Climate-Economic Nexus MCP

Climate-economic integrated assessment via the Model Context Protocol gives AI agents quantitative tools for scenario analysis, tipping cascade detection, carbon market forecasting, and physical risk attribution. This MCP server orchestrates 18 live environmental and economic data sources — NOAA, GDACS, World Bank, IMF, FRED, GBIF, IUCN, Eurostat, and more — then applies 8 published mathematical frameworks to produce structured, decision-grade outputs.

The server runs persistently on the Apify platform in Standby mode. Any MCP-compatible client — Claude Desktop, Cursor, Windsurf, Cline — connects via a single endpoint URL with no infrastructure to manage. All 8 tools follow pay-per-event pricing: you pay only for the analysis you actually run.

⬇️ What data can you access?

Why use Climate-Economic Nexus MCP?

Building your own climate-economic analytics pipeline means licensing data from NOAA, World Bank, IUCN, and a dozen other sources, implementing published damage functions, running Monte Carlo simulations, and maintaining all of it. That is months of engineering work — and it still doesn't give you integrated cross-source analysis.

This MCP server handles the entire data collection and computation layer. Your AI agent calls a single tool with a natural-language query, and the server fans out to all 18 data sources in parallel, calibrates the mathematical models from live data, and returns structured JSON results ready for decision support.

Platform capabilities your agent inherits automatically:

• Standby mode — the server stays warm between calls, so there is no cold-start latency on inference requests
• Parallel execution — all 18 actor calls run concurrently via Promise.all, reducing per-tool latency to the slowest single source
• Spending limits — set a maximum budget per session; all tools check the limit before charging
• API access — connect from any MCP client or trigger analysis directly from Python, JavaScript, or cURL
• Monitoring — Apify platform logging captures every data source call, result count, and error for debugging
• Integrations — output can be piped to Zapier, Make, Google Sheets, webhooks, or HubSpot via Apify integrations

⬆️ MCP Tools

Use cases for climate-economic integrated assessment

Climate risk financial disclosure

Chief Risk Officers and sustainability teams building TCFD-aligned disclosures need quantified scenario analysis under RCP 2.6, 4.5, and 8.5 pathways. Use simulate_integrated_assessment for physical risk trajectories and downscale_spatial_impacts to map temperature and damage exposure at asset-level granularity. quantify_damage_uncertainty provides the confidence intervals required for robust disclosure narratives.

Carbon market strategy and ETS trading

Carbon trading desks and corporate sustainability teams managing emissions obligations need to anticipate price regime transitions before they happen. forecast_carbon_price_regimes applies regime-switching jump-diffusion with Hamilton filter smoothing to identify when the EU ETS or other markets are approaching a policy-transition regime, including the 90% confidence interval for price in each regime.

Insurance catastrophe modeling and loss-and-damage assessment

Actuaries pricing climate-related property insurance and reinsurance need to separate the anthropogenic climate signal from natural variability in historical loss data. attribute_climate_damages uses optimal fingerprinting with Total Least Squares — the same framework used in Allen and Stott (2003) — to compute what fraction of observed damages is attributable to anthropogenic forcing, supporting defensible actuarial assumptions and climate litigation analysis.

Adaptation investment planning under deep uncertainty

Infrastructure planners, government agencies, and development finance institutions need adaptation portfolios that perform well across a wide range of future scenarios, not just the central projection. optimize_robust_adaptation applies Patient Rule Induction Method (PRIM) bump-hunting to discover the specific scenario combinations where each strategy fails, and MOEA multi-objective optimization to produce a Pareto front balancing cost, damage reduction, and robustness.

Tipping point early warning and systemic risk monitoring

Risk analysts tracking interconnected earth system elements need quantitative proximity-to-bifurcation metrics, not qualitative descriptions. detect_tipping_cascades parameterizes AMOC, Amazon dieback, West Antarctic Ice Sheet, coral reefs, and permafrost thaw as coupled cusp catastrophe systems and runs 5,000 Euler-Maruyama Monte Carlo paths to estimate cascade probability and which tipping chains are most likely.

Biodiversity and natural capital accounting

Investment managers, regulatory bodies, and corporates reporting under TNFD frameworks need to quantify the economic value of ecosystem services at risk from habitat loss. assess_biodiversity_economic_loss combines GBIF occurrence data, IUCN threat assessments, and World Bank country indicators to compute species-area loss estimates and dollar-equivalent ecosystem service losses per biome, using percolation theory to identify connectivity collapse thresholds.

How to connect this MCP server

This server runs in Apify Standby mode and exposes a persistent /mcp endpoint. No deployment or configuration is needed beyond adding it to your MCP client.

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "climate-economic-nexus": {
      "url": "https://climate-economic-nexus-mcp.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}

Cursor / Windsurf / Cline

Add the endpoint https://climate-economic-nexus-mcp.apify.actor/mcp in your MCP server settings with your Apify API token as the Bearer authorization header. The server is immediately available — no spin-up wait.

Direct HTTP (cURL)

# Call simulate_integrated_assessment directly
curl -X POST "https://climate-economic-nexus-mcp.apify.actor/mcp" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_APIFY_TOKEN" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
      "name": "simulate_integrated_assessment",
      "arguments": {
        "query": "global warming 2050 economic impact",
        "horizon_years": 80,
        "carbon_tax_init": 75
      }
    },
    "id": 1
  }'

Python

import httpx
import json

APIFY_TOKEN = "YOUR_APIFY_TOKEN"
MCP_URL = "https://climate-economic-nexus-mcp.apify.actor/mcp"

payload = {
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
        "name": "simulate_integrated_assessment",
        "arguments": {
            "query": "Southeast Asia flood risk economic damage 2050",
            "horizon_years": 100,
            "carbon_tax_init": 50
        }
    },
    "id": 1
}

response = httpx.post(
    MCP_URL,
    json=payload,
    headers={"Authorization": f"Bearer {APIFY_TOKEN}"}
)

result = response.json()
content = json.loads(result["result"]["content"][0]["text"])

print(f"Social Cost of Carbon: ${content['socialCostOfCarbon']:.2f}/tCO2")
print(f"Peak Warming: {content['peakWarming']:.2f}°C")
print(f"Year of 2°C: {content['yearOf2C']}")
print(f"Carbon Budget Remaining: {content['carbonBudgetRemaining']:.0f} GtCO2")

JavaScript

const APIFY_TOKEN = "YOUR_APIFY_TOKEN";
const MCP_URL = "https://climate-economic-nexus-mcp.apify.actor/mcp";

const response = await fetch(MCP_URL, {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    "Authorization": `Bearer ${APIFY_TOKEN}`
  },
  body: JSON.stringify({
    jsonrpc: "2.0",
    method: "tools/call",
    params: {
      name: "detect_tipping_cascades",
      arguments: {
        query: "AMOC collapse ice sheet tipping point",
        n_paths: 5000,
        horizon_years: 100
      }
    },
    id: 1
  })
});

const result = await response.json();
const data = JSON.parse(result.result.content[0].text);

console.log(`System Risk: ${(data.systemRisk * 100).toFixed(1)}%`);
console.log(`Cascade Probability: ${(data.cascadeProbability * 100).toFixed(1)}%`);
console.log(`Elements Near Bifurcation: ${data.tippedCount}`);

for (const element of data.elements) {
  console.log(`  ${element.name}: ${element.bifurcationType} (discriminant: ${element.discriminant.toFixed(3)})`);
}

Output examples

simulate_integrated_assessment

{
  "socialCostOfCarbon": 187.42,
  "peakWarming": 2.81,
  "yearOf2C": 2061,
  "carbonBudgetRemaining": 312.7,
  "discountedUtility": 4821.6,
  "optimalCarbonTaxPath": [
    50.0, 51.5, 53.1, 54.7, 56.3, 58.0, 59.7, 61.5, 63.4, 65.3
  ],
  "trajectorySnapshots": [
    {
      "year": 2025, "temperature": 1.10, "temperatureOcean": 0.33,
      "carbonAtm": 851.0, "carbonUpper": 460.0, "carbonDeep": 1740.0,
      "output": 1.08e14, "damages": 0.00287, "abatement": 0.00914,
      "netOutput": 1.06e14, "carbonTax": 50.0, "emissions": 36.2
    },
    {
      "year": 2035, "temperature": 1.42, "temperatureOcean": 0.46,
      "carbonAtm": 906.3, "carbonUpper": 471.2, "carbonDeep": 1742.1,
      "output": 1.23e14, "damages": 0.00478, "abatement": 0.01123,
      "netOutput": 1.20e14, "carbonTax": 67.2, "emissions": 31.8
    }
  ],
  "totalTimesteps": 100
}

detect_tipping_cascades

{
  "systemRisk": 0.347,
  "tippedCount": 2,
  "cascadeProbability": 0.183,
  "elements": [
    {
      "name": "AMOC",
      "state": -0.412,
      "controlA": -1.823,
      "controlB": 0.091,
      "discriminant": -21.6,
      "nearBifurcation": true,
      "bifurcationType": "fold",
      "criticalDistance": 0.048
    },
    {
      "name": "Amazon",
      "state": 0.218,
      "controlA": -0.724,
      "controlB": 0.034,
      "discriminant": -0.34,
      "nearBifurcation": false,
      "bifurcationType": "stable",
      "criticalDistance": 0.312
    }
  ],
  "cascadeChains": [
    { "trigger": "AMOC", "affected": ["Amazon", "PermafrostArctic"], "severity": 0.71 }
  ],
  "pathCount": 5000
}

forecast_carbon_price_regimes

{
  "currentRegime": 1,
  "regimeNames": ["low-stable", "policy-transition", "high-volatile"],
  "regimeProbabilities": [0.18, 0.61, 0.21],
  "forecastMean": 84.3,
  "forecastStd": 22.7,
  "forecast90CI": [48.2, 124.6],
  "jumpProbability": 0.094,
  "transitionMatrix": [
    [0.92, 0.07, 0.01],
    [0.05, 0.88, 0.07],
    [0.03, 0.12, 0.85]
  ],
  "filteredPrices": [52.1, 54.3, 58.8, 61.2, 67.4, 71.9, 75.0, 78.3, 81.7, 84.3]
}

Output fields

simulate_integrated_assessment

detect_tipping_cascades

quantify_damage_uncertainty

optimize_robust_adaptation

downscale_spatial_impacts

forecast_carbon_price_regimes

assess_biodiversity_economic_loss

attribute_climate_damages

How much does it cost to run climate-economic analysis?

All tools use pay-per-event pricing. Each tool call triggers a single charge event — compute costs are included. There are no subscriptions or minimum commitments.

The Apify Free plan includes $5 of monthly credits — enough for approximately 140–160 tool calls per month at no cost. You can set a maximum spending limit per session in your MCP client; the server checks the limit before charging and returns a structured error if the limit is reached rather than silently failing.

Compare this to enterprise climate data platforms (Bloomberg NEF, MSCI Climate, Verisk) that charge $50,000–$200,000 per year for comparable scenario modeling capabilities.

How Climate-Economic Nexus MCP works

Phase 1: Parallel data collection from 18 actors

Every tool call triggers parallel execution of up to 18 Apify actors via Promise.all. Each actor call has a 180-second timeout and 256 MB memory allocation. The actor client in actor-client.ts handles failure gracefully — a single failing data source returns an empty array rather than aborting the entire run. Results are bundled into typed collections (climateData, economicData, disasterData, biodiversityData, geoData) before being passed to the scoring engine.

Phase 2: Data-calibrated model initialization

Each mathematical model is calibrated from live data before running. The DICE model, for example, extracts numeric temperature anomalies from NOAA and GDACS records to set baseTemp, pulls GDP figures from World Bank and IMF to set baseGdp, and reads CO2 concentration data to initialize baseEmissions. If a data source returns no usable numeric values, the model falls back to published IPCC/Nordhaus baseline constants. This means results reflect actual current conditions rather than static parameter sets.

Phase 3: Algorithm execution

Eight distinct mathematical frameworks run after calibration:

DICE Integrated Assessment implements Nordhaus's three-reservoir carbon cycle: M_ATM(t+1) = φ₁₁·M_ATM + φ₂₁·M_UP + E(t) with transfer coefficients from the 2017 DICE calibration (φ₁₁ = 0.88, φ₁₂ = 0.0472, etc.). The two-box energy balance model advances temperature using T_ATM(t+1) = T_ATM + ξ₁·(F(t) − λ·T_ATM − ξ₃·(T_ATM − T_OCEAN)) with climate feedback λ = 1.18 and forcing coefficient η = 3.68 W/m². Carbon tax ramps at 3% per year from the initial value, with abatement fraction computed from μ = min(carbonTax / backstopPrice, 1).

Cusp Catastrophe Tipping parameterizes each of the 5 tipping elements using the Thom cusp potential V(x;a,b) = x⁴/4 + ax²/2 + bx. The discriminant Δ = 8a³ + 27b² determines proximity to the bifurcation set. Monte Carlo paths use Euler-Maruyama discretization of coupled SDEs with Heaviside coupling: x_i(t+dt) = x_i(t) + f_i(x)·dt + Σⱼ αᵢⱼ·H(xⱼ − θⱼ)·dt + σᵢ·√dt·Z. n_paths defaults to 5,000.

Gaussian Process Spatial Downscaling uses a Matern-3/2 kernel k(r) = σ²·(1 + √3·r/l)·exp(−√3·r/l) with Lanczos gamma function and modified Bessel K_ν approximation for the general Matern form. GP prediction follows μ* = K*ᵀ·(K + σₙ²·I)⁻¹·y with posterior variance Σ* = K** − K*ᵀ·(K + σₙ²·I)⁻¹·K*. The Cholesky-like inversion uses a seeded Mulberry32 PRNG for reproducibility given the same input query.

Hamilton Filter for Carbon Price Regimes performs forward recursion: ξ_{t|t} = (f(y_t|s_t) ⊙ ξ_{t|t−1}) / (1ᵀ·f(y_t|s_t) ⊙ ξ_{t|t−1}) across 3 regimes with jump-diffusion dynamics superimposed on regime-specific drift and volatility parameters.

Phase 4: Structured JSON response

Results are serialized to JSON and returned as MCP text content items. Large arrays are trimmed before returning (e.g., trajectory snapshots at every 10th step, spatial points capped at 50, filtered prices at last 20) to keep response sizes within MCP client limits while preserving analytical completeness.

Tips for best results

1. Include geographic specificity in queries. "coastal Bangladesh flood risk" produces better-calibrated results from NOAA, FEMA, and Nominatim than "flood risk". Nominatim resolves place names to coordinates that anchor the spatial downscaling grid.

2. Start with simulate_integrated_assessment for new topics. It calls all 18 data sources and returns the broadest cross-section of climate-economic context. Use the SCC and temperature trajectory outputs to frame more specific follow-up queries to other tools.

3. Use quantify_damage_uncertainty before citing numbers in reports. The tail risk at the 95th and 99th percentile is often 3–5× the mean estimate. The modelDisagreement field tells you how much inter-regional variance exists — high disagreement means regional breakdowns matter more than global averages.

4. Pair detect_tipping_cascades with attribute_climate_damages. Cascades identify which elements are near bifurcation; attribution tells you what fraction of the driving forcing is anthropogenic. Together they answer both "how close are we?" and "how much is our fault?".

5. Set horizon_years to 50 for near-term planning and 150 for long-term strategy. The DICE model runs annual timesteps, so 300 years (the maximum) adds computation time. For infrastructure with 30-year lifespans, 50-year horizons with a higher initial carbon tax are most informative.

6. Use forecast_carbon_price_regimes quarterly. Carbon markets are regime-dependent and transition probabilities change as new policy signals emerge. Running the tool with updated queries like "EU ETS reform 2025" or "California AB 32 auction" helps detect emerging regime shifts before they price in.

7. Chain downscale_spatial_impacts into asset-level risk. Pass the hotspots array (grid points with the highest predicted anomaly) to a geocoder or GIS system to identify which physical assets fall within high-anomaly zones.

Combine with other Apify actors

Limitations

• Models are simplified implementations, not IPCC-grade simulators. DICE, for example, uses annual timesteps and linear carbon tax optimization rather than full non-linear dynamic programming. Results are appropriate for scenario analysis and directional risk assessment, not regulatory submission without expert review.
• Data calibration depends on query quality. If a query returns no numeric values from a data source (e.g., NOAA returns no temperature records for an abstract query), that source falls back to baseline constants. Specific, geographically grounded queries produce better-calibrated results.
• All 18 actor calls run on each invocation. There is no caching between calls. Running the same query twice incurs the full per-event charge twice.
• Spatial downscaling grid resolution is limited to 1–10 degrees. Sub-degree or asset-level resolution is not available within this tool. For finer spatial analysis, export the hotspot coordinates and run them through a dedicated geospatial service.
• Carbon price forecasting covers near-to-medium term regime dynamics. Long-run structural carbon price modeling (decades-out absolute price levels) is not the intended use case; the regime-switching model is calibrated to identify transition probabilities over 1–5 year horizons.
• Attribution analysis applies to aggregate damage signals. The optimal fingerprinting implementation cannot attribute individual events in isolation — it operates on the statistical ensemble of disaster records returned by GDACS and FEMA. For single-event attribution, domain-expert counterfactual analysis remains necessary.
• Biodiversity percolation uses a 50×50 default lattice. Increasing lattice_size to 100 gives finer spatial resolution at the cost of O(n²) compute time. Lattice values above 100 are not recommended.
• Exchange rate normalization is USD-centric. Damage estimates are normalized to USD using live exchange rates. For multi-currency portfolio analysis, interpret regional damage fractions (percentage of GDP) rather than absolute dollar figures.

❓ FAQ

How many tool calls can I make per month on the free Apify plan? The free plan includes $5 of monthly credits. At $0.030–$0.040 per call, that is approximately 125–165 tool calls per month. Paid plans start at $49/month with significantly higher credit allowances.

Does Climate-Economic Nexus MCP use real climate models? The mathematical frameworks implement published peer-reviewed methodologies: the Nordhaus DICE integrated assessment model, Thom cusp catastrophe theory for tipping element analysis, Allen and Stott optimal fingerprinting for attribution, and Lempert PRIM for robust decision-making. They run on real data from 18 authoritative sources. Results are scenario analyses, not deterministic predictions, and should be interpreted by domain-qualified professionals for high-stakes decisions.

How accurate are the temperature and damage projections? Projections reflect calibrated model runs, not ensemble IPCC outputs. Uncertainty is explicitly quantified: quantify_damage_uncertainty returns 95th and 99th percentile tail risks; detect_tipping_cascades returns cascade probability distributions across 5,000 Monte Carlo paths; downscale_spatial_impacts returns posterior standard deviation at every grid point. Use the uncertainty bands, not just the point estimates.

What time horizons does the server cover? simulate_integrated_assessment supports up to 300-year horizons (default 100 years, annual timesteps from 2025). forecast_carbon_price_regimes covers near-to-medium term regime transitions (1–5 years). downscale_spatial_impacts produces a current-state snapshot rather than a forward projection. detect_tipping_cascades runs over the specified horizon_years (default 100 years).

Can I use this for TCFD, EU Taxonomy, or other regulatory climate disclosures? The tools produce scenario analysis outputs consistent with TCFD physical risk frameworks and support qualitative scenario narratives required by the EU Taxonomy. Compliance filings require review by qualified climate risk professionals. The server does not produce regulatory filings — it produces the quantitative inputs that support them.

How is this different from Bloomberg NEF or MSCI Climate? Bloomberg NEF and MSCI Climate are enterprise platforms with human analyst teams, proprietary models, and annual licenses starting at $50,000+. This MCP server is a self-service API that implements published open-source methodologies on top of publicly available data. It is suitable for teams building climate analytics into their own tools, AI workflows, or research pipelines without enterprise licensing costs.

Is it legal to use the underlying data sources? All 18 data sources — NOAA, USGS, GDACS, OpenAQ, World Bank, IMF, FRED, OECD, Eurostat, GBIF, IUCN, FEMA, and others — provide publicly available data under open or government open data licenses. Use of this server does not involve scraping private or paywalled content. See Apify's guide on web scraping legality for general guidance.

How long does each tool call take? Tool call duration depends on data source response times. The server fans out all actor calls in parallel, so wall-clock time is approximately equal to the slowest responding data source (typically 30–120 seconds). simulate_integrated_assessment calls 18 actors and is typically the slowest at 60–150 seconds. forecast_carbon_price_regimes calls fewer high-latency economic sources and typically completes in 30–90 seconds.

Can I run multiple tools in sequence in one agent conversation? Yes. Each tool call is independent. A common pattern is: run simulate_integrated_assessment first for context, then detect_tipping_cascades for cascade risk, then quantify_damage_uncertainty for tail-risk bounds. The spending limit check on each tool prevents accidental overrun.

What happens if one of the 18 data sources is unavailable? The actor client handles failures gracefully. Each individual actor call is wrapped in a try-catch that returns an empty array on failure. The model then falls back to baseline constants for the missing data source. Results are degraded but not broken. The tool logs which actors failed so you can investigate data source availability.

Can I connect this server to a custom AI agent framework? Yes. The /mcp endpoint implements the Model Context Protocol (MCP) over HTTP with StreamableHTTP transport. Any framework that speaks MCP — including LangChain with the MCP adapter, LlamaIndex, AutoGen, and custom agent loops — can connect to it. The server also accepts raw tools/call JSON-RPC requests, so it works with any HTTP client without an MCP library.

How is the DICE social cost of carbon calculated? The social cost of carbon is computed as the present value of the marginal damage from one additional GtCO2 of emissions, using the Nordhaus quadratic damage function Ω(T) = 1/(1 + π₁·T + π₂·T²) with π₂ = 0.00236, discounted at ρ = 1.5% pure time preference. The three-reservoir carbon cycle uses the DICE-2016 transfer coefficients (φ₁₁ = 0.88) to track atmospheric carbon buildup.

Help us improve

If you encounter unexpected results or a tool call fails, enable run sharing so we can diagnose faster:

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

This lets us see the run logs and data source responses when something goes wrong. Your data is only visible to the actor developer, not publicly.

Integrations

• Zapier — Trigger climate scenario analysis from Zapier workflows; pipe social cost of carbon outputs into Google Sheets or Airtable for reporting
• Make — Automate monthly carbon price regime monitoring; route regime-transition alerts to Slack or email
• Google Sheets — Export DICE trajectory snapshots and regional damage estimates directly into spreadsheets for board-level reporting
• Apify API — Trigger analysis programmatically from Python or JavaScript pipelines; integrate into existing climate risk platforms
• Webhooks — Receive notifications when long-running DICE simulations or cascade analyses complete
• LangChain / LlamaIndex — Register all 8 tools as LangChain tools or LlamaIndex query engines for climate-aware RAG pipelines

Troubleshooting

Tool returns default/baseline values despite a specific query. This usually means the query returned no numeric data from one or more data sources. Check that your query contains geographic context and relevant domain terms (e.g., "Amazon deforestation tipping 2050" rather than "tipping points"). Abstract queries like "climate" produce fewer numeric matches across NOAA, GDACS, and World Bank.

Tool call times out in your MCP client. The server fans out to 18 actors in parallel with a 180-second timeout each. Total wall-clock time can reach 2–3 minutes for tools querying all 18 sources. Configure your MCP client's tool timeout to at least 300 seconds. Claude Desktop's default timeout is 60 seconds — increase it in your client settings if possible.

Spending limit reached error on the first call. This means the eventChargeLimitReached flag fired before the tool ran. Check your Apify account balance and top up credits, or increase the run spending limit in your actor configuration. The free plan's $5 of credits is the most common cause.

cascadeProbability is 0.0 for all tipping elements. The cascade simulation uses a seeded Mulberry32 PRNG initialized from the query string hash. Very short or single-word queries may produce degenerate parameter initializations. Try a more descriptive query: "AMOC Atlantic thermohaline circulation weakening Arctic warming" produces a richer parameterization than "AMOC".

attributableFraction returns exactly 1.0. This indicates the natural variability baseline in the disaster data was near zero, making the anthropogenic fraction dominate by default. Check that your query includes a specific extreme event type (e.g., "2024 hurricane season damage attribution") so GDACS and FEMA return varied disaster records for the baseline calibration.

Responsible use

• All 18 underlying data sources provide publicly available data under open or government open data licenses.
• Model outputs are probabilistic scenario analyses. Do not present them as deterministic forecasts in regulatory, legal, or financial contexts without expert review.
• Comply with applicable data protection and financial regulations when using outputs in investment decisions or public disclosures.
• For guidance on web scraping legality, see Apify's guide.

Support

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