Rating systems must differentiate risk meaningfully with obligor and facility grades, assign every exposure before approval of credit and review assignments at least annually, ensure integrity of assignment (independent approval, overrides documented), document design and operation, keep historical data on ratings, defaults and losses, run sound stress tests of capital adequacy, base PD, LGD and conversion factor estimates on long data histories with margins of conservatism, and use the systems in risk management, credit approval, internal capital allocation and governance. Institutions must validate estimates robustly: compare realised default rates with estimated PDs per grade (and realised LGDs and conversion factors with estimates), analyse deviations, use other quantitative tools and external data, and update the analysis and documentation at least annually.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.