Expected Shortfall Calculator: CVaR for TradingView Backtest Results

I ran a strategy for six months that looked great on win rate. Then I checked the worst five trades and realized my average loser was a lot worse than I thought. Expected Shortfall, also called CVaR (Conditional Value at Risk), is the measure that caught it.

Expected Shortfall asks: "In the worst 5% (or 1%) of my trades, how much do I lose on average?" Unlike Value at Risk (VaR) which gives you a threshold (the 95th percentile loss), Expected Shortfall looks past that threshold at the whole tail. If VaR says your 95th percentile loss is $500, Expected Shortfall tells you the average of all losses beyond that $500 mark. If the tail is fat, Expected Shortfall can be dramatically higher than VaR.

The key difference is simple: VaR tells you the minimum loss in the tail. Expected Shortfall tells you the average loss inside the tail. CVaR is always at least as strict as VaR, and when a strategy has a few blowout losers it catches risk that VaR misses entirely.

CVaR is expressed in the same unit as your trades (dollars or percentage). A positive CVaR means you have net gains even in the tail (rare with directional strategies). A negative CVaR is the expected loss per trade in the tail.

Lower (more negative) is worse. A CVaR_{95%} of -$200 means your worst 5% of trades average a $200 loss. The same strategy might have VaR_{95%} of only -$120, giving a false sense of safety.

The gap between VaR and CVaR matters. If VaR_{95%} is -$120 and CVaR_{95%} is -$400, your tail is heavy with outlier losses. If they are close (VaR -$120, CVaR -$150), the loss distribution is well-behaved and the worst trades are not much worse than the threshold.

At 99% confidence, CVaR will naturally be more extreme because it isolates the very tail. I always check both levels when I upload a new CSV. The difference between the two tells me whether the risk is concentrated in a few extreme trades or spread across a broader tail.