GUIDED SIMULATION

01 / 06

HOW THE SYSTEM WORKS

Banks depend on shared cloud providers.

In this simulation, we see how twenty synthetic banks depend on three shared cloud providers for critical digital capacity.

SYSTEM STRUCTURE 20 banks · 3 providers

SYNTHETIC MODEL

The guided story uses the same deterministic scenario engine as Scenario Lab.

READY SCENE 01 / 06

HOW THE SYSTEM WORKS Banks are connected to shared infrastructure, not isolated technology stacks.

Affected banks8 / 20baseline outage

Unmet capacity—capacity units

Workload restored—SCFR baseline

System resilience—SCFR score

SCENARIO CONTROLSAdjust assumptions

SCENARIO LAB

Test the recovery mechanism.

Change the assumptions below. The Earth, outcomes and comparison update from the same deterministic model.

ShockWhich shared dependency fails?

Failed cloud providers 1 selected

Blue Cloud8 banks Orange Cloud7 banks Green Cloud5 banks

Select one or more providers. Multiple providers can fail simultaneously.

Draft shock Blue Cloud

Prepared resilienceHow much capacity is available?

Immediate backup capacity 20% 0%25%50% Prepared reserve 25% 0%30%60%

CoordinationHow is pooled reserve allocated?

SCFR allocation rule

SCENARIO READY Adjust the controls, then run the scenario to apply the new assumptions.

WHAT STAYS FIXED Same bank network and model equations Only the selected scenario assumptions change. The synthetic bank sample and calculation logic remain constant.

VIEW RECOVERY MECHANISM Scenario stays fixed while the allocation mechanism changes.

ShockBlue Cloud Market20% Reserve25%

AFFECTED BANKS — of 20 synthetic banks

UNMET CAPACITY — units after selected mechanism

WORKLOAD RESTORED — critical workload

SYSTEM RESILIENCE — / 100

COMPARE RECOVERY MODELS Same shock. Same reserve budget. Different coordination.

Blue Cloud · 20% market · 25% reserve

CONTROLLED TEST Individual Reserves and SCFR receive the same aggregate reserve pool. The allocation mechanism changes; the reserve budget does not.

WHY THE MAP CHANGES

Market view shows the capacity shortage created by simultaneous recovery demand.

MODEL & EVIDENCE

Understand what the simulator is doing — and what it is not claiming.

Resilience Atlas is a deterministic mechanism stress-test. It uses synthetic institutions to isolate how a shared-provider shock and different reserve-allocation rules change recovery outcomes.

REAL-WORLD MOTIVATION Shared ICT dependency and operational-resilience risk Grounded in BIS, EBA and DORA material.

SYNTHETIC MODEL Banks, workloads, readiness and numerical outputs Illustrative test bed — not estimates of real institutions.

BUILT FOR DISCUSSION BY

Bank resilience teamsExplore shared-provider recovery bottlenecks.

SupervisorsMake concentration and coordination assumptions visible.

ResearchersStress-test mechanisms without presenting synthetic outputs as forecasts.

ANNOTATED SYSTEM MAP One network model, explained visually

Illustrative topology · no real-bank locations

MODEL FLOW

From shock to resilience outcome

The same sequence is used by Guided Simulation and Scenario Lab.

01ShockA shared provider fails.

→

02DemandDependent banks need recovery capacity.

→

03MarketImmediate backup capacity is allocated.

→

04ReserveIndividual or pooled reserve is applied.

→

05ReadinessCapacity is converted into restored workload.

→

06OutcomeRecovery and systemic resilience are measured.

CORE EQUATIONS

Four calculations drive the mechanism

These are the equations implemented in model/simulation.js.

CAPACITY RATIO min(allocated capacity / critical load, 1) How much of the required workload has capacity behind it.

RESTORED FRACTION capacity ratio × failover readiness Capacity alone is not enough; readiness limits effective recovery.

RESERVE POOL total system load × reserve % Individual and SCFR scenarios receive the same aggregate reserve budget.

SYSTEMIC RESILIENCE SCORE Σ(loadᵢ × importanceᵢ × restoredᵢ) / Σ(loadᵢ × importanceᵢ) × 100 A synthetic weighted recovery metric. It is not an official regulatory score.

EXPERIMENT DESIGN

What stays fixed — and what changes

The central comparison isolates coordination rather than giving one mechanism more total reserve.

HELD CONSTANT

Failed provider
Affected synthetic banks
Bank workloads, readiness and importance
Emergency-market assumption
Total aggregate reserve budget

MECHANISM CHANGES

Market: immediate post-shock capacity only
Individual: reserve remains bank-specific
SCFR: the same reserve pool can move across affected banks

SCFR ALLOCATION RULE

Systemic: critical load × stylized importance
Equal: capacity shared across banks still in need
Readiness: higher failover readiness first

ROBUSTNESS SWEEP

Does coordination only help in one chosen scenario?

The simulator reruns a local 3×3 grid around the current market and reserve assumptions. Each cell shows SCFR resilience minus Individual Reserve resilience.

POSITIVE UPLIFT — tested synthetic combinations

MAX UPLIFT — resilience-score points

CURRENT SHOCK — provider selection held fixed

Synthetic diagnostic only. A positive cell means pooled SCFR coordination restores more systemically weighted workload than ring-fenced Individual Reserves under the same aggregate reserve budget for that tested assumption pair.

EVIDENCE BASE

Why the research question is plausible

These sources motivate studying correlated third-party ICT disruption. They do not validate the prototype's synthetic numerical outputs.

BIS / FSI · 2023 Managing cloud risk

Discusses systemic implications of financial-sector dependence on a concentrated cloud market.

Official source ↗ EBA · 2026 Risk Assessment Report

Highlights operational-resilience risks associated with interconnected third-party ICT dependencies.

Official source ↗ EU DORA OVERSIGHT Critical ICT third parties

Addresses systemic and concentration risks from reliance on critical external ICT providers.

Official source ↗ CPMI–IOSCO · 2026 Reliance on third-party providers

Examines how critical third-party dependencies can amplify risk for financial market infrastructures.

Official source ↗

Evidence boundary

The external evidence supports the problem motivation. It does not establish that SCFR is feasible or that real banks would achieve the resilience scores produced here.

LIMITATIONS

What v0.1 deliberately does not model

These boundaries define the prototype rather than being hidden assumptions.

Market pricing & bidding Provider compatibility Migration time Network bottlenecks Data synchronization Contractual constraints Cross-border rules Reserve cost Recovery-time objectives Multi-provider failures

REPRODUCIBILITY

Inspect the assumptions, code and current scenario

Same inputs produce the same outputs; the recovery engine contains no random draw.

MODELsimulation.jsScenario and allocation logic ↗ INPUTSbanks.jsSynthetic bank/provider dataset ↗ TESTSAutomated checksModel invariants and UI regressions ↗ SOURCESEvidence baseResearch motivation and source notes ↗

CURRENT SCENARIO Blue Cloud · 20% market · 25% reserve · systemic rule

PROTOTYPE TRANSPARENCY

AI tools materially assisted brainstorming, code drafting, debugging, documentation and interface iteration. The project author selected the research framing and assumptions, reviewed outputs and is responsible for the final prototype.