[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-327327-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-327327-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","meta-benchmarks-for-financial-services-llm-evaluation","Meta-Benchmarks for Financial-Services LLM Evaluation","","Public LLM leaderboards optimise for global average performance and miss the cognitive demands of financial-services work, where strong results on general tasks can mask weaknesses in document-grounded compliance reasoning or multi-turn customer interactions. The paper presents a meta-benchmarking framework that organizes 452 publicly reported benchmarks into a 41 work-activity taxonomy and aggregates them into 38 BIAN banking business domains. A multiplicative weighting scheme (discrimination, coverage, recency) and an Elo-based aggregation produce comparable work-activity and domain capability scores, demonstrated on a June 2026 snapshot of 288 models across 25 organizations.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/meta-benchmarks-for-financial-services-llm-evaluation/327327/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/meta-benchmarks-for-financial-services-llm-evaluation/327327.png","ImageObject",300,407,{"name":42,"@type":43},"Emma Wilson","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-11-02","2026-09-21",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",11,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"Why are public LLM leaderboards insufficient for regulated financial-services use?","Question",{"text":63,"@type":64},"Public leaderboards blend performance across broad tasks, but financial-services needs require more specific capabilities. A model can lead on general benchmarks while still underperforming on document-grounded compliance reasoning or multi-turn customer interaction tasks.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"What does the meta-benchmarking framework aggregate and how?",{"text":68,"@type":64},"It aggregates 452 publicly reported benchmarks into 41 O*NET generalized work activities, then further aggregates those into 38 BIAN banking business domains spanning sales, operations, risk, and support work.",{"name":70,"@type":61,"acceptedAnswer":71},"How does the framework weight benchmarks when computing capability scores?",{"text":72,"@type":64},"It uses a multiplicative weighting scheme based on discrimination, coverage, and recency over a rolling model window. The weights reward benchmarks actively contested at the frontier and suppress saturated or rarely attempted legacy tests.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},327327,1790648773,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":140,"language":141,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":12,"update_tm":145,"read_time":146,"preview_page_count":4},3848291630094,"https://eur-avatar.wpscdn.com/davatar_085a072bc5b1113ac321206ff7593b45","arXiv :2607 .0 1740v 1 [ cs .AI] 2 Jul 2026  \nMeta-Benchmarks for Financial-Services LLM Evaluation  \nBlair Hudson  \nCommonwealth Bank of Australia  \n[blair@cba.com.au](blair@cba.com.au)  \nJuly 2026  \nAbstract: Public LLM leaderboards optimise for global average performance and do not capture the specific cognitive demands of financial-services work: a model that leads on MMLU-Pro may underperform on documentgrounded compliance reasoning, and a coding leader may handle multi-turn customer interactions poorly. We present a meta-benchmarking framework that organises 452 publicly reported benchmarks into 41 O*NET Generalized Work Activities and aggregates those into 38 BIAN banking business domains spanning sales, operations, risk, and support work. A multiplicative weighting scheme (discrimination × coverage × recency), computed over a rolling model window, rewards benchmarks that still separate the best models, are widely reported, and remain in active use, suppressing saturated legacy tests automatically. These weights scale the K-factor in a pairwise Elo tournament, producing cross-benchmark-comparable work-activity scores without raw score normalisation; businessdomain scores are weighted averages of the constituent work-activity Elos. We demonstrate the framework on a point-in-time public snapshot covering 288 models across 25 organisations as of June 2026, and describe the methodology, full taxonomy, design decisions, and limitations with the aim of making the approach reproducible for institutions facing similar selection and governance challenges.  \nKeywords: LLM evaluation, benchmark aggregation, financial services, work-activity taxonomy, capability scoring, metabenchmarking  \n1 Introduction  \nPublic LLM leaderboards present aggregate scores that blend performance across tasks ranging from undergraduate trivia to competitive programming. For an organisation selecting and governing models for specific operational purposes, these aggregates are insufficient: a model that leads on MMLU-Pro may underperform on document-grounded compliance reasoning; a coding leader may score poorly on multi-turn customer interaction tasks. Selecting and governing AI models for regulated industries requires a more precise tool.  \nWe describe a meta-benchmarking framework developed in a financial-services research context to address this gap. It introduces no new benchmarks. Instead, it aggregates 452 existing publicly reported benchmarks into 41 work activities drawn from the O*NET Generalized Work Activities taxonomy (National Center for O*NET Development 2024), and further aggregates those into 38 banking business domains drawn from the BIAN Service Landscape 14.0.0 (Banking Industry Architecture Network 2024) . A dynamic weighting scheme rewards benchmarks that are actively contested at the frontier and suppresses those that have been saturated or are rarely attempted by recent models.  \nThe contribution is fourfold. First, we articulate a principled taxonomy mapping benchmarks to standardised work activities and banking business domains, grounded in established external standards (O*NET and BIAN) . Second, we describe a weighting and normalisation methodology that handles the heterogeneous and rapidly evolving landscape of public benchmarks without requiring manual score curation. Third, we  \n452  \n[LLM-Stats.com](LLM-Stats.com)[ ](LLM-Stats.com)Public Benchmarks  \nFigure 1: The evaluation pyramid. Reading bottomup, 288+ models are scored on 452 public benchmarks, which are mapped to 41 ONET Generalized Work Activities, aggregated into 38 BIAN business domains, and grouped under five BIAN Business Areas.  \ndescribe practical applications of the resulting capability profiles in preliminary model comparison, riskinformed screening, and governance research. Fourth, we provide sufficient methodological detail to make the taxonomy and aggregation approach reproducible for institutions facing similar model-selection and governance challenges.","cbCaifm6RXQLW0lq","https://ap.wps.com/l/cbCaifm6RXQLW0lq","pdf",864523,27,"English","# Introduction\n## Why not evaluate on internal benchmarks?","[{\"question\":\"Why are public LLM leaderboards insufficient for regulated financial-services use?\",\"answer\":\"Public leaderboards blend performance across broad tasks, but financial-services needs require more specific capabilities. A model can lead on general benchmarks while still underperforming on document-grounded compliance reasoning or multi-turn customer interaction tasks.\"},{\"question\":\"What does the meta-benchmarking framework aggregate and how?\",\"answer\":\"It aggregates 452 publicly reported benchmarks into 41 O*NET generalized work activities, then further aggregates those into 38 BIAN banking business domains spanning sales, operations, risk, and support work.\"},{\"question\":\"How does the framework weight benchmarks when computing capability scores?\",\"answer\":\"It uses a multiplicative weighting scheme based on discrimination, coverage, and recency over a rolling model window. The weights reward benchmarks actively contested at the frontier and suppress saturated or rarely attempted legacy tests.\"}]","Meta-Benchmarks for Financial-Services LLM Evaluation | PDF",1789976985,68]