Why mean-reversion is the closest thing finance has to gravity

There's no Newton's law in finance. The closest thing is mean-reversion: assets that move far from their long-term averages tend to drift back toward those averages. Volatility clusters then dissipates. Sentiment cycles between euphoria and despair. Spreads widen during crises and compress during calm.

This works as a tendency, not a law. Treating it as a law is how most retail traders blow up.

Why mean-reversion exists

The mechanism is partly economic, partly psychological:

Economic side: a company trading at 100× earnings is implicitly being valued as if it will grow 30% a year forever. Forever is a long time. Eventually competition, regulation, or market saturation slows growth. The valuation has to compress. Symmetrically, a company trading at 5× earnings is being valued as if it will shrink forever. Eventually management changes, restructuring happens, the cycle turns. The valuation has to expand.

Psychological side: humans are bad at extrapolating. Recency bias makes traders assume current conditions will persist. When markets are up for three years, everyone is bullish. When markets are down for six months, everyone is bearish. The actual probability of the next move is closer to the long-run average than to the current trend. Smart traders systematically take the other side of consensus extrapolation.

When mean-reversion works

It works best when:

1. The fundamentals haven't changed. A stock down 40% on a single quarter's earnings miss is usually a mean-reversion opportunity (if the long-term story is intact). A stock down 40% on a fundamental disruption is not.
2. The time horizon is long enough. Mean-reversion happens over months to years, not days. Traders who buy oversold stocks expecting to sell them three days later are gambling on noise, not exploiting reversion.
3. You can survive the journey. A position can mean-revert 100% in your direction over 18 months while drawing down another 30% along the way. If you're forced to sell during the drawdown, you missed the reversion.

When mean-reversion fails (loudly)

It fails when there's been a regime change:

• Permanent business model disruption: Kodak's mean-reversion never came. Neither did Blockbuster's. Some companies don't revert because the world that made them valuable no longer exists.
• Permanent macro shift: Japanese real estate didn't revert from 1990 highs. Some asset classes can take a generation to come back, which for most traders is effectively never.
• Bankruptcy or major dilution: A company that issues 60% more shares during a crisis doesn't revert on a per-share basis even if the business recovers.

A practical framework

Before any mean-reversion trade, ask three questions:

1. What's the mean I'm reverting to? Be specific. "The 5-year average P/E" is concrete. "What it should trade at" is too vague to act on.

2. What's the mechanism that gets it there? "Sentiment changes" is weak. "Earnings beat fixes the narrative" or "interest rates normalize" or "management buyback compresses share count" are concrete mechanisms with timelines.

3. What's my drawdown tolerance? If you're sized at 5% of portfolio and willing to ride a 50% additional drawdown, you can survive almost any mean-reversion journey. If you're sized at 25% and worried about a 20% drawdown, you'll get shaken out before the trade works.

A worked example: oil in 2020

In April 2020, oil futures briefly traded negative. Storage was full, demand collapsed, producers were paying buyers to take barrels off their hands. Mean-reversion thesis: this is unprecedented and unsustainable; the long-run price of oil is somewhere between \$40 and \$80; this should revert hard.

The trade: buy oil futures at \$5-15.

The journey: by mid-2021, oil was at \$70. By 2022, \$100+. The reversion was massive — 10-20× on the entry. But the journey took 18 months and required surviving multiple 30%+ drawdowns along the way.

Most retail traders who tried this didn't make it. They either:

• Sized too large and got margin-called during the drawdowns
• Sized correctly but sold too early when oil rallied to \$25 and looked "fine"
• Got the timing right but used futures with rolling costs that ate most of the return

The mean-reversion thesis was correct. Execution killed most of the people who tried to express it.

Try this in the simulator

1. Find a stock currently trading more than 30% below its 200-day moving average.
2. Investigate the fundamentals: has the long-term story actually changed, or is this a temporary dislocation?
3. If the story is intact, build a position over 3-5 weeks (not in one day — averaging in reduces the cost of being early).
4. Set a target: when does this revert to the 200-day, and what's your time horizon?
5. Track it for 6 months. Was your mean-reversion thesis right? Did you size correctly?

Most students will find that their thesis was right about 60% of the time, but their sizing was wrong about 80% of the time. Sizing is harder than thesis.

Next lesson: Position sizing — Kelly, fractional Kelly, and why most students get it wrong.