Open your portfolio and count the coins. Five, maybe eight. BTC, ETH, a couple of large caps, one or two bets you're quietly proud of. It looks spread out. It feels safe.

It isn't. You're holding one position wearing eight costumes.

Ethereum, BNB, and Litecoin have historically shown a correlation with Bitcoin between 0.70 and 0.90. A correlation of 1.0 means two assets move in lockstep.

So when your "diversified" bag sits at 0.85, you don't own eight different things. You own Bitcoin, leveraged by however many altcoins you stacked on top.

Correlation is the risk almost nobody prices in. You size each coin like it's an independent bet. It isn't. And the moment you actually need diversification — the crash — is the exact moment it disappears.

In 2022, Bitcoin fell 77%. That part gets talked about. What gets skipped is what the altcoins did around it. They didn't cushion the fall. They amplified it, most of them down 80% or 90%, because in a real drawdown the correlation between crypto assets doesn't just stay high. It climbs toward one. Investors stop selling selectively. They sell everything they can, all at once.

This isn't a crypto quirk. In 2008, asset classes that were supposed to zig while others zagged all went down together. The mechanism is the same every time. When people are scared, the reason they own something stops mattering. Liquidity matters. Getting out matters. Your careful thesis about a Layer 2 token means nothing when the whole market is being liquidated in the same three hours.

Where most traders get it backwards

They think diversification is about the number of things you hold. Ten coins feels safer than three. But diversification was never about count. It's about how those things behave when one of them breaks. Ten assets that all crash together give you exactly as much protection as one — you've just paid more fees to feel comfortable.

The comfort is the trap. A basket of correlated altcoins looks calm in a bull market because everything's green and nobody's checking the correlation matrix. Then the regime flips, the coins that rose together fall together, and you find out your risk was concentrated the whole time. You were long one trade at 8x. You just didn't label it that way.

I've watched people do this to themselves, and I did a version of it myself years ago. The account looks robust right up until the day it doesn't. What kills you isn't the coin you were worried about. It's the correlation you never measured.

What actually helps

Start by sizing for what happens in the crash, not the calm. If five of your positions move together, treat them as one position and size the group, not each name. That "eight coins" portfolio might really be two or three independent bets. Size it like two or three, and your risk math suddenly matches reality.

Real diversification in crypto is rare and it doesn't come from holding more of the same beta. It comes from genuinely different behavior — different theses, different return drivers, ideally something that doesn't need the whole market to go up for it to work. That last part is the interesting one. A strategy's edge can come from how it trades, not just what it holds. A system that goes to cash or flips short in a downtrend has a different return profile than a bag of coins that can only pray for a bounce. That's a real diversifier. Another spot coin usually isn't.

The one thing to take from this: before you add a position, don't ask whether you believe in it. Ask how it behaves when Bitcoin drops 40% in a week. If the honest answer is "same as everything else I own," you didn't diversify. You just concentrated with extra steps.

Pull up a correlation chart on your current holdings this week. Most people have never looked at one for their own portfolio. It takes ten minutes and it tends to be uncomfortable, because the diversification you thought you had usually isn't there.

If removing the emotional guesswork from that decision sounds appealing, systematic strategies with a defined behavior in downtrends are worth a look. We publish the full backtest data — either way, measure the correlation before the market measures it for you.

See the data →