Correlation Is Not Causation: The Most Dangerous Shortcut in Crypto

What You Will Learn

The difference between correlation and causation — and why confusing them is fatal for traders
Three specific ways correlation tricks you, with crypto-native examples
A practical checklist for evaluating any data-driven signal before you trade on it

The Core Idea

“When hashrate goes up, BTC price follows.” “Social sentiment predicts altcoin moves.” “Rising TVL means bullish fundamentals.”

You’ve heard claims like these. They sound logical. They might even come with charts. But they share a common flaw: they confuse correlation (two things moving together) with causation (one thing making the other happen).

This distinction isn’t academic. If you build a trading strategy on a relationship that’s correlative but not causal, the relationship can vanish without warning — and you won’t understand why your edge disappeared.

What Correlation Actually Means

Correlation measures how two variables move in relation to each other, on a scale from -1 to +1.

+1: Perfect positive correlation. They move in lockstep.
0: No correlation. Their movements are unrelated.
-1: Perfect negative correlation. One goes up, the other goes down.

A high correlation between two assets or metrics tells you they’ve moved together in the past. It says nothing about why they moved together — and nothing about whether they’ll continue to do so.

This is where traders get burned. Finding a high correlation feels like finding an edge. But correlation without a causal mechanism is a pattern without a foundation. And patterns without foundations break.

Three Ways Correlation Tricks You

1. Confounding Variables

BTC price and altcoin prices are highly correlated. Does BTC cause altcoins to move? Not directly. Both are responding to a common underlying factor: broad market risk sentiment. When macro conditions favor risk assets, both BTC and alts rise. When fear hits, both drop.

The confounding variable — global risk appetite — creates the correlation. If you built a strategy that trades alts based on BTC movements, you’d be right much of the time. Until you weren’t — because the real driver was macro, and the correlation broke the moment BTC and alts decoupled (as they sometimes do during alt-specific events).

2. Spurious Correlation

The number of Nicolas Cage films released in a year correlates with swimming pool drowning deaths. Obviously, one doesn’t cause the other. With enough data series and enough time, you will find correlations everywhere. Most of them are noise.

Crypto is particularly fertile ground for spurious correlations because the market is young, data series are short, and there are thousands of metrics to test against each other. Run enough comparisons and something will look statistically significant. It won’t be.

3. Reverse Causation

“Exchange inflows increase before price drops, so exchange inflows cause selling pressure.”

Maybe. Or maybe falling prices cause traders to move assets to exchanges — planning to sell because the price is already dropping. The arrow of causation can point in either direction, and the data alone can’t tell you which.

This is one of the most common traps in on-chain analytics. An on-chain metric moves before price — but “before” doesn’t automatically mean “because.” The metric and the price may both be responding to the same information, with the metric reacting slightly faster.

Crypto-Specific Examples

Hashrate and price. Bitcoin hashrate and price are correlated over long timeframes. But hashrate doesn’t drive price. Both are driven by miner profitability expectations, which themselves depend on hardware costs, energy prices, and market sentiment. In the short term, the correlation frequently breaks down.

Social sentiment and price. Twitter/X sentiment often moves with altcoin prices. But does positive sentiment cause buying? Or does a price rally cause positive sentiment? In practice, it’s mostly the latter — sentiment is a lagging indicator dressed up as a leading one.

TVL and token price. Total Value Locked in a DeFi protocol correlates with its token price. But this is often circular: when the token price rises, the dollar value of assets locked in the protocol rises mechanically — even if no new capital entered. Analysts then point to “growing TVL” as bullish fundamentals, creating a self-referencing loop that collapses when the price reverses.

A Checklist Before You Trust Any Signal

Before trading on any data-driven signal, run it through these four questions:

1. Can you explain the mechanism? Why would A cause B? If you can’t articulate a plausible causal chain — not just “they move together” — the signal is suspect.

2. Is the time sequence correct? Does A consistently occur before B? And even if it does, is there a third factor that precedes both?

3. Have you ruled out confounders? What else could explain the relationship? Common macro factors, shared liquidity pools, or overlapping investor bases can all create correlations that look like signals.

4. Does it hold out of sample? The signal worked on historical data. Does it work on data it hasn’t seen? If you only tested it on one time period, you’re looking at a backtest, not an edge.

If a signal can’t pass all four checks, treat it as noise until proven otherwise.

Common Failure Modes

Getting excited by a high correlation — finding two data series that move together and immediately treating it as a tradable signal, without asking why they’re correlated.
Cherry-picking time windows — the correlation worked from March to June, so you ignore the months it didn’t. This is curve-fitting, not analysis.
Assuming causation because “everyone says so” — popular narratives become accepted truths through repetition, not verification. “Exchange inflows = dump” is treated as law, but the evidence is far more nuanced.
Freezing when the correlation breaks — you’re in a position based on a correlated relationship, the correlation decouples, and you don’t know what to do because you never understood why it worked.