{
  "docId": "019dd923-5ca1-7489-b634-a2709765f07b",
  "docSlug": "62c9b2539bf1585b",
  "documentTitle": "Whats Next Insurance Pricing",
  "authorId": "BCG",
  "authorName": "BCG",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "strategy_consulting",
  "sourceTypeLabel": "Strategy consulting",
  "presentationDate": null,
  "orientation": "portrait",
  "aspectRatio": 0.773,
  "pageNumber": 6,
  "pageCount": 13,
  "prevPage": 5,
  "nextPage": 7,
  "slideType": "industry_trends",
  "function": "diagnose",
  "density": "dense",
  "nDataPoints": 0,
  "notes": "Illustrates the bias-variance tradeoff in machine learning models.",
  "elementsJson": [
    "line_chart",
    "paragraph",
    "footnote"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5ca1-7489-b634-a2709765f07b/6",
  "deckHref": "/decks/019dd923-5ca1-7489-b634-a2709765f07b",
  "deckJsonHref": "/decks/019dd923-5ca1-7489-b634-a2709765f07b.json",
  "deckAnchorHref": "/decks/019dd923-5ca1-7489-b634-a2709765f07b#slide-6",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "Diminishing returns for adding variables... and those returns may only be illusory",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd951-acb8-762a-a73e-82d52b34f138",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.35,
        "w": 0.85,
        "x": 0.07,
        "y": 0.14
      },
      "kind": "chart",
      "text": "Performance vs Complexity chart showing Training vs Actual performance curves",
      "attrs": null,
      "subkind": "line",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "32c7367b-1598-4845-b277-7ce380aa2897",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Model performance",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd951-acb8-762a-a73e-861db0393c1b",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.25,
        "w": 0.85,
        "x": 0.07,
        "y": 0.58
      },
      "kind": "paragraph",
      "text": "Recent advances in predictive modeling, as well as the propagation of machine learning techniques, are improving risk modeling capabilities, enabling insurers to more accurately segment risk and predict future losses. Insurers are increasingly trying to employ GBMs (gradient boosting machines) over GLMs (generalized linear models), because the former does not assume a linear relationship between variables and can better identify interactive effects through the if-then structure of the insurers’ models. However, GBMs’ lower transparency presents an obstacle for regulatory acceptance, leading to some instances where an insurer will build a GLM on top of the output of a GBM.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "11c1e61d-a091-416d-ac80-a202ef14bccf",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.03,
        "w": 0.85,
        "x": 0.07,
        "y": 0.52
      },
      "kind": "source-note",
      "text": "Note: Common illustrative example of overfitting vs underfitting. Techniques can be employed to help avoid performance degradation",
      "attrs": null,
      "subkind": null,
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "d00c9fc7-87c1-416d-9e31-c169905c726f",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.85,
        "x": 0.07,
        "y": 0.07
      },
      "kind": "title",
      "text": "Exhibit 2 - After a certain point, insurers see diminishing returns to model performance with increasing number of variables",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "beb9230e-d417-4de8-ace7-444a1c335724",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-163f-72bd-8a0c-79bb757aac3e",
      "evidence": "Title states the insight: 'diminishing returns to model performance'",
      "confidence": 90
    },
    {
      "name": "Annotation",
      "slug": "annotation",
      "agent": "Designer",
      "layer": "slide",
      "matchId": "019dd95a-163f-72bd-8a0c-7fe92f2168da",
      "evidence": "Two callouts label exact spots: 'Diminishing returns...' and '...returns may only be illusory'",
      "confidence": 90
    },
    {
      "name": "Chart Selection Guide",
      "slug": "chart-selection-guide",
      "agent": "Designer",
      "layer": "slide",
      "matchId": "a4d39810-907f-467d-adae-73c4435baca8",
      "evidence": "Performance vs Complexity chart showing Training vs Actual performance curves",
      "confidence": 0.7
    },
    {
      "name": "Data-Ink Ratio",
      "slug": "data-ink-ratio",
      "agent": "Designer",
      "layer": "slide",
      "matchId": "019dd95a-163f-72bd-8a0c-87ee7bb9d450",
      "evidence": "Minimal axes, no gridlines, two clean lines with direct labels",
      "confidence": 70
    },
    {
      "name": "Unexpected Pattern",
      "slug": "unexpected-pattern",
      "agent": "Storyteller",
      "layer": "slide",
      "matchId": "019dd95a-163f-72bd-8a0c-82fef11d4c9d",
      "evidence": "Actual curve flattens then declines while Training keeps rising - counterintuitive",
      "confidence": 78
    },
    {
      "name": "Von Restorff Effect",
      "slug": "von-restorff-effect",
      "agent": "Designer",
      "layer": "slide",
      "matchId": "019dd95a-163f-72bd-8a0c-8a5ca3e00ab1",
      "evidence": "Red dotted line and blue callout stand out against green/black palette",
      "confidence": 65
    }
  ],
  "frameworks": [
    {
      "name": "overfitting-vs-underfitting",
      "slug": null,
      "matchId": "e1f9ea3c-744e-4d2b-96e3-62ee40f58989",
      "evidence": "Chart illustrates the divergence between training and actual performance as complexity increases.",
      "confidence": 0.9
    }
  ],
  "arcBeats": [
    {
      "to": 6,
      "from": 5,
      "beatId": "019dd95a-0702-74a3-87e4-099544b2493d",
      "arcName": "The Consultant's Gambit",
      "arcSlug": "consultants-gambit",
      "beatName": "Problem & Complication",
      "beatSlug": "consultants-gambit-problem-complication",
      "evidence": "Exhibit 1 siloed process + Exhibit 2 diminishing returns curve",
      "position": 2,
      "confidence": 85,
      "parentBeatName": "Complication",
      "parentBeatSlug": "complication"
    }
  ],
  "loops": [
    {
      "to": 7,
      "from": 5,
      "name": "Zoom In",
      "slug": "06-zoom-in",
      "bestFor": "Technical deep-dives, case studies, detailed analysis",
      "matchId": "019dd95a-088c-724c-b310-03aff0e780a5",
      "evidence": "Exhibit 1 process map -> Exhibit 2 modeling detail -> Exhibit 3 implication of GBM transition",
      "position": 2,
      "objective": "Drill from process big picture into modeling detail and its implication",
      "structure": "The Big Picture -> Key Area of Focus -> Specific Detail -> Implication",
      "confidence": 78,
      "description": "Start broad, then progressively focus on specific details that prove your point"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}