLI) mostly cover small entrypoints (analyze(), WFA JSON). They do not include a separate runnable script per block.

Same pipeline as on /methodology (product)

Data Quality Guard
  → Walk-Forward
  → Benchmark Metrics
  → Benchmark Comparison
  → Parameter Sensitivity
  → Cost Drag
  → Action Plan
  → Risk Metrics
  → Robustness Score
  → Final Verdict

Where each step is implemented (engine core)

Methodology block (product name)Typical role in TestResultDataPrimary Open Core modules (under core/src/)
Data Quality GuarddataQualityGuardResultdataQualityGuard.ts
Walk-ForwardwalkForwardAnalysis, WFE / WFA transformswfaProfessional.ts, wfa/, buildDiagnosticsFromWfa.ts, related
Benchmark MetricsproBenchmarkMetricsproBenchmarkMetrics.ts, benchmarkCore.ts, canonicalMetrics.ts
Benchmark ComparisonbenchmarkComparisonbenchmarkFromEquity.ts, tryBuildBenchmarkComparisonFromEquityPath, benchmarkKlinesResolver.ts (market data inputs are supplied by the host)
Parameter SensitivityparameterSensitivity / WFA diagnosticsparameterSensitivityContract.ts, buildDiagnosticsFromWfa.ts, analysis helpers
Cost DragturnoverAndCostDragturnoverAndCostDrag.ts, turnoverCore.ts
Action Planstrategy action plan payloadsstrategyActionPlanPrecomputed.ts
Risk MetricsriskAnalysisriskCore.ts, riskNarratives.ts
Robustness ScorerobustnessScorerobustnessScoreFromWfa.ts (and inputs from other blocks)
Final VerdictverdictPayload / summaryfinalVerdictEngine.ts, summaryBlockEngine.ts, decisionArtifacts.ts, killSwitch.ts

Assembly of the full object is buildTestResultDataFromUnified.ts plus mapPayloadToUnified.ts (types in @kiploks/engine-contracts). The host orchestrates I/O (exchange data, storage, API). Orchestration details are outside this repo.

What the OSS examples cover

ExampleTouches methodology blocks
[01-minimal-analyze.md]None of the full report (summary + metadata only).
[02-wfa-from-trades.md], [03-wfa-from-windows.md]Walk-Forward slice of the pipeline (public WFA / WFE JSON); not full DQG/benchmark/verdict.
[04-csv-to-trades.md]Input only (trades → engine).
[0507]CLI and validation; same partial surface as above.
[08-result-shape-and-kiploks-ui.md], [result-layout-demo.html]Conceptual mapping + static demo; not full TestResultData.
[09-full-report-vs-public-wfa.md]Explains why public WFA JSON has available: false on many blocks.

There is no separate example file per methodology section by design: reproducing the full chain requires a unified integration payload and a host that calls buildTestResultDataFromUnified; that wiring is not duplicated as ten runnable demos in Open Core.

If you need the full pipeline

  1. Read [OPEN_CORE_LOCAL_USER_GUIDE.md] (scope of Open Core vs full report).
  2. Use the Kiploks methodology documentation (public site) for definitions of each block; use this table to jump to engine source for formulas.

[Back to documentation index]