ICE Data Services MCP: Bond Trade Data for Claude

ICE Data Services Brings Bond Trade Data Directly Into Claude

In mid-2026, Intercontinental Exchange (ICE) launched an official Claude MCP Connector for ICE Data Services — the first major financial data provider to deploy a native Model Context Protocol integration with Claude. For licensed ICE clients, this connector surfaces FINRA TRACE corporate and securitized bond trade data, MSRB RTRS municipal bond trades, credit ratings, coupon details, and maturity reference data directly inside Claude conversations and Claude Code sessions.

The significance is practical: instead of downloading TRACE reports, running Excel lookups, and switching between terminals, a portfolio manager or compliance analyst can ask Claude natural-language questions about bond trade activity and receive structured answers drawn from authoritative ICE data in real time. This page covers what the connector does, how it works, and the most productive Claude prompt patterns for fixed income teams.

What Data the ICE MCP Connector Covers

The connector exposes three primary data domains:

FINRA TRACE — Corporate, Agency, and Securitized Bond Trades

FINRA's Trade Reporting and Compliance Engine (TRACE) is the authoritative post-trade transparency system for U.S. over-the-counter fixed income markets. ICE's MCP connector allows Claude to query TRACE trade data covering investment-grade and high-yield corporate bonds, agency bonds, mortgage-backed securities, asset-backed securities, and collateralized mortgage obligations. Data coverage begins January 2026 (some ICE sources cite September 2025 for certain products — confirm with your ICE relationship manager).

Query performance considerations: single-day and weekly date ranges return fastest. Queries spanning more than one month may time out — plan multi-month analysis as sequential weekly queries or use ICE's bulk data products for longer lookups.

MSRB RTRS — Municipal Bond Trade Data

The Municipal Securities Rulemaking Board's Real-Time Transaction Reporting System (RTRS) is the primary transparency mechanism for the $4 trillion municipal bond market. ICE's connector exposes muni trade data through the same interface — allowing analysts to compare muni trade volume, pricing, and spread relationships against corporate bond data without switching between systems.

Bond Reference Data — Ratings, Coupons, and Maturities

Beyond trade data, the connector includes a structured reference data layer: CUSIP-level bond characteristics including Moody's/S&P/Fitch credit ratings, coupon rate and type (fixed/floating/zero), maturity date, sector and sub-sector classification, and callable/puttable features. A built-in knowledge layer provides ICE's definitions for financial thresholds and classifications — ensuring consistent, auditable responses when Claude categorizes bonds into investment-grade vs. high-yield, or short vs. intermediate vs. long duration buckets.

Setting Up the ICE MCP Connector in Claude

The connector is delivered as a standard MCP server. Once ICE provisions access, configuration follows the same pattern as other Claude MCP integrations:

1. Obtain your ICE MCP credentials from the ICE Data Services portal or your account manager. Credentials include an API endpoint URL and an OAuth token or API key.
2. Add the server to your Claude Desktop claude_desktop_config.json under "mcpServers", specifying the ICE endpoint URL and your authorization header.
3. Restart Claude Desktop. The ICE tools should appear as available in the tool panel — typically named query_trace_trades, query_msrb_trades, get_bond_reference, and related functions.
4. For team deployment, Claude Code users can add the MCP server to the project's CLAUDE.md configuration for consistent access across the team.

ICE's connector also supports the Claude API for programmatic workflows — useful for building internal tools that embed bond data queries into risk dashboards or compliance monitoring systems.

Corporate Bond Analysis Workflows

Fixed income analysts and portfolio managers use the ICE MCP connector most heavily for spread analysis, trade surveillance, and credit monitoring. The following prompts illustrate productive patterns:

• "Use the ICE TRACE data to pull all corporate bond trades for Ford Motor Credit (FORD) cusips in the past week. Compute: (1) total traded volume by day; (2) volume-weighted average price (VWAP) for each day; (3) implied yield from VWAP and maturity date; (4) spread over the interpolated Treasury benchmark. Flag any days where daily volume was more than 2x the weekly average — those are likely news-driven."
• "I'm running a credit monitor on our HY corporate bond portfolio. For each of these 12 CUSIPs, pull the last 5 trading days of TRACE data and compute: daily VWAP yield, VWAP spread over T+5Y, day-over-day spread change in bps, and 5-day spread change. Rank by largest 5-day spread widening. Flag any position with more than 50 bps of spread widening — those need credit team review."
• "Compare the TRACE trade activity for two similar HY issuers: Issuer A (Ba2/BB, 5.75% 2029) and Issuer B (B1/B+, 6.25% 2029) over the past month. Analyze: (1) trade frequency (number of trades per day), (2) average trade size distribution, (3) bid-offer spread implied by customer buy vs. sell trade price differentials, (4) any divergence in price trend. Which issuer shows better secondary market liquidity?"
• "Pull TRACE data for the CCC-rated corporate bond segment (filtered by credit rating from ICE reference data) over the past two weeks. Compute: total daily volume, number of unique CUSIPs trading, average trade size, and proportion of volume in tickets under $1MM face value (retail) vs. over $1MM (institutional). Compare Monday vs. Thursday pattern — is retail participation higher on Mondays?"

Municipal Bond Analysis with MSRB Data

Municipal bond markets are less liquid and less transparent than corporates — which makes the MSRB RTRS data surfaced through the ICE connector particularly valuable for muni portfolio managers and advisors:

• "Query MSRB RTRS data for all trades in California GO bonds (state general obligation, rated Aa2/AA-) over the past 5 trading days. Compute: daily traded volume, VWAP yield by maturity bucket (1-5Y, 5-10Y, 10-20Y, 20-30Y), and spread over the MMD (Municipal Market Data) benchmark. Has the California GO curve steepened or flattened over this period?"
• "I need a liquidity analysis for these 8 hospital revenue bonds in our portfolio. For each CUSIP, pull the past 30 days of MSRB RTRS data and compute: total number of trades, total face value traded, average days between consecutive trades, and median trade size. Rank by days between trades (least liquid first). Flag any bond that has traded fewer than 3 times in 30 days — those need a liquidity premium review."
• "Compare the MSRB trade data for muni bonds in two tax brackets: (1) AMT-exempt munis vs. (2) AMT-subject munis, both 10-year maturity range, BBB-rated. Over the past 2 weeks: how large is the yield premium on AMT-subject bonds? Has this premium changed? What does trade volume tell us about investor demand for each type?"
• "Use ICE reference data to identify all Puerto Rico GO bonds and COFINA bonds in the MSRB RTRS system. Pull 5-day trading data. Compute VWAP yield and spread over the 10-year MMD for both issuer types. Has post-restructuring trade activity normalized, or do these still trade at crisis-era spreads? How does recent volume compare to par outstanding?"

Credit Surveillance and Watchlist Monitoring

Compliance teams and credit risk managers use the ICE connector to build automated surveillance workflows — flagging unusual trade patterns that may indicate market intelligence or pre-announcement activity:

• "Build a credit surveillance report for our 25-bond HY watchlist. For each CUSIP, use ICE TRACE data to check the past 10 trading days for: (1) unusual volume spikes (any day where volume exceeds 3x the 10-day average), (2) price moves of more than 2 points in a single day, (3) spread widening of more than 75 bps over the 10-day period. List all flags with date, magnitude, and direction (widening vs. tightening). This feeds our morning credit meeting."
• "Run a pre-earnings trade pattern analysis for XYZ Corp (insert CUSIP). Pull TRACE data for the 15 trading days before each of the last 4 quarterly earnings dates. For each pre-earnings window: (1) average daily volume vs. prior 30-day baseline, (2) trend in VWAP spread (widening or tightening into earnings), (3) any large individual trades (over $10MM face). Identify if there's a consistent pre-earnings pattern and what it implies about how bond investors position ahead of XYZ's reports."
• "I need to document an unusual bond trade for our compliance file. CUSIP XYZ123 saw $50MM face trade at 3:42 PM last Tuesday, 2 points below the prior VWAP, with 3x normal volume. Use ICE reference data to provide: (1) bond characteristics (issuer, coupon, maturity, rating), (2) TRACE trade detail context (time, price, size), (3) comparison to 30-day trade history, (4) any material news events that day per ICE's knowledge layer. Format this as a compliance surveillance memo."

Portfolio-Level Analytics Combining TRACE and Reference Data

The most powerful workflows combine ICE's trade data with its bond reference data to produce portfolio-level analytics:

• "I have a 40-bond corporate portfolio (paste CUSIP list). Use ICE reference data to classify each bond by: rating category (IG: AAA-BBB- / HY: BB+ and below), maturity bucket (0-3Y, 3-7Y, 7-15Y, 15Y+), sector (financials, industrials, utilities, TMT). Then use TRACE data to pull this week's VWAP spread for each. Compute: (1) portfolio average spread by rating category and maturity bucket, (2) duration-weighted average spread, (3) contribution to spread by sector. Output as a table — this is my weekly portfolio report."
• "Compare the spread levels in my portfolio against the ICE BofA Corporate Bond Index benchmark (investment-grade). For each of my 40 IG bonds, pull the TRACE VWAP spread. Also get the current ICE BofA IG index OAS from ICE reference data. For each bond: (1) spread vs. index OAS, (2) over/under-performing the index by how many bps, (3) which bonds are >50 bps wide of the IG index on a rating-adjusted basis. Show which sectors are cheapest and richest vs. the benchmark."

Integration with Other MCP Data Sources

FinSkilz's MCP ecosystem allows combining ICE bond trade data with other financial data sources in a single Claude session. Particularly productive combinations:

• ICE + SEC EDGAR MCP: Pull a company's latest 10-Q credit metrics from EDGAR, then query ICE TRACE to see how the bond market has priced credit risk before and after the filing. Ask Claude to reconcile the accounting-based credit metrics with market-implied spreads.
• ICE + Portfolio Risk MCP: Feed TRACE spread data into FinSkilz's Portfolio Risk MCP to compute spread VaR, run spread-based stress tests (e.g., what if IG spreads widen 100 bps?), and generate full portfolio credit risk reports.
• ICE + OFAC MCP: For compliance teams: pull bond reference data from ICE to identify issuers, then cross-reference against OFAC's SDN list using FinSkilz's OFAC MCP to flag any sanctioned entities in bond portfolios or trade counterparty lists.

Limitations and What ICE MCP Doesn't Cover

Understanding the connector's scope prevents misuse. Key limitations:

• No live pricing: The connector surfaces trade data (completed transactions), not real-time bid/offer quotes or dealer runs. For pre-trade pricing, ICE's evaluated pricing products are separate from the MCP connector.
• Data lag: TRACE data is disseminated on a 15-minute delay for most transactions (same-day T+0 for FINRA member firms). The MCP connector surfaces post-dissemination data, so "real-time" queries reflect T-15 minutes, not instantaneous.
• License-gated data: ICE's full dataset — including 144A-only bonds, restricted securities, and real-time data — requires specific license tiers. Confirm your entitlements with your ICE account team before building workflows that depend on data you haven't confirmed access to.
• Query timeouts for long date ranges: Multi-month queries across large CUSIP sets can time out. Structure analysis as rolling 5-day or weekly queries and aggregate in Claude for longer-period analysis.