Cross-Model Conflict and Double-Counting Resolver

Install and import#

bash

npm install fintech-algorithms

import { crossModelConflictAndDoubleCountingResolver } from "fintech-algorithms/fundamental-analysis-and-valuation/integrated-equity-scoring/cross-model-conflict-and-double-counting-resolver";

Signature#

crossModelConflictAndDoubleCountingResolver(data)

Groups scored components by the evidence they share, caps each group's weight so correlated evidence cannot be counted twice, and flags groups whose members disagree by more than the conflict threshold.

Parameters#

Name	Type	Notes
`data`	`{ group_cap: number; conflict_threshold: number; components: Array<{ name?: string; evidence_group: string; score: number; weight: number }> }`	`group_cap` is the largest weight any one evidence group may carry and must be greater than 0 and at most 1. `conflict_threshold` is the within-group score range that marks a conflict and must be greater than 0 and at most 100. Each component needs a text `evidence_group`, a `score` between 0 and 100 and a positive `weight`; `name` falls back to `component`.

Returns#

{ state: string; method: string; raw_score: number; resolved_score: number; double_counting_adjustment: number; groups: Array<{ group: string; raw_weight: number; capped_weight: number; mean_score: number; range: number; conflict: boolean }>; conflict_groups: string[]; conflict_count: number; group_cap: number; conflict_threshold: number }

raw_score is the weighted mean over all components as supplied and resolved_score the weighted mean of group means under the capped weights, with double_counting_adjustment their difference. groups carries one row per evidence group sorted by name, conflict_groups and conflict_count the groups whose range reaches the threshold, and state is conflict-detected when any exists, otherwise resolved.

Errors#

When components is not a nonempty list, or a component lacks a text evidence_group — throws TypeError
When group_cap or conflict_threshold is not a finite number — throws TypeError
When group_cap is not in (0, 1], or conflict_threshold is not in (0, 100] — throws RangeError
When a component score is below 0 or above 100, or its weight is not positive — throws RangeError

Complexity: time O(n log n), space O(n).

Worked example#

executed Captured by running this function on the input its own test provides. Real output of real code — but not asserted against a published figure.

Input#

data

{
  "group_cap": 0.4,
  "conflict_threshold": 30,
  "components": [
    {
      "name": "altman_liquidity",
      "score": 82,
      "weight": 0.3,
      "evidence_group": "liquidity"
    },
    {
      "name": "current_ratio",
      "score": 76,
      "weight": 0.2,
      "evidence_group": "liquidity"
    },
    {
      "name": "piotroski_quality",
      "score": 70,
      "weight": 0.25,
      "evidence_group": "quality"
    }
  ]
}

Call#

crossModelConflictAndDoubleCountingResolver(data)

Returns#

object with 10 fields: state, method, raw_score, resolved_score, double_counting_adjustment, groups, conflict_groups, conflict_count, …

{
  "state": "resolved",
  "method": "evidence-group-cap-and-conflict-resolver",
  "raw_score": 69.45833333333333,
  "resolved_score": 69.64,
  "double_counting_adjustment": 0.18166666666667197,
  "groups": [
    {
      "group": "liquidity",
      "raw_weight": 0.5,
      "capped_weight": 0.4,
      "mean_score": 79.6,
      "range": 6,
      "conflict": false
    },
    {
      "group": "quality",
      "raw_weight": 0.5,
      "capped_weight": 0.4,
      "mean_score": 57.5,
      "range": 25,
      "conflict": false
    },
    {
      "group": "resilience",
      "raw_weight": 0.2,
      "capped_weight": 0.2,
      "mean_score": 74,
      "range": 0,
      "conflict": false
    }
  ],
  "conflict_groups": [],
  "conflict_count": 0,
  "group_cap": 0.4,
  "conflict_threshold": 30
}

Other exports#

This module also exports calculate, pointInTimeStockScoringInputAssembly, stockScoringPeerCohortResolver, fundamentalMetricDirectionAndPeerNormalization, modelApplicabilityAndVariantRouter, accountingFinancialHealthComposite, earningsQualityComposite, dividendSafetyScore, balanceSheetResilienceScore, distressModelEnsemble, overallExplainableStockScore, scoreConfidenceMissingDataPenaltyAndAbstention, marketWideStockScreeningAndRanking, stockScoreHistoryMigrationAndChangeAttribution. Every module additionally exports run as an alias of its primary function, and a meta object carrying its catalog id, domain, family, shape and article URL.