Comparing financials across major payment processors is rarely as simple as pulling identical line items. A recent workflow demonstration using Claude and the Equibles Model Context Protocol (MCP) server highlights the subtle traps embedded in financial data, specifically mismatched fiscal calendars and restated figures. The example focuses on Visa (V), Mastercard (MA), American Express (AXP), and PayPal (PYPL), showcasing how an AI agent can be prompted to handle these nuances rather than blindly aggregating numbers.

Handling Fiscal Calendar Discrepancies

The first major pitfall identified is the variation in fiscal year-end dates. When querying FY2025 revenue and net income, the Equibles tool correctly identifies that Visa’s fiscal year ends on September 30, 2025, whereas the other three companies end theirs on December 31, 2025. This means Visa’s 'FY2025' actually covers a period from October 2024 to September 2025. The tool prevents this error from slipping through by explicitly displaying the period end date for each entry, allowing the analyst to adjust comparisons to a common timeframe, such as the twelve months ended June 30, 2026.

The Danger of Restated Figures and Valuation Nuances

Growth calculations require careful handling of restated figures. The workflow demonstrates retrieving FY2024 data for the same four companies, where the values are sourced from newer 10-K filings that may have restated previous numbers. This ensures that growth rates are calculated on a consistent accounting basis. Furthermore, the valuation multiples reveal critical distinctions between business models. American Express, acting as a lender, has an enterprise value (EV) structure that differs fundamentally from network operators like Visa and Mastercard. Consequently, EV/EBIT ratios are often irrelevant or blank for lenders, while P/E ratios remain comparable. The tool correctly leaves gaps rather than providing misleading estimates when data is missing, such as Visa’s P/E being blank due to unavailable diluted EPS for the period.

Key Takeaways

  • Always verify fiscal year-end dates when comparing companies; Visa’s September end date creates a quarter-shift compared to Mastercard, Amex, and PayPal.
  • Use restated figures from the latest 10-K filings to calculate accurate year-over-year growth rates.
  • Recognize that business model differences (lender vs. network) invalidate certain valuation ratios like EV/EBIT for companies like American Express.
  • Leverage the Equibles MCP server’s explicit handling of missing data (blanks instead of estimates) to avoid false precision.

The Bottom Line

This workflow proves that AI agents are only as good as their context and tools; without an MCP server that understands fiscal nuances, LLMs would likely hallucinate comparability where none exists.