What Is FRTB? Trading Book, Expected Shortfall & the Math

FRTB, the Fundamental Review of the Trading Book, is the Basel Committee's rewrite of how banks hold capital against market risk. The two questions on the agenda are simple: does it apply to us, and what would it cost? This post uses a fictional regional bank, Ridgeway Regional, to explain the spirit of the rules, the two approaches, the basic math, and where the United States stands as of mid-November 2026.

<iframe width="560" height="315" src="https://www.youtube.com/embed/FWOSRVbQj8k" title="What Is FRTB? Trading Book, Expected Shortfall & the Math" frameborder="0" allowfullscreen></iframe>

Watch on YouTube

Why FRTB exists

Ridgeway Regional is a fictional bank with about $9 billion in assets, mostly loans and deposits. It sells interest-rate swaps to commercial borrowers and hedges them with dealers, and it runs a small bond trading account: about $150 million of trading assets and liabilities. Every Ridgeway number is made up for teaching.

Looking back at the 2007-08 crisis, the Basel Committee wrote that capital for trading activities proved insufficient to absorb losses. In 2009 came a quick patch, often called Basel 2.5, which added capital, including a stressed version of VaR, or value at risk. Then the full rewrite: FRTB, published in January 2016 and revised in January 2019. Basel's agreed start date was January 1, 2023, but each country writes its own rules and sets its own timing.

The spirit: capital that matches real trading risk, including the bad tail and positions you can't sell quickly.

Four weak spots FRTB targets

1. A blurry line between the trading book and the banking book, so positions could drift to wherever capital was lower.
2. VaR shows where the bad tail starts, not how deep it goes. VaR is the loss line on a normal bad day.
3. Liquidity. The old rules assumed you could exit positions over ten days. The crisis proved that false.
4. A standardized approach too crude to be a credible fallback when models fail.

The boundary: why do you hold this position?

FRTB starts with one question: why do you hold this position? If it's for short-term resale, short-term profit, arbitrage, or to hedge those positions, it goes in the trading book, marked to market every day. Everything else, like loans held to collect, lives in the banking book under the credit-risk rules.

Switching books is rare, needs approval, and can't cut your capital. Any reduction comes back as a surcharge.

Two roads to market-risk capital

The standardized approach is the default: the regulator's formulas. Internal models are a privilege, approved desk by desk. Even banks that use models must still calculate the standardized number for every desk, as a fallback and a benchmark. And Basel offers a simplified version for banks with small, simple trading books.

The standardized approach, in numbers

The standardized approach has three building blocks.

One: the sensitivities-based method. Measure how much each position moves when a risk factor moves, then multiply by a risk weight set by Basel.

Two: the default risk charge, for the jump-to-default loss on bonds and other credit positions.

Three: the residual risk add-on, a small charge for exotic payoffs the formulas can't capture.

Add the three together, and that's standardized capital.

Here's the core math. Say Ridgeway's net swap position loses $25,000 for every one-basis-point rise in 10-year rates. That's its sensitivity. Basel's risk weight for a 10-year rate is 1.1 percentage points. Think of it as a 110-basis-point shock. Multiply: $25,000 times 110 is $2.75 million. That's the capital for this one risk. Real portfolios combine many of these with set correlations, run three ways, low, medium and high, keeping the worst.

Next, default risk. Say the bond account holds $10 million of a triple-B rated senior corporate bond. If the issuer defaulted, Basel assumes you lose 75% of a senior bond: $7.5 million. Triple-B carries a 6% default risk weight, so the charge is $450,000.

And the residual risk add-on is simple: 1% of notional for exotic underlyings, like weather, and 0.1% for other hard-to-replicate payoffs.

Internal models: expected shortfall, horizons, and desk tests

FRTB swaps VaR for expected shortfall at 97.5%: the average loss on the worst 2.5% of days. In a perfect bell curve, the two are almost identical: 99% VaR is 2.33 standard deviations, and 97.5% expected shortfall is 2.34. But real markets have fat tails. When the worst days get worse, VaR can stay put, while expected shortfall rises. In calm markets, it's about the same number. In a crisis, expected shortfall sees how deep the tail goes. And it's calibrated to a stressed 12-month period, not just last year.

FRTB also asks: in a stressed market, how long would it take to get out of this risk? That's the liquidity horizon. Interest rates in major currencies like the dollar: 10 days. Investment-grade corporate credit spreads: 40 days. High yield: 60. Some credit risks: 120. The model starts from a 10-day expected shortfall and scales up. For a single risk factor, multiply by the square root of the horizon over 10. So 40 days doubles it.

Model approval now happens desk by desk, and each desk has to keep passing two tests. Backtesting: more than 12 exceptions in a year at 99%, or more than 30 at 97.5%, and the desk loses its model. And the profit-and-loss attribution test: does the risk model's P&L track the desk's own P&L? Green needs a rank correlation above 0.8 and a close match in the distributions. Fail backtesting, or land in the attribution red zone, and that desk drops to the standardized approach.

One more model rule: every risk factor needs real data. Roughly, at least 24 real prices a year, with no long gaps. If not, it's a non-modellable risk factor. It's pulled out of the model and charged with its own stress scenario, which can be expensive.

Does this apply to a regional bank?

In the U.S. today, the market risk rule is still VaR-based, and it applies once trading assets plus liabilities reach $1 billion, or 10% of total assets.

On March 19, 2026, the Fed, the FDIC and the OCC re-proposed the Basel III endgame, including an FRTB-style market risk rule: expected shortfall for models, and a standardized default. As proposed, it would cover the largest banks, plus any bank with more than $5 billion of trading activity, or trading of at least 10% of its assets. Comments closed on June 18. As of mid-November 2026, there is still no final rule and no start date.

For Ridgeway, $150 million of trading is about 1.7% of assets. That's outside the rule today, and outside it as proposed. Track the thresholds as your swap business grows, and plan before you cross one. Document why every position sits in its book. That's good discipline either way. Remember that your dealers' capital rules can show up in the hedge prices they quote you. And borrow the ideas: sensitivities, tail risk under stress, and how long it takes to get out.

Five questions for your next ALCO

1. Which positions are in our trading book, and why?
2. How far are we from the market risk rule's thresholds?
3. What's our biggest sensitivity, and what shock would Basel apply to it?
4. Are dealer hedging costs changing what we pay?
5. If trading grows, could we run the standardized approach on day one?

Recap

FRTB is Basel's post-crisis rewrite of market-risk capital. Boundary first: the reason you hold a position decides its book. Standardized capital is sensitivities times risk weights, plus a default charge and a residual add-on. And models mean expected shortfall at 97.5%, liquidity horizons, and desk-level tests.

Watch more Bank Risk Decoded videos on the Bank Risk Decoded YouTube channel.