Geometric vs. Arithmetic Mean in Portfolio Growth
The brutal mathematics of volatility drag, the illusion of simple averages, and how to calculate the true compounding power of your algorithmic edge.
The financial industry is built on a mathematical lie. When a hedge fund, a proprietary trading firm, or a retail algorithm vendor markets their performance, they almost universally quote their Arithmetic Mean—the simple average of their returns. They do this because simple averages look impressive on a pitch deck.
In the reality of quantitative execution, the Arithmetic Mean is entirely useless. It assumes that capital is not reinvested and that volatility has no penalty. But trading is a game of sequential, compounding probabilities. If you lose capital today, you have less capital to compound tomorrow.
To survive in algorithmic trading, you must strip away the marketing metrics and judge your system strictly by its Geometric Mean (the Compound Annual Growth Rate). If you do not understand the mathematical divergence between these two numbers, you will over-leverage your account, miscalculate your edge, and inevitably blow up your portfolio.
The Arithmetic Lie: Why Averages Destroy Capital
The Arithmetic Mean is calculated by adding up all your percentage returns and dividing by the number of periods. It is the math you learned in primary school. Let's look at how this destroys a $100,000 trading account over a two-year period with extreme volatility.
- Year 1: You have a massive bull run. Your algorithm generates a +100% return. Your $100,000 account is now worth $200,000.
- Year 2: The market crashes. Your algorithm enters a heavy drawdown and loses -50%. Your $200,000 account is cut in half, leaving you with exactly $100,000.
The Illusion
Arithmetic Mean = (+100% - 50%) / 2 Years = +25% Average Return per year.
If you put this on a resume, you claim you average 25% a year. But look at your bank account. You started with $100,000. You ended with $100,000. You made absolutely zero money. This discrepancy is the exact reason retail traders who risk too much per trade inevitably go bankrupt, even if their "average" trade is profitable.
The Variance Penalty (Volatility Drag)
The reason the Arithmetic Mean lies to you is a mathematical phenomenon called Volatility Drag (or the Variance Penalty). In geometric compounding, losses are fundamentally heavier than gains.
If you lose capital, the remaining capital must work exponentially harder just to get back to the starting line. The deeper the drawdown, the more aggressive the mathematical penalty becomes.
| Portfolio Drawdown | Required Gain to Break Even |
|---|---|
| -10% | +11.1% |
| -20% | +25.0% |
| -50% | +100.0% |
| -90% | +900.0% |
This table illustrates the central thesis of quantitative risk management: Capital preservation is infinitely more important than capturing the peak of a trend. If your algorithm sizing allows for a 50% drawdown, you must literally double your remaining money just to recover. High variance mathematically destroys geometric compounding.
The Geometric Truth
The Geometric Mean is the true measure of your algorithm's efficiency. It explicitly factors in the variance penalty. It calculates the exact continuous rate of return required to get from your starting capital to your ending capital.
This approximation formula is the most important piece of math a quantitative developer will ever look at. It proves definitively that as volatility (variance) increases, your actual geometric compounding rate drops, even if the "average" winning trade remains identical.
If you have two trading algorithms that both boast a 20% Arithmetic Mean, but Algorithm A has massive variance and Algorithm B has low variance, Algorithm B will drastically outperform Algorithm A over a 10-year period. Algorithm A is constantly fighting the friction of deep drawdowns, while Algorithm B geometrically compounds with frictionless efficiency.
The Kelly Criterion Solution
This is precisely why the Kelly Criterion is the undisputed king of position sizing. The Kelly formula is not designed to maximize your simple average return; the Kelly formula is the mathematical derivative that strictly maximizes the Geometric Mean.
When you execute at optimal Kelly, the formula has perfectly balanced your statistical edge against the devastating effects of Volatility Drag. It pushes your leverage exactly high enough to maximize compounding, and strictly limits your leverage the moment the variance penalty begins to overwhelm the geometric return. By stepping down to Half Kelly or Quarter Kelly, you sacrifice a small amount of peak geometric growth in exchange for massively crushing the variance penalty, resulting in a smoother, psychologically survivable equity curve.
Protect Your Geometric Growth
Stop optimizing your algorithms for high average returns, and start optimizing them for low-variance geometric compounding. Input your trading edge into the EV Kelly engine to calculate the exact fractional sizing required to limit your volatility drag and prevent mathematical ruin.
Open the Calculator Engine