Mean Reversion — Pro — Growth Capital Group

The Concept

From "That Looks Stretched" to a Testable Rule

Many prices and spreads oscillate around some underlying average rather than trending forever. Mean reversion strategies formalize the intuition that "this has gone too far" into a precise, repeatable rule: measure how far the current price has strayed from its historical average, in statistical terms, and act when that distance crosses a threshold.

The standard tool for measuring "how far" is the z-score: (current price − historical mean) ÷ historical standard deviation. A z-score of +2 means the price is two standard deviations above its average — statistically unusual, and historically likely (though never certain) to pull back toward the mean. A z-score of -2 signals the opposite: unusually cheap relative to its own history.

Pairs trading and statistical arbitrage extend this same idea to the spread between two related assets rather than a single price — betting that when two historically-linked securities drift apart, the gap itself will revert, regardless of which direction the market as a whole moves.

⚖️ Illustrative Example: Z-Score Signal Zones

A stock's price has a 60-day historical mean of $100 and a standard deviation of $5. Here's how different current prices translate into z-scores and, under a simple rule, mean-reversion signals.

Current Price	Distance From Mean	Z-Score	Illustrative Signal
$115	+$15	+3.0	Strong Sell / Fade
$110	+$10	+2.0	Sell / Fade Threshold
$101	+$1	+0.2	No Signal — Near Mean
$90	-$10	-2.0	Buy / Fade Threshold
$82	-$18	-3.6	Strong Buy / Fade

A common rule of thumb enters a mean-reversion trade once the z-score crosses ±2 (roughly the top/bottom 5% of a normal distribution) and exits as the price reverts back toward zero — illustrative only, and real markets don't move in a neat bell curve.

Watch For This

5 Things to Know About Mean Reversion

Not every asset mean-reverts — the strategy assumes a stable underlying average exists; a genuinely trending asset can keep making the z-score more extreme instead of reverting.
"Historical mean" is a moving target — using too short a lookback window makes the mean noisy; too long a window makes it slow to reflect real structural change.
This is the same math as Bollinger Bands — a Bollinger Band touch is essentially a z-score extreme, expressed visually rather than numerically.
"Catching a falling knife" is the classic failure mode — a stock cheap for a genuine, lasting reason (bad news, a broken business) can keep getting cheaper well past a -2 z-score.
Pairs trading isolates the spread, not market direction — a well-constructed pair can profit in both up and down markets, since it bets on the relationship between two assets, not the market's overall direction.

Put It Into Practice

4 Things to Check Before Trusting a Mean-Reversion Signal

📏 Choose Your Lookback Window Deliberately

Test multiple windows (e.g. 20, 60, 120 days) — a signal that only appears with one specific window is a red flag.

🧭 Ask Why It's Stretched

A statistically extreme z-score driven by a fundamental change, not noise, may never revert — check for news, not just numbers.

🧪 Backtest With Realistic Costs

Mean reversion trades are often frequent and short-lived — trading costs and slippage can erase an edge that looks great on paper.

🛑 Define an Exit If It Doesn't Revert

Set a stop for when the z-score keeps extending rather than reverting, so a broken assumption has a bounded cost.

🧮 Related lessons: Bollinger Bands (Intermediate) covers the same z-score logic visually, and Backtesting (next in this track) shows how to properly test a rule like this before risking real money.

Worth knowing: this lesson explains the statistical reasoning behind mean-reversion strategies using illustrative numbers and historical framing — it isn't personalized financial advice, and no strategy described here is a recommendation to trade. Past patterns, including reversion to a mean, are never guaranteed to repeat. Speak to a licensed advisor about what's appropriate for your situation.

Activity

Try It Yourself: Z-Score Calculator

Enter a current price and its historical mean and standard deviation — see the z-score and where it falls relative to common ±2 signal thresholds.

Current Price ($)

Historical Mean ($)

Historical Std. Deviation ($)

Signal Threshold (±z)

Distance From Mean

—

Z-Score

—

vs. Threshold

—

Model: z-score = (current price − historical mean) ÷ historical standard deviation. This is a statistical measure of stretch, not a prediction — historical patterns are never guaranteed to repeat.

End of Lesson