{"schema_version":1,"research_id":"eoa_inverse_innovation_exp04_external_evaluation_20260802","source_assessment_id":"invariant_mode_decomposition_design__economics_finance:PROPOSAL_FIRST:v0","cell_id":"invariant_mode_decomposition_design__economics_finance","search_queries":["site:bis.org Basel Committee intraday liquidity monitoring tools payment settlement liquidity stress indicators","site:bankofengland.co.uk systemic liquidity risk monitoring network model banks payment data stress","site:ecb.europa.eu bank liquidity stress testing network contagion payment systems model","spectral early warning bank liquidity stress eigenvalue network transition model research","site:occ.treas.gov liquidity handbook early warning indicators banks supervision funding risk 2024","site:federalreserve.gov bank supervision liquidity risk monitoring early warning indicators payment data","site:bis.org BCBS 239 risk data aggregation principles supervisors data lineage accuracy timeliness liquidity","site:imf.org systemic liquidity stress testing network banks payment flows early warning","bank liquidity early warning principal components dynamic factor model supervisory indicators paper","financial systemic risk dynamic mode decomposition eigenvalues bank liquidity network early warning","bank liquidity VAR eigenvalue stability transition matrix systemic risk monitoring","site:federalreserve.gov SR 11-7 model risk management validation supervisory guidance","site:bls.gov/oes/2024/may data scientists annual mean wage economists operations research analysts","site:bls.gov/ooh math data scientists median pay 2024","site:bls.gov/oes/current oes152051 data scientists","\"Evaluation of systemic risk in a financial system using dynamic mode decomposition\"","\"dynamic mode decomposition\" \"systemic risk\" financial system Zavialov Ikeda"],"sources":[{"source_id":"S1","title":"Monitoring tools for intraday liquidity management","publisher":"Basel Committee on Banking Supervision, Bank for International Settlements","url":"https://www.bis.org/publ/bcbs248.htm","source_class":"STANDARD","publication_date":"2013-04-11","accessed_at":"2026-08-02","claims_supported":["Intraday liquidity is a recognized supervisory concern tied to timely payment and settlement obligations.","The Basel framework supplies seven quantitative monitoring tools, stress scenarios, application issues, and a reporting regime.","The tools are explicitly for supervisory monitoring and complement ordinary liquidity standards, establishing a relevant baseline but not a modal method."]},{"source_id":"S2","title":"Section 252.34—Liquidity Risk-Management Requirements","publisher":"Board of Governors of the Federal Reserve System","url":"https://www.federalreserve.gov/frrs/regulations/section-25234-liquidity-risk-management-requirements.htm","source_class":"GOVERNMENT_OR_REGULATOR","publication_date":"Current regulation; page does not state one consolidated publication date","accessed_at":"2026-08-02","claims_supported":["Covered bank holding companies must monitor emerging liquidity stress and maintain tailored early-warning indicators.","Requirements cover funding concentrations, collateral, legal entities, currencies, business lines, intraday inflows and outflows, and payment-system obligations.","Senior management, boards, risk committees, and independent review functions have defined liquidity-risk responsibilities, demonstrating identifiable authorizers and workflows."]},{"source_id":"S3","title":"Systemic illiquidity in the interbank network","publisher":"Bank of England","url":"https://www.bankofengland.co.uk/working-paper/2016/systemic-illiquidity-in-the-interbank-network","source_class":"PRIMARY_RESEARCH","publication_date":"2016-04-08","accessed_at":"2026-08-02","claims_supported":["Systemic illiquidity can propagate over multiple days through an interbank funding network.","The study used bank daily cash flows, short-term interbank funding, and liquid-asset buffers, supporting the feasibility and relevance of coupled cross-bank state data.","Network-based simulation and importance-weighted macroprudential policy are close prior art to system-level liquidity analysis."]},{"source_id":"S4","title":"Integrating contagion risk into the 2025 EU-wide stress test: a system-wide analysis with amplification effects between banks and non-banks","publisher":"European Central Bank","url":"https://www.ecb.europa.eu/press/financial-stability-publications/macroprudential-bulletin/html/ecb.mpbu202511_03.en.html","source_class":"GOVERNMENT_OR_REGULATOR","publication_date":"2025-11","accessed_at":"2026-08-02","claims_supported":["An identifiable supervisor already operates a granular system-wide network stress-testing tool using supervisory and commercial data.","The ECB reports that modeled second-round effects added 29 basis points of average CET1 depletion, increasing first-round effects by 12%, showing that interconnected amplification can matter.","ISA models iterative liquidity redistribution, fire sales, liquidity shortfalls, and defaults across 96 banks and other financial institutions, constituting close workflow and technical prior art."]},{"source_id":"S5","title":"Systemwide Liquidity Stress Testing Tool","publisher":"International Monetary Fund","url":"https://www.imf.org/en/publications/wp/issues/2022/12/16/systemwide-liquidity-stress-testing-tool-527046","source_class":"PRIMARY_RESEARCH","publication_date":"2022-12-16","accessed_at":"2026-08-02","claims_supported":["System-wide liquidity stress testing is an established practice with tools that map counterparty exposures and spillovers across sectors.","The IMF identifies data constraints and hard-to-model behavior as central implementation limitations and states that no uniformly accepted systemic-liquidity model exists.","A simple balance-sheet-matrix tool can identify important financial linkages and inform macroprudential or liquidity-support discussions, providing a lower-complexity comparator."]},{"source_id":"S6","title":"Using dynamic mode decomposition to extract cyclic behavior in the stock market","publisher":"Physica A / Elsevier; author-uploaded full text hosted by ResearchGate","url":"https://www.researchgate.net/publication/288917827_Using_dynamic_mode_decomposition_to_extract_cyclic_behavior_in_the_stock_market","source_class":"PRIMARY_RESEARCH","publication_date":"2015-12","accessed_at":"2026-08-02","claims_supported":["Dynamic mode decomposition and Koopman spectral analysis have already been applied to high-dimensional financial data.","Financial modes can be associated with complex eigenvalues representing growth rates and frequencies.","The paper tests reproducibility across data partitions and shows that many apparent modes are not robust, directly supporting the candidate's resampling and mode-stability concerns."]},{"source_id":"S7","title":"Supervisory Guidance on Model Risk Management","publisher":"Board of Governors of the Federal Reserve System, Office of the Comptroller of the Currency, and Federal Deposit Insurance Corporation","url":"https://www.federalreserve.gov/frrs/guidance/supervisory-guidance-on-model-risk-management.htm","source_class":"OFFICIAL_GUIDANCE","publication_date":"2026-04-17","accessed_at":"2026-08-02","claims_supported":["Traditional statistical and quantitative models used for financial-risk purposes require risk-based governance commensurate with model materiality.","Sound practice includes conceptual review, outcome analysis, back-testing, ongoing monitoring, documentation, defined roles, and objective effective challenge.","Model limitations, performance deterioration, misuse, common dependencies, and reliance on proprietary products require explicit controls and complementary analysis."]},{"source_id":"S8","title":"Math Occupations","publisher":"U.S. Bureau of Labor Statistics","url":"https://www.bls.gov/ooh/math/home.htm","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"2025-08-28","accessed_at":"2026-08-02","claims_supported":["May 2024 median annual pay was $104,620 for mathematical occupations, $112,590 for data scientists, $104,350 for mathematicians and statisticians, $91,290 for operations-research analysts, and $125,770 for actuaries.","These wage observations provide a public labor-cost anchor for broad resource-equivalent estimates, although they exclude benefits, supervisory specialization, secure infrastructure, and procurement overhead."]}],"problem_evidence":{"support":"MODERATE","rationale":"Official standards and regulation recognize emerging liquidity stress, intraday payment obligations, collateral constraints, funding concentrations, and the need for early-warning monitoring. Bank of England and ECB work demonstrates propagation and measurable second-round amplification through interconnections. However, the stronger proposal-specific assertion—that ordinary institution thresholds and averages systematically miss a stable, weakly damped cross-bank transition mode—is not directly demonstrated by the opened evidence.","source_ids":["S1","S2","S3","S4","S5"]},"stakeholder_evidence":{"support":"STRONG","rationale":"Banking supervisors, central banks, bank boards, risk committees, and independent review functions are identifiable authorizers. Federal regulation expressly requires early-warning indicators and multidimensional liquidity monitoring, BCBS supplies supervisory tools, and the ECB already operates a granular system-wide contagion model. None specifically requests eigendecomposition-based prioritization, so pull is strong for the problem and workflow but only inferential for this particular method.","source_ids":["S1","S2","S4","S7"]},"prior_art":{"proximity":"ADJACENT_PRIOR_ART","closest_analogues":[{"name":"BCBS intraday-liquidity monitoring tools","similarity":"Uses multiple liquidity and payment indicators, stress scenarios, and supervisory reporting to detect vulnerabilities.","remaining_difference":"It combines prescribed indicator information and scenarios rather than estimating a cross-bank transition operator and ranking its modes by gain and consequence sensitivity.","source_ids":["S1"]},{"name":"Bank of England systemic-illiquidity network model","similarity":"Uses daily bank cash-flow, funding, and liquid-buffer data to model multi-day propagation of systemic illiquidity.","remaining_difference":"It is a calibrated contagion simulation and network-importance analysis, not a shadow alert system based on empirically estimated transition eigenmodes with spectral-gap and residual gates.","source_ids":["S3"]},{"name":"ECB Interconnected System-wide stress test Analytics","similarity":"Uses granular entity-level exposures and iterative liquidity and loss propagation to reveal second-round effects missed by sector-by-sector analysis.","remaining_difference":"It propagates specified adverse scenarios through a structural network model; it does not discover locally growing observed-state modes or compare mode-prioritized alerts against ordinary early-warning dashboards.","source_ids":["S4"]},{"name":"IMF Systemwide Liquidity Stress Testing Tool","similarity":"Maps exposure matrices, liquidity shortfalls, and cross-sector spillovers to support macroprudential decisions.","remaining_difference":"It is an aggregate, scenario-based balance-sheet tool rather than an institution-variable transition decomposition or mode-drift monitor.","source_ids":["S5"]},{"name":"Financial dynamic mode decomposition and Koopman-mode analysis","similarity":"Applies DMD to high-dimensional financial time series, interprets eigenvalues as modal growth and frequency, and tests mode reproducibility across subsamples.","remaining_difference":"The opened study concerns stock-market cycles, not confidential supervisory liquidity variables, payment disruption, sensitivity-ranked review actions, or governed comparison with liquidity-monitoring baselines.","source_ids":["S6"]}],"distinctive_claim_remaining":"On a preregistered, held-out supervisory history and at the same false-alert budget, prioritizing reviews by stable, traceable cross-bank transition modes using both modal gain and consequence sensitivity will produce earlier or uniquely correct liquidity-escalation signals than ordinary thresholds, PCA/common-factor monitoring, exposure centrality, and the same transition model used only for direct multivariate forecasting; the claimed advantage must disappear when conditioning, spectral-gap, drift, or residual gates fail.","confidence":"MODERATE"},"implementation_evidence":{"support":"MODERATE","rationale":"The required linear algebra and financial DMD methods are technically established, and official institutions demonstrate that granular network stress models and relevant cash-flow data can exist. A retrospective shadow workflow avoids direct supervisory action. Feasibility is nevertheless limited by confidential data access, missingness and scaling choices, short stress histories, regime changes, behavioral endogeneity, non-normal or ill-conditioned operators, mode identity instability, and the need for independent validation and documented governance. No opened source demonstrates this complete dashboard on supervisory liquidity data.","source_ids":["S3","S4","S5","S6","S7"]},"scores":{"meaningful_impact":{"score":4,"rationale":"Liquidity failures can disrupt payment obligations and propagate across institutions; official ECB modeling finds material second-round amplification. The candidate addresses a high-consequence monitoring problem, although its realized impact is unmeasured.","source_ids":["S1","S2","S3","S4"]},"stakeholder_pull":{"score":4,"rationale":"Regulators mandate early-warning and multidimensional liquidity monitoring, and the ECB already uses system-wide network analytics. Pull for an additional modal layer is not explicit.","source_ids":["S1","S2","S4"]},"incremental_advantage":{"score":3,"rationale":"Gain-ranked transition modes could reveal dynamics not captured by indicator magnitude, variance, or static network prominence, but no held-out comparison shows earlier or more accurate escalation.","source_ids":["S3","S4","S5","S6"]},"distinctiveness_plausibility":{"score":2,"rationale":"Network liquidity stress testing and financial dynamic-mode analysis both pre-exist. The governed combination and specific comparator claim may remain distinctive, but world novelty and proprietary deployments were not assessed.","source_ids":["S3","S4","S5","S6"]},"technical_implementability":{"score":3,"rationale":"Decomposition, resampling, reconstruction, and monitoring are implementable with standard quantitative methods; the difficult parts are valid state construction, adequate temporal samples, non-normality, regime stability, and confidential-data engineering.","source_ids":["S3","S5","S6","S7"]},"adoption_authority_feasibility":{"score":3,"rationale":"A supervisor can plausibly authorize a nonbinding retrospective shadow study using data already held, and relevant governance roles exist. No named partner, data custodian, jurisdiction-specific authorization, or review approval has been secured.","source_ids":["S1","S2","S4","S7"]},"evidence_readiness":{"score":3,"rationale":"A bounded retrospective design with objective comparators and falsifiers is available, but decisive evidence requires proprietary supervisory data, rare-event labels, and independent validation.","source_ids":["S3","S4","S5","S7"]},"safety_net_benefit":{"score":4,"rationale":"Keeping ordinary monitoring in parallel, prohibiting automatic enforcement, limiting outputs to shadow review, and halting on model-validity failures substantially bound downside. Confidentiality and stigmatization risks still require formal controls.","source_ids":["S2","S7"]},"scalability":{"score":3,"rationale":"The computations are scalable, and official network models cover many entities, but calibration, data lineage, legal permissions, stress labels, and regimes are jurisdiction-specific and may dominate replication cost.","source_ids":["S4","S5","S7"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"Preregister and execute one access-controlled retrospective study on an already assembled frozen dataset, including data-quality review, transition estimation, modal diagnostics, four comparator models, held-out scoring, and an audit report.","confidence":"MODERATE","assumptions":["Approximately 0.5-1.2 FTE-years split among a liquidity-domain analyst, quantitative researcher, data scientist, and independent reviewer.","Existing authorized data, secure workspace, and outcome labels are available; acquiring or reconstructing them is excluded.","BLS mathematical-occupation wages are used as the salary anchor, with additional allowance for benefits, specialization, and secure-environment overhead.","No live alerts, bank contacts, procurement, or production integration."],"source_ids":["S7","S8"]},"initial_deployment_startup":{"band_2026_usd":"1M_TO_5M","scope":"Build a production-quality secure data pipeline and dashboard; establish lineage and access controls; implement monitoring, audit logs, model inventory, independent validation, documentation, and integration with the existing supervisory dashboard.","confidence":"LOW","assumptions":["Approximately 5-12 FTE-years across quantitative modeling, liquidity supervision, data engineering, security, validation, legal/compliance, product, and operations.","Existing supervisory feeds are reusable but need reconciliation and production controls.","The estimate excludes purchasing new bank-level data, major legacy-system replacement, and cross-jurisdiction rollout.","Range is resource-equivalent, not a vendor quote or appropriated budget."],"source_ids":["S2","S7","S8"]},"operational_launch":{"band_2026_usd":"250K_TO_1M","scope":"Run a time-bounded parallel shadow launch, train analysts, tune only preregistered operational thresholds, conduct incident and rollback exercises, obtain governance approvals, and complete launch review.","confidence":"LOW","assumptions":["Approximately 1.5-4 FTE-years over a six-to-twelve-month parallel run.","No automatic supervisory action or public output is enabled.","Independent validation and data-governance functions already exist institutionally.","Material remediation discovered during parallel operation would move costs into the higher startup band."],"source_ids":["S2","S7","S8"]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Operate secure pipelines and dashboards; investigate alerts; monitor data, residuals, conditioning, gaps, drift, and performance; perform periodic recalibration, independent review, documentation, and access-control audits.","confidence":"LOW","assumptions":["Approximately 1.5-4 recurring FTE equivalents plus moderate secure-compute and storage costs.","The system remains advisory and parallel to ordinary monitoring.","At least annual independent review plus more frequent outcome and drift monitoring.","Major redesign following regime change is excluded and would be treated as new startup work."],"source_ids":["S7","S8"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"Official standards, regulation, and central-bank research establish multidimensional liquidity risk, payment obligations, propagation, and system-wide amplification, though not the proposal-specific modal miss mechanism.","source_ids":["S1","S2","S3","S4"]},"externally_credible_adopter_or_authorizer":{"status":"YES","reason":"Banking supervisors and central banks are identifiable authorizers; Federal Reserve rules prescribe relevant monitoring and the ECB already operates system-wide contagion analytics.","source_ids":["S1","S2","S4"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"The claim can be tested at a fixed false-alert budget against thresholds, PCA/common factors, centrality, and direct transition forecasting, with lead time and unique correct escalations as outcomes.","source_ids":["S3","S4","S5","S6"]},"bounded_next_evidence_step":{"status":"YES","reason":"A frozen-data, preregistered, retrospective shadow comparison with an untouched temporal holdout, explicit validity gates, and no live decisions is bounded.","source_ids":["S6","S7"]},"no_unresolved_safety_or_authority_stop":{"status":"UNCERTAIN","reason":"The proposed shadow use is low-authority and reversible, but no named supervisory partner, data custodian, jurisdiction, confidentiality classification, legal approval, or independent model owner is identified. These must be resolved before accessing data.","source_ids":["S2","S7"]},"credible_cost_scope_and_range":{"status":"YES","reason":"All four ranges name concrete work packages and FTE assumptions, use official wage data as a public anchor, and explicitly exclude data acquisition and legacy replacement. Confidence remains low outside the first study because no partner architecture or procurement quote exists.","source_ids":["S7","S8"]}},"next_evidence_step":"With a named supervisor and written data-custodian approval, freeze one historical panel containing calm periods and all available liquidity or payment-stress episodes. Before viewing the temporal holdout, specify variables, scaling, missing-data handling, regime windows, event labels, a false-alert budget, conditioning and spectral-gap limits, drift thresholds, and a consequence-weighted residual budget. Fit only on the earlier segment. Compare (1) the ordinary threshold dashboard, (2) PCA or a dynamic common-factor monitor, (3) exposure-network centrality, (4) a regularized direct VAR or state-transition forecast without modal prioritization, and (5) the proposed gain-plus-sensitivity modal rule. On the untouched later segment, measure false alerts per quarter, precision-recall, lead time to documented escalation or payment disruption, unique correct alerts, calibration, reconstruction error, mode stability under bootstrapping and reasonable rescaling, and incremental value over the direct forecast. Falsify the intervention if it has no preregistered lead-time or unique-correct-alert improvement at the same false-alert budget; if results depend materially on reasonable scaling or sampling choices; if the operator is ill-conditioned or lacks a usable gap; if identities drift beyond tolerance; if consequence-weighted residuals exceed budget; or if a directly observable institution or variable explains the events equally well. Keep all output access-controlled and nonbinding.","blocking_evidence":["No opened source shows that institution-level thresholds or system averages actually missed a reproducible weakly damped liquidity mode before a documented supervisory escalation.","No named supervisor, data custodian, approved dataset, event-label definition, or legal/confidentiality determination is available.","The prevalence and separability of stress events are unknown; rare events may make lead-time and false-alert comparisons statistically underpowered.","No empirical evidence establishes that eigenmodes of the estimated transition are stable under reasonable variable scaling, missing-data treatment, rolling windows, or policy-regime changes.","No evidence shows incremental value over the same transition model used for direct prediction; this comparator is necessary to isolate the modal prioritization contribution.","The system's non-normality, eigenvector conditioning, spectral gaps, residual consequence distribution, and mode-coupling behavior are unknown.","A highly relevant 2026 SSRN result titled \"Evaluation of systemic risk in a financial system using dynamic mode decomposition\" appeared in search at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6031928, but opening returned HTTP 403; it was not treated as verified evidence and requires full-text differentiation.","Production cost estimates lack partner-specific architecture, procurement, security-accreditation, and staffing evidence.","World novelty, patentability, freedom to operate, market size, and realized impact remain unmeasured."],"research_disposition":"PARTNERED_RESEARCH_PROGRAM","world_novelty_boundary":"The search establishes adjacent public prior art in regulatory liquidity monitoring, interbank and system-wide contagion simulation, and financial dynamic-mode/Koopman analysis. It did not find an opened source demonstrating the complete combination of confidential supervisory liquidity-state transition estimation, gain-and-consequence mode prioritization, traceability to banks and variables, explicit residual/conditioning/gap/drift gates, and a governed shadow-review workflow. This is only a contrastive search boundary, not a world-novelty conclusion. Patents, proprietary supervisory systems, non-English literature, vendor deployments, and the inaccessible 2026 SSRN lead were not resolved.","arm":"PROPOSAL_FIRST","candidate_version":0,"controller_recommendation":{"action":"STOP_EMPIRICAL_RESEARCH_NEEDED","repairable":true,"material_progress_observed":true,"progress_targets":["Secure a named supervisory partner, data custodian, written access authority, confidentiality controls, and an independent model-validation owner.","Produce a frozen data dictionary and feasibility audit covering temporal depth, missingness, reporting changes, stress-event labels, bilateral exposures, and outcome timestamps.","Preregister the ordinary-dashboard, PCA/common-factor, centrality, and direct-transition-forecast comparators plus a fixed false-alert budget and primary lead-time metric.","Predefine numerical falsifiers for conditioning, non-normality, spectral separation, bootstrap and scaling stability, mode drift, and consequence-weighted residual error.","Complete the held-out retrospective study and demonstrate incremental value over the direct forecast, not merely over univariate thresholds.","Differentiate the proposal against the inaccessible 2026 systemic-risk DMD paper and any nonpublic supervisory analytics located by the partner.","Replace resource-equivalent estimates with partner-specific security, integration, validation, and recurring-operation estimates before deployment inquiry."],"reason":"Bounded web research supports the underlying systemic-liquidity problem, credible authorizers, and extensive adjacent practice, while leaving a narrow, falsifiable workflow claim. The decisive questions—whether stable liquidity modes exist in supervisory data and improve held-out prioritization—require proprietary data and empirical testing. Under the stated controller rule, further web research cannot establish those results, so the correct action is STOP_EMPIRICAL_RESEARCH_NEEDED."}}