Sovereign Debt Contagion & Geopolitical MCP

Sovereign debt contagion modelling and geopolitical risk analysis are now available as a single MCP server your AI agent can call directly. This server gives Claude, Cursor, and any MCP-compatible client 8 quantitative finance tools that fuse live data from 18 international sources — IMF, World Bank, OECD, Eurostat, FRED, BLS, UN COMTRADE, exchange rate feeds, CoinGecko, OFAC, OpenSanctions, Interpol, World Bank Projects, Grants.gov, USAspending, and GDACS — into institutional-grade sovereign risk models. The result is a fully automated pipeline from raw macroeconomic data to crisis probability scores, contagion chains, tipping point detection, and geopolitical alignment maps.

The server implements seven distinct quantitative algorithms: a 4th-order tensor CP decomposition for network structure, Euler-Maruyama Monte Carlo simulation of coupled sovereign debt SDEs, Jacobian-based bifurcation analysis, HHI-weighted commodity vulnerability with 95th-percentile VaR, Markov-switching currency crisis detection via the Hamilton filter, a modified IMF Assessing Reserve Adequacy metric, and latent space ideal point estimation via SVD for geopolitical positioning. Each tool runs the relevant data collection actors in parallel, calibrates the model from live data, and returns structured analytical output in seconds.

What data can you access?

MCP tools

Why use Sovereign Debt Contagion & Geopolitical MCP Server?

Building sovereign risk models manually means assembling data from a dozen international databases, normalising currencies and time periods, writing the SDE discretisation code, implementing tensor decomposition, and interpreting bifurcation diagrams — a project that takes a quantitative analyst weeks and requires Bloomberg Terminal or Refinitiv licenses running $20,000-30,000 per year.

This MCP server eliminates that entirely. Your AI agent calls a single tool, the server fetches fresh data from 18 public sources in parallel, runs the calibrated model, and returns structured analytical results. No data sourcing. No numerical methods implementation. No subscription required.

• Scheduling — run daily or weekly sovereign monitoring for a country portfolio via Apify's built-in task scheduler
• API access — trigger the server from Python, JavaScript, or any MCP-compatible client, including Claude and Cursor
• Parallel data collection — all 18 actors run simultaneously; total data collection time is bounded by the slowest single source, not the sum
• Monitoring — configure Slack or email alerts when runs fail via Apify's notification system
• Integrations — connect results to Zapier, Make, Google Sheets, or webhooks for downstream reporting pipelines

Features

• 4th-order tensor network construction — builds T[creditor][debtor][exposure_type][time] from 18 sources, decomposed via Alternating Least Squares CP decomposition (T ≈ Σ_r λ_r · a_r ⊗ b_r ⊗ c_r ⊗ d_r) to reveal latent cross-exposure structures
• Euler-Maruyama Monte Carlo with contagion shocks — discretises the coupled SDE system dd(t) = (r(t) − g(t))d(t)dt + pb(t)dt + σ_d·d(t)·dW(t) with up to 10,000 paths; when country j exceeds distress threshold, injects contagion shock r_i += exposure_ij × severity_j into creditor country i
• Bifurcation classification — computes the Jacobian at the current fixed point, finds dominant eigenvalues via power iteration, then sweeps interest rate spread parameters to locate tipping points and classifies them as saddle-node, Hopf, or transcritical
• HHI × VaR commodity vulnerability — Herfindahl-Hirschman Index of the commodity export basket multiplied by commodity-specific 95th-percentile log-normal VaR, scaled by commodity exports to GDP; outputs four risk tiers (LOW / MEDIUM / HIGH / CRITICAL)
• Markov-switching currency crisis detection — two-regime model with Hamilton forward recursion; augmented by Exchange Market Pressure index EMP = Δe/e − (σ_e/σ_r)(ΔR/R) + (σ_e/σ_i)Δi; cryptocurrency volume spikes as supplementary capital flight indicator; crisis flag at P(crisis) > 0.70
• Modified IMF ARA reserve adequacy — RA = reserves / (0.3×STD + 0.2×portfolio + 0.1×M2 + 0.1×exports); adjusted for current crisis probability and geopolitical alignment score derived from latent space proximity to major reserve currency issuers
• Latent space geopolitical alignment — SVD on the country-feature matrix (trade flows, sanctions co-occurrence, aid relationships) via power iteration and deflation, yielding low-dimensional ideal point estimates analogous to DW-NOMINATE; assigns countries to US_ALIGNED, EU_ALIGNED, CN_ALIGNED, or NON_ALIGNED blocs
• Composite sovereign risk report — integrates all 7 algorithms into a single overallSystemicRisk score (0-1), letter grade (A-F), ranked top risk factors, and actionable recommendations
• Seeded deterministic numerics — uses Mulberry32 PRNG seeded from country identifiers; results are reproducible across runs with identical inputs
• Spending limit enforcement — each tool checks Actor.charge() before executing; stops gracefully when the run budget is reached
• 18-actor parallel orchestration — runActorsParallel() fires all actor calls via Promise.all(); individual actor failures return empty arrays rather than crashing the pipeline

Use cases for sovereign debt contagion analysis

Sovereign risk portfolio stress testing

Fixed income portfolio managers holding emerging market sovereign bonds need to quantify correlated default risk. This tool runs Monte Carlo SDE simulations across a custom country basket, returning per-country default probabilities, expected default times, and the specific contagion chains that propagate distress across the portfolio.

Tipping point and systemic fragility detection

Macro risk analysts advising policymakers or institutional clients need to know how close a sovereign system is to a cascading failure. The bifurcation tool computes Jacobian eigenvalues at the current system state and sweeps parameter space to locate the exact interest rate spread or growth shock that would push the system from stable to crisis equilibrium.

Commodity-dependent economy assessment

Sovereign credit analysts and development finance institutions evaluating resource-rich countries need quantified exposure to commodity price shocks. The HHI × VaR vulnerability tool identifies Dutch disease risk, resource curse dynamics, and which single commodity shock would be most destabilising.

Currency crisis early warning

Central bank research departments and international financial institutions monitoring emerging market FX stability can use the Markov-switching currency tool to track regime transition probabilities in real time, combining Exchange Market Pressure indices with cryptocurrency volume as a capital flight proxy.

Geopolitical alignment and sanctions exposure mapping

Geopolitical risk consultancies and corporate treasury teams managing country exposure need to understand how diplomatic alignment affects sovereign resilience. The SVD-based ideal point tool places countries in a 2D geopolitical latent space, assigns bloc membership, and computes isolation scores that feed directly into reserve adequacy adjustments.

Country risk due diligence for investment and lending

Development banks, export credit agencies, and cross-border lenders performing country risk due diligence can run the full generate_sovereign_risk_report tool to receive a composite systemic risk score, risk grade, and ranked recommendations — covering debt dynamics, currency risk, commodity exposure, reserve buffers, and geopolitical positioning in a single call.

How to connect this MCP server

The server runs in Apify Standby mode and exposes an MCP endpoint at /mcp. Connect it once and your AI agent can call all 8 tools on demand.

Step 1: Get your Apify API token

Go to Apify Console > Settings > Integrations and copy your API token.

Step 2: Add to your MCP client

Claude Desktop — add to claude_desktop_config.json:

{
  "mcpServers": {
    "sovereign-debt-contagion-geopolitical": {
      "url": "https://sovereign-debt-contagion-geopolitical-mcp.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}

Cursor — add to Cursor MCP settings:

{
  "mcpServers": {
    "sovereign-debt-contagion-geopolitical": {
      "url": "https://sovereign-debt-contagion-geopolitical-mcp.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}

Other MCP clients — use the URL https://sovereign-debt-contagion-geopolitical-mcp.apify.actor/mcp with your token in the Authorization: Bearer header.

Step 3: Run your first analysis

Ask your AI agent: "Analyse sovereign debt contagion risk for Greece, Italy, and Argentina using the sovereign debt contagion MCP server" — the agent will call build_sovereign_exposure_network followed by simulate_debt_crisis_cascade and return structured results.

MCP tool parameters

build_sovereign_exposure_network

simulate_debt_crisis_cascade

detect_bifurcation_points

compute_commodity_vulnerability

detect_currency_crisis_signal

assess_reserve_adequacy

map_geopolitical_alignment

generate_sovereign_risk_report

Parameter tips

• Start with 3-6 countries — network analysis is most interpretable with a focused set; adding more countries increases cost and run time proportionally
• Lower monte_carlo_paths to 1,000-2,000 for exploratory runs; use 10,000 for final risk estimates or regulatory reporting
• Set distress_threshold_pct below critical_debt_pct — the distress threshold triggers contagion propagation; the critical threshold triggers default
• Add a specific query for scenario analysis (e.g. "debt restructuring default 2024") to focus data pulls from the underlying actors

Output example

Output from generate_sovereign_risk_report for countries ["GR", "IT", "AR"]:

{
  "overallSystemicRisk": 0.72,
  "riskGrade": "C",
  "topRisks": [
    "Argentina default probability exceeds 60% within 5-year horizon",
    "Italy debt-to-GDP trajectory unsustainable under base interest rate scenario",
    "Cross-exposure contagion chain: AR → GR via multilateral creditor linkages",
    "Currency crisis probability elevated for ARS (P=0.84) and TRY (P=0.71)",
    "Greece reserve adequacy MARGINAL after crisis-probability adjustment"
  ],
  "recommendations": [
    "Reduce gross AR sovereign exposure; current contagion weight to GR is 0.23",
    "Monitor ECB rate path for IT bifurcation proximity (margin to tipping: 0.18)",
    "Argentina commodity HHI=0.61 (soy-dominant): hedge commodity price tail risk",
    "Greece geopolitical alignment score 0.78 (EU_ALIGNED) supports ARA adjustment"
  ],
  "networkSummary": {
    "countries": 3,
    "edges": 47,
    "tensorRank": 5
  },
  "crisisCascadeSummary": {
    "systemicRisk": 0.69,
    "defaultProbabilities": {
      "AR": 0.63,
      "IT": 0.18,
      "GR": 0.24
    },
    "contagionChains": 4
  },
  "bifurcationSummary": {
    "systemStable": false,
    "tippingPointCount": 2,
    "nearestTippingPoint": {
      "parameter": "interest_rate_spread_IT",
      "criticalValue": 4.8,
      "currentValue": 4.62,
      "marginToTipping": 0.18,
      "type": "saddle_node",
      "affectedCountries": ["IT", "GR"]
    }
  },
  "commoditySummary": {
    "systemicRisk": 0.41,
    "criticalCountries": 1
  },
  "currencySummary": {
    "systemicRisk": 0.58,
    "countriesInCrisis": 1
  },
  "reserveSummary": {
    "systemicRisk": 0.39,
    "inadequateCountries": 1
  },
  "geopoliticalSummary": {
    "blocs": [
      { "name": "EU_ALIGNED", "members": ["GR", "IT"] },
      { "name": "NON_ALIGNED", "members": ["AR"] }
    ],
    "svdExplainedVariance": [0.44, 0.22, 0.11]
  }
}

Output fields

generate_sovereign_risk_report

simulate_debt_crisis_cascade

detect_bifurcation_points

compute_commodity_vulnerability

detect_currency_crisis_signal

assess_reserve_adequacy

map_geopolitical_alignment

How much does it cost to run sovereign debt contagion analysis?

This MCP server uses pay-per-event pricing — you pay per tool call. The underlying actor runs that fetch live data are charged separately at Apify's standard compute rates. Typical all-in cost per tool call is $0.04-$0.055 for the MCP event plus $0.50-$3.00 in actor compute costs depending on the number of countries and data sources involved.

You can set a maximum spending limit per run in Apify Console to control costs. The server enforces the limit gracefully — if the budget is reached mid-session, the tool returns a structured error rather than crashing.

Compare this to Bloomberg DLIB or Refinitiv sovereign risk modules at $20,000-$30,000 per year. Most users of this server spend $20-$80 per month for regular sovereign monitoring workflows.

How Sovereign Debt Contagion & Geopolitical MCP Server works

Phase 1: Parallel data collection from 18 sources

Every tool call begins with runActorsParallel(), which fires all relevant actor calls simultaneously via Promise.all(). Each call has a 180-second timeout (240 seconds for the full report). Individual actor failures return empty arrays rather than propagating errors — the pipeline completes with available data even when individual sources are temporarily unreachable. The full network and risk report tools call all 18 actors in parallel: IMF, World Bank, OECD, Eurostat, FRED, BLS, UN COMTRADE, exchange rate tracker, exchange rate history, ECB rates, CoinGecko, OFAC, OpenSanctions, Interpol, World Bank Projects, Grants.gov, USAspending, and GDACS.

Phase 2: Sovereign network construction and tensor decomposition

buildSovereignNetwork() ingests all 18 result arrays and constructs SovereignNode objects from IMF/World Bank/OECD data (extracting debt-to-GDP, interest rates, GDP growth, primary balance, FX volatility, reserves, short-term debt, M2, exports, and commodity exports). It builds ExposureEdge objects from trade, sanctions, and aid data, then constructs the 4th-order tensor T[creditor][debtor][exposure_type][time] — with time represented as 4 quarterly snapshots using exponential decay weighting (e^{-0.05q}). CP decomposition via ALS iterates up to 100 times with a 1e-6 tolerance to yield creditor, debtor, exposure type, and time factor matrices plus component weights λ_r, converging to identify latent cross-exposure structures not visible in bilateral data alone.

Phase 3: Algorithm execution

Depending on the tool called, scoring.ts runs one or more of the seven algorithms against the constructed network:

• SDE simulation: Euler-Maruyama timesteps discretise each country's debt dynamics; contagion shocks propagate when distress thresholds are crossed; Mulberry32 PRNG seeded from country identifier hash ensures reproducibility
• Bifurcation analysis: N×N Jacobian computed at the current fixed point (∂f_i/∂d_j); dominant eigenvalue found via power iteration with deflation; 1D parameter sweep over interest rate spreads locates sign changes; bifurcation type classified from imaginary eigenvalue component
• Commodity vulnerability: UN COMTRADE data mapped to commodity categories; HHI = Σ(share_i²); log-normal VaR at 95th percentile with commodity-specific volatility; scaled by commodity exports/GDP
• Markov-switching: Hamilton filter forward recursion P(s_t=k|r_{1:t}) ∝ f(r_t|s_t=k) × P(s_t=k|r_{1:t-1}); two regimes with distinct mu and sigma; EMP index computed; crypto volume overlaid as capital flight indicator
• Geopolitical alignment: country-feature matrix from trade, sanctions, and aid; SVD via power iteration and deflation; bloc membership assigned from cosine similarity to anchor countries (US, DE, CN)

Phase 4: Composite report assembly

computeSovereignRiskReport() weights the seven risk dimensions (cascade 30%, bifurcation 25%, currency 15%, commodity 15%, reserve 10%, geopolitical 5%) to produce overallSystemicRisk, converts it to a letter grade, and generates ranked topRisks and recommendations by sorting risk factors from all dimensions.

Tips for best results

1. Start with build_sovereign_exposure_network for exploratory work. The network output tells you which bilateral exposures are most significant before committing to the more expensive cascade simulation.

2. Use detect_bifurcation_points before simulate_debt_crisis_cascade for tipping point workflows. Bifurcation analysis runs faster and tells you whether you are near a regime transition before investing in 10,000-path Monte Carlo.

3. Compare the allianceAdjusted and araMetric fields in reserve adequacy. A large gap between the raw ARA and the alliance-adjusted score indicates that geopolitical isolation is the binding constraint on reserve adequacy, not the raw reserve level.

4. Interpret contagionChains directionally. Each chain has a trigger (the first country to default) and affected[] (the downstream contagion path). Sort by severity to identify your highest-priority exposure reduction targets.

5. Cross-reference the geopolitical latent space distanceMatrix with your reserve adequacy results. Countries with large geopolitical distances from USD, EUR, and CNY issuers have their ARA scores penalised — this is the formal mechanism linking geopolitical risk to liquidity risk.

6. Use tensor_rank: 3 for small country sets (3-5 countries) and tensor_rank: 5-7 for larger sets. Over-parameterising the tensor rank relative to the number of countries will produce noise rather than signal.

7. For recurring monitoring, add a query parameter with the current year and specific risk context (e.g. "sovereign debt restructuring 2026 EM") to focus the underlying data pulls on the most relevant recent publications.

8. Set monte_carlo_paths: 1000 during prototyping. This reduces run time and cost by 10x relative to the 10,000-path default while still providing directionally accurate risk rankings. Use full 10,000 paths for final deliverables.

Combine with other Apify actors

Limitations

• No confidential bilateral exposure data — the tensor network is built from publicly available IMF, World Bank, and OECD data; actual cross-border bank exposures from BIS confidential data are not included, meaning bilateral linkages are estimated from trade and aid proxies rather than direct financial claims
• Euler-Maruyama approximation error — discretisation of continuous-time SDEs introduces step-size-dependent numerical error; the default time step is 0.1 years; results are approximations, not exact solutions
• Two-regime Markov-switching limitation — the currency crisis model assumes exactly two regimes (stable and crisis); multi-regime or regime-within-regime dynamics (e.g. managed float vs. free float vs. peg break) are not modelled
• Local linearisation in bifurcation analysis — the Jacobian captures local stability; the full nonlinear system may exhibit global bifurcation behaviour not captured by local eigenvalue analysis
• Geopolitical alignment is estimated, not observed — ideal point positions are inferred from observable proxies (trade, sanctions, aid); they do not reflect classified diplomatic intelligence, military alliances, or informal political relationships
• Monte Carlo paths capped at 10,000 — the monte_carlo_paths parameter is bounded for computational feasibility; confidence intervals on low-probability tail events (p < 0.01) may be wide at this sample size
• Data freshness depends on source update cadence — IMF WEO is updated twice yearly; FRED and exchange rate data are near-real-time; analysis mixing slow-updating and fast-updating sources may reflect mixed time periods
• Country coverage limited to public data availability — countries with limited IMF and World Bank reporting (e.g. North Korea, some Pacific island states) will have sparse node attributes and less reliable risk estimates

Integrations

• Zapier — trigger a sovereign risk report on a weekly schedule and push results to a Google Sheet or Airtable dashboard for portfolio monitoring
• Make — build automated alert workflows that run the currency crisis signal tool daily and send a Slack notification when any country's crisisProbability exceeds 0.70
• Google Sheets — pipe the defaultProbabilities and contagionChains output directly into a live country risk tracker spreadsheet
• Apify API — integrate the MCP endpoint into a Python or JavaScript risk analytics dashboard for programmatic access to all 8 tools
• Webhooks — configure webhook notifications when bifurcation analysis detects that systemStable has changed from true to false across your country set
• LangChain / LlamaIndex — use this MCP server as a tool in a multi-agent workflow where a reasoning agent orchestrates calls across sovereign risk analysis, sanctions screening, and macroeconomic data retrieval

Troubleshooting

Tool returns partial data or empty arrays for some countries. This typically means one or more of the 18 underlying actors returned no results for the given country query. Individual actor failures are handled gracefully — the pipeline completes with the available data. Check the Apify run log for Actor X failed warnings. Try adding a more specific query parameter (e.g. "Greece sovereign debt GDP 2024") to improve data retrieval from IMF and World Bank actors.

systemStable: true but high overallSystemicRisk. Stability and systemic risk are distinct measures. systemStable refers to local eigenvalue stability at the current equilibrium; overallSystemicRisk is the Monte Carlo probability of default. A system can be locally stable (negative eigenvalues) while having high default probability under stochastic shocks.

Run times exceeding 5 minutes for the full risk report. The generate_sovereign_risk_report tool calls all 18 actors in parallel with 240-second individual timeouts. Total wall time is dominated by the slowest actor. If a specific actor is consistently slow, reduce maxResults by adding a more targeted query. You can also run individual tools (commodity, currency, reserves) in separate calls rather than the full report.

bifurcations[] returns an empty array. This means the system is not near any detected tipping points under the current parameter sweep. The bifurcation analysis sweeps interest rate spreads by ±50% in 20 steps; if no sign change in the leading eigenvalue is detected in that range, no bifurcation is reported. This is a valid result indicating the system is robustly stable under the analysed parameter range.

Spending limit reached error. Each tool call charges the MCP event before executing the actor calls. If eventChargeLimitReached is true, increase your per-run spending limit in Apify Console under the actor's run settings, or set a higher maxTotalChargeUsd when triggering via API.

Responsible use

• This server only accesses publicly available data from international organisations (IMF, World Bank, OECD), government databases (FRED, BLS, OFAC, USAspending), and open data platforms (OpenSanctions, GDACS, CoinGecko).
• Sovereign risk scores and default probabilities produced by this server are model outputs, not investment advice. They should be used as inputs to human analytical processes, not as standalone recommendations.
• Sanctions data from OFAC and OpenSanctions should be verified against primary sources before use in compliance decisions.
• Comply with the terms of service of all 18 underlying data sources when using outputs for commercial purposes.
• For guidance on web scraping legality, see Apify's guide.

FAQ

How does sovereign debt contagion analysis work in this server? The server models contagion using coupled stochastic differential equations. When country j's debt-to-GDP ratio crosses the distress_threshold_pct, all creditor countries i receive a shock to their effective interest rate: r_i += exposure_ij × severity_j. This raises i's debt accumulation rate, potentially pushing it toward its own distress threshold and creating a cascade. The Monte Carlo simulation runs this process 1,000-10,000 times to estimate empirical default probabilities.

What is the 4th-order tensor decomposition used for? The tensor T[creditor][debtor][exposure_type][time] captures four dimensions of bilateral sovereign exposure simultaneously. CANDECOMP/PARAFAC (CP) decomposition factorises this into a sum of rank-1 components (T ≈ Σ_r λ_r · a_r ⊗ b_r ⊗ c_r ⊗ d_r) via Alternating Least Squares. The resulting latent structures reveal clusters of creditor-debtor relationships that co-move across exposure types and time — patterns not visible in bilateral exposure matrices alone.

How accurate are the default probability estimates? The accuracy depends on data quality from the 18 sources and the validity of the log-normal SDE model assumption. The model is calibrated from public data which typically lags current conditions by 3-12 months. Treat outputs as scenario-based probability estimates rather than point forecasts. At 10,000 paths, the standard error on a 50% default probability is approximately ±0.5 percentage points.

How is geopolitical alignment different from traditional geopolitical risk scores? Traditional geopolitical scores (e.g. from Eurasia Group or Control Risks) are qualitative assessments made by analysts. This server estimates geopolitical positioning quantitatively using Singular Value Decomposition on a country-feature matrix built from observable data — trade flows, sanctions co-occurrences, and aid relationships. The method is analogous to DW-NOMINATE ideal point estimation used to map legislative voting positions. The result is a reproducible, data-driven positioning that updates automatically with new trade and sanctions data.

How does sovereign debt contagion analysis differ from IMF stress testing? The IMF uses DSA (Debt Sustainability Analysis) which projects debt paths under deterministic macro scenarios. This server uses stochastic differential equations with Monte Carlo simulation to produce empirical default probability distributions, and couples them with a network contagion model to capture second-order cross-country effects. The bifurcation analysis component identifies structural tipping points not captured by linear scenario analysis.

Can I use this server to monitor a country in real time? Near-real-time monitoring is supported for variables that update frequently: exchange rates (ECB, tracker), crypto volumes (CoinGecko), disaster alerts (GDACS), and sanctions changes (OFAC, OpenSanctions). IMF, World Bank, and OECD data update on quarterly to annual schedules. You can schedule the currency crisis signal tool to run daily for high-frequency monitoring and the full risk report weekly for comprehensive updates.

How many countries can I analyse in one call? There is no hard limit, but performance degrades non-linearly above 10-15 countries due to the tensor construction and SDE simulation. For large sets (20+ countries), consider running the lighter tools (commodity_vulnerability, currency_crisis_signal) across the full set and reserving the full cascade simulation for a focused subset.

Is it legal to use the data from these 18 sources? All 18 sources provide publicly available data under open access or government transparency mandates. IMF, World Bank, OECD, Eurostat, FRED, BLS, and USAspending data are published under open data licences. OFAC and Interpol data are publicly accessible for compliance and research purposes. CoinGecko provides a public API. See Apify's guide on web scraping legality for general guidance.

What does the marginToTipping value mean in bifurcation results? marginToTipping is the distance between the current parameter value and the critical value at which the system crosses a bifurcation boundary. For example, if parameter is interest_rate_spread_IT with currentValue: 4.62 and criticalValue: 4.80, the margin is 0.18 percentage points of spread widening before the system transitions to the crisis equilibrium. Smaller margins indicate higher proximity to instability.

How does the reserve adequacy tool incorporate geopolitical alignment? The modified IMF ARA metric is first computed as RA = reserves / (0.3×STD + 0.2×portfolio + 0.1×M2 + 0.1×exports). This is then adjusted by crisis probability from the Markov-switching model and multiplied by an alliance alignment score derived from the country's proximity in geopolitical latent space to the US, EU, and China — the three major reserve currency issuers. Countries closely aligned with these issuers receive higher adjusted scores because they have better access to emergency swap lines and IMF support programmes.

Can I combine this with other MCP servers in Claude? Yes. Claude Desktop supports multiple MCP servers simultaneously. You can combine this server with other data or analysis servers. A typical workflow would be: use this server's geopolitical alignment tool to identify high-risk country blocs, then use a separate news or filing search MCP server to pull recent developments for those specific countries.

Help us improve

If you encounter issues, you can help us debug faster by enabling run sharing in your Apify account:

1. Go to Account Settings > Privacy
2. Enable Share runs with public Actor creators

This lets us see your run details when something goes wrong, so we can fix issues faster. Your data is only visible to the actor developer, not publicly.

Support

Found a bug or have a feature request? Open an issue in the Issues tab on this actor's page. For custom sovereign risk models, additional data source integrations, or enterprise deployments, reach out through the Apify platform.