The Basel Committee's post-crisis overhaul of trading-book capital — the Fundamental Review of the Trading Book (FRTB). It redraws the boundary between the banking book and the trading book to curb regulatory arbitrage, and replaces the Basel 2.5 regime with two revised approaches: a risk-sensitive standardised approach (SA) and an internal models approach (IMA). The headline methodological change is that the IMA's risk measure moves from 99% Value-at-Risk to 97.5% Expected Shortfall, calibrated to a period of stress and scaled by liquidity horizons, with model approval granted per trading desk subject to backtesting and a profit-and-loss attribution test.
"At the trading desk level, backtesting must compare each desk's one-day VaR measure ... at both the 97.5th percentile and the 99th percentile, using at least one year of current observations of the desk's one-day P&L."
This is where the academic argument the wiki records — Artzner-Delbaen-Eber-Heath showed VaR is not subadditive and expected shortfall is its coherent repair — became binding capital regulation for the world's banks. FRTB is the concrete answer to "so what if VaR isn't coherent?": the IMA charge is now 97.5% ES, and the document's real complexity is the machinery bolted around that switch — liquidity horizons (because tail risk is inseparable from how long you're stuck holding), stressed calibration, the non-modellable-risk-factor regime, and desk-level approval gated by backtesting and P&L attribution. The one genuine tension it embodies is the open problem noted elsewhere in the wiki: ES is harder to backtest than VaR (it is not elicitable in isolation), which is exactly why the standard still backtests VaR at two percentiles even though it capitalises on ES — a pragmatic compromise between the theoretically preferred measure and the one you can actually validate against realized P&L. For the wiki it is the applied, regulatory capstone of the risk-measure thread.