The bank must include in its models all risk factors used for pricing and those in the standardised approach, justifying omissions, must not model securitisations, and must use at least six yield curve risk factors for material exposures in major currencies (MAR31.2-31.6). A risk factor is modellable only if it passes the RFET quarterly: at least 24 real price observations per year with no 90-day period of fewer than four in the last 12 months (monitored monthly), or at least 100 observations over the previous 12 months, counting one per day (MAR31.12-31.13). Backtesting and PLA must be passed at bank-wide and desk level, with a one-year report for approval (MAR32.1, MAR32.3), and at least 10% of aggregate market risk capital must come from qualifying desks, checked quarterly (MAR32.2). Bank-wide backtesting uses a 99th percentile one-day VaR over 250 days with green, amber and red zones and multipliers from 1.50 to 2.00 (MAR32.5-32.9); every exception must be documented (MAR32.12). A desk with more than 12 exceptions at 99% or 30 at 97.5% in 12 months moves to the standardised approach (MAR32.18-32.19). PLA compares RTPL with HPL over 250 days: green needs Spearman correlation above 0.80 and KS below 0.09; red is correlation below 0.7 or KS above 0.12, forcing standardised capital (MAR32.35-32.43).
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.