Five Position-Sizing Mistakes That Kill Crypto Accounts
The Fluxy Team
-Five Position-Sizing Mistakes That Kill Crypto Accounts
There's a dirty secret in systematic trading: the signal matters less than everyone thinks, and the sizing matters more. A mediocre signal sized well survives to compound. A good signal sized badly dies on an ordinary Tuesday. In crypto — where 5% daily moves are unremarkable and 15% happens every year — sizing is the survival skill.
These are the five mistakes we see most often, roughly in order of bodies.
1. Fixed size in a regime market
Trading "1 BTC" or "25% of the account" regardless of conditions means your risk is whatever the market decides. BTC's realized volatility has ranged roughly 30% to 150% annualized over recent years — a fixed position is a 5× different trade depending on the month, and the max-size months are the violent ones. The fix is volatility targeting: pick the risk, let the size float. It's less an optimization than a survival trait.
2. Sizing up after losses
Martingale has a seductive logic — "it has to bounce eventually" — and a century of ruined practitioners. Doubling into a drawdown converts a survivable losing streak into a terminal one, and crypto serves losing streaks generously. The tell is any sizing rule where position size grows as account equity shrinks. Disciplined sizing does the opposite: size follows equity down, which is what makes drawdowns recoverable. (Sizing as a fixed fraction of current equity does this automatically — one reason it beats fixed-dollar sizing.)
3. Correlation blindness
Five positions at 10% each feels diversified. In crypto it usually isn't: in a risk-off move, most alts are one trade wearing twenty tickers — correlations that sit at 0.6 in calm weeks go to 0.9+ in a crash, exactly when it matters. Portfolio-level sizing has to treat correlated positions as one exposure. Our engine surfaces correlated positions at the book level for precisely this reason: per-position discipline with portfolio blindness is discipline theater.
4. Ignoring the leverage cliff
Leverage doesn't scale risk linearly; it moves your liquidation price into the range of ordinary noise. At 2×, BTC has to halve to liquidate you. At 10×, a 9% move does it — and 9% moves are a monthly event. Worse, venue leverage brackets shrink your allowed leverage as size grows, so the same strategy behaves differently at different account sizes. Sizing that respects venue brackets and liquidation distance — with hard caps the strategy cannot override — is the difference between a drawdown and a zero.
5. Sizing by conviction
"I'm really sure about this one" is how concentrated positions happen, and concentrated positions are how accounts end. Conviction is not information about variance. The empirical reality is humbling: even good systematic signals are barely better than coin flips per-trade; their edge lives in repetition. A sizing rule that lets today's confidence override the risk budget converts a statistical edge into a sequence of coin flips for the whole stack.
What disciplined sizing looks like
Every fix above is the same fix wearing different clothes: decide risk in advance, mechanically, at the portfolio level — vol-targeted size, fraction-of-equity scaling, correlation-aware book limits, leverage caps, no manual overrides. Then let the machine keep the promise on the day you'd break it yourself, because that day is exactly when the rule earns its keep. This is why sizing in our platform lives in strategy config executed by the engine, not in the heat of the moment.
None of this makes a bad strategy good. It makes a decent strategy survivable — and survival is the only property that lets compounding do its work.
Vol targets, leverage caps, and portfolio-level limits — declared once, enforced mechanically. Size positions like a desk →
Related reading