The number that fools people is the average. Somebody shows me a track record, or an asset's history, and the headline is the mean annual return. Say it averaged 20 percent a year. The instinct is to think a dollar left alone for a decade turns into whatever 20 percent compounded gets you. But the average return and the return you actually walk away with are two different numbers, and the gap between them is almost entirely a function of how bumpy the ride was. The bumpier the asset, the bigger the gap, and it grows faster than most people expect.
This is volatility drag, and once you see it you cannot unsee it. It is the single most underappreciated reason that a high-flying, wildly swinging asset can average a great number on paper and still leave you with less money than a boring one that averaged less.
The simplest version of the problem
Take an asset that goes up 50 percent one year and down 50 percent the next. The arithmetic average of those two returns is zero. Sounds like you broke even. You did not. A dollar goes to 1.50, then loses half to land at 0.75. You are down 25 percent over two years while the average return says you made nothing.
The reason is that percentage gains and losses are not symmetric. A 50 percent loss requires a 100 percent gain to recover, not another 50 percent. Losses compound against you harder than gains compound for you, and the more your returns swing around, the more this asymmetry bleeds you. The average return is an arithmetic mean. The return you keep is a geometric mean, the actual compounded growth rate, and the geometric mean is always lower than the arithmetic mean whenever there is any volatility at all. If returns never varied, the two would be equal. They never do, so it is always a tax.
The sigma-squared-over-two shortcut
There is a clean approximation for how big this tax is, and it is worth memorizing because you can do it in your head. The geometric return is roughly the arithmetic return minus one half of the variance. In symbols people write it as:
geometric return is approximately equal to the arithmetic mean minus sigma squared over two.
Sigma is the standard deviation of returns, your volatility. Square it to get variance, halve that, and subtract. That halved variance is the drag. The thing to notice is that volatility enters squared, which is why it punishes you so disproportionately. Doubling the volatility of an asset quadruples the drag it imposes on your compounding. This is not a linear penalty you can shrug off. It accelerates.
A rough worked example. Say an asset averages 10 percent a year with a volatility of 20 percent. Volatility of 20 percent is 0.20, squared is 0.04, halved is 0.02. So the drag is about 2 percent, and your real compounding rate is closer to 8 percent than the 10 percent the average advertised. Now take something that averages the same 10 percent but swings with 60 percent volatility, which is not unusual for a single volatile crypto asset. That is 0.60 squared, 0.36, halved, 0.18. The drag is roughly 18 percent. Your compounding rate is negative. Same average return, and one of them quietly destroys money over time while the other builds it.
How to run this on your own holdings
You do not need anything fancy. Here is the workflow I use when I want to know what an asset is actually doing to a portfolio rather than what its headline says.
- Pull a few years of periodic returns, monthly is fine, weekly if you have it. More frequent sampling gives you a cleaner volatility estimate.
- Take the arithmetic average of those returns and annualize it. That is your advertised number, the one that lies to you a little.
- Take the standard deviation of those same returns and annualize it too. For monthly data you multiply by the square root of 12. This is your sigma.
- Square the annualized sigma, divide by two, and subtract it from the arithmetic annual return. What is left is roughly your geometric rate, the honest one.
- Compare the honest number across your holdings. Some assets that looked like your best performers on average will drop below duller ones once the drag is taken out.
If you want to skip the approximation entirely and get the exact figure, just compute the geometric mean directly. Multiply all the period growth factors together, so 1 plus each return, take the nth root where n is the number of periods, subtract one. That is your true compounded return with no approximation. The sigma-squared-over-two version is useful because it tells you why the exact number came out where it did, and it lets you reason about changes before you make them.
Why this should change how you size things
The practical payoff is in allocation and in any decision about leverage. Because the drag scales with the square of volatility, position sizing on a volatile asset depends on more than how much you are willing to lose in a bad month. It is about how much compounding you are surrendering every single year just to hold the thing. A smaller allocation to a high-volatility asset can produce more long-term wealth than a larger one, because past a certain point you are adding more drag than expected return. There is an optimal size, and oversizing sails right past it into territory where more exposure means less money.
Leverage is where this gets genuinely dangerous, and it is the failure mode I have watched wreck the most accounts. Leverage multiplies your returns, but it multiplies your volatility by the same factor, and since drag goes with volatility squared, the drag grows with the square of your leverage. Two times leverage does not double your drag, it roughly quadruples it. So you take an asset with a healthy positive geometric return, lever it up chasing a bigger number, and the ballooning drag can flip your real compounding rate negative even while the average leveraged return still looks fat and green on a chart. This is exactly the mechanism behind why leveraged daily-rebalanced products decay so badly in choppy, sideways markets. It is not a bug in the product. It is volatility drag doing precisely what the math says it will.
The rule of thumb I keep in my head is simple enough. Whenever someone quotes an average return, mentally subtract half the variance before I believe it, and if the asset is volatile enough that I cannot ballpark that in my head, that itself is the warning that the drag is large. The calmer path with the lower average often wins the decade, and the number that tells you so is the geometric one, not the one on the brochure.