The Trap of Volatility Drag

Volatility Drag, also called 'Variance Drain,' is the phenomenon whereby the larger the swings in price, the lower the actual compound growth rate becomes compared to the simple average return. It's a concept named by Tom Messmore in a 1995 piece titled 'Variance Drain.'

The key lies in the fact that the 'simple average (arithmetic mean)' and the 'actual compounding (geometric mean)' differ. The 'average annual return' we casually cite is often the arithmetic mean, but the money that actually grows in your account follows the geometric mean. And as long as there's volatility, the geometric mean is always less than or equal to the arithmetic mean.

Simply put, the more an asset churns, the more you can fall into the trap where 'the average looks plausible but what you end up with is less.'

Let's look at the most intuitive example. You start with ₩1 million.

Day 1: +10% → ₩1.1 million Day 2: -10% → ₩990,000

You clearly experienced '+10%' and '-10%' equally, yet the result is not ₩1 million but ₩990,000. The reason is simple. Day 2's -10% applies not to the ₩1 million principal but to the grown ₩1.1 million. So the loss amount is larger than the gain amount.

This small ₩10,000 difference is exactly the identity of volatility drag. The more the ups and downs repeat, the more this tiny loss keeps piling up.

The real reason volatility drag is scary lies in the 'asymmetry of losses.' Rising back by the amount it fell does not restore the principal.

-10% loss → needs +11% to recover -25% loss → needs +33% to recover -50% loss → needs +100% to recover -75% loss → needs +300% to recover

For an asset cut in half (-50%) to return to principal, it has to rise 'double (+100%).' The deeper the loss, the more the required recovery rate grows exponentially. So once you take a big drawdown, no matter how good the years that follow, it's hard for your actual money to grow by the 'average return.'

This is why you must weigh the maximum drawdown (MDD) as heavily as the return.

Volatility drag is roughly expressed by this approximation.

Geometric mean (actual compounding) ≈ arithmetic mean − (standard deviation² ÷ 2)

That is, the compound return you actually end up with is close to the simple average minus 'half the variance' (this is an approximation under assumptions of continuous compounding and small volatility, not an exact equality). The key is that the larger the volatility, the more sharply the amount you must subtract grows.

This problem shows up in the extreme with leveraged products (e.g., 2x and 3x leveraged ETFs). Scaling leverage by L times raises expected returns by L times (linearly), but volatility drag grows roughly in proportion to L squared. Doubling leverage roughly quadruples the drag.

According to one asset manager's data, a 3x leveraged ETF assuming 48% volatility could leak about 11.5% a year as volatility-drag cost. There's also an illustrative calculation that in a situation of very high volatility (e.g., assuming 50% a year), even an arithmetic mean of +10% could shrink to less than half the principal over the long run. Such figures are just examples under specific assumptions, but the direction is clear.

Volatility drag is a trap that's easy to miss if your eyes are fixed only on the number 'average return.' The conclusion of this article isn't to buy or sell a particular product, but that when looking at an asset, you should always look at the 'average return' and the 'volatility (and the maximum drawdown)' together.

The more volatile an asset, the lower the actual compound performance that stays in your account can be, even for the same average return. Conversely, diversification that lowers volatility helps protect long-term compounding by reducing the drag.

On 'The Return of Almost Everything,' you can directly simulate how the drawdown and recovery period during crisis windows affected the actual performance when you rolled the same period as a lump sum or as recurring investing. Check with your own eyes the 'result that actually rolled out,' not the average.