Crypto is the only market open 24/7, the youngest, the most volatile and the worst narrated: half the world describes it as the future of finance, the other half as a scam. I care about a third description: a market with specific microstructure, specific costs and specific opportunities — to be measured, not cheered. It's the market of my most rigorous quant research and of my latency arbitrage bot.
Spot, perpetuals, CFDs: three different exposures
- Spot (exchange): you buy and own the asset. Low costs (~0.1% commissions), custody on you, little or no leverage. For the long term, the efficient route.
- Perpetual futures: the heart of professional crypto trading. Contracts without expiry anchored to spot via the funding rate: every 8 hours (typically) longs and shorts pay each other depending on which way the market leans. Nominal leverage up to 100x (which nobody sane uses), automatic liquidations, deep order books. Funding is a real flow: on positions held for weeks it can be worth more than the price move — which is why funding/basis carry strategies are one of the few areas where my research found hypotheses still alive.
- CFDs (broker): the regulated European retail perimeter: no custody, but wider spreads, daily swaps, limited leverage. Inefficient for holding; for trading, compare total costs — and the fact that the CFD price follows the reference exchange's is exactly the structural asymmetry my bot exploits.
Volatility: why it attracts and why it kills
Bitcoin does in a day what an index does in a month. It looks like an advantage ("more movement = more opportunity") but the accounting must be done in risk units: if moves are 5x wider, your size must shrink 5x to keep the same per-trade risk — and then percentage costs weigh just the same. Volatility doesn't gift edge: it scales everything, including your losses and your tails. And perpetuals' liquidation cascades produce moves no "mental" stop survives: here the stop in the market (better: the right size) isn't advice, it's the only defense.
What the real data says (not the YouTube channels)
I tested the planet's most popular hypothesis — price action patterns on 5-minute Bitcoin — on 210,240 audited bars, with leak-free features and an out-of-sample verdict. Result: gross edge ≈ zero against 6.5 bps of costs, on every pattern and every horizon. Not even filtering by volatility, volume, session or order flow. The full research is public, replicable with the Python + Binance API pipeline I use — the data is free, the excuses aren't.
Where structure does pay: real order-flow information lives at tick and order-book level (not aggregated bars), mechanism asymmetries (feed latency, funding, basis) are measurable, and the absence of daily closes eliminates gaps — but introduces the opposite problem: no moment when the market lets you sleep. For an automated system, an advantage; for a manual trader, attrition.
Security: the risk that isn't market risk
- Counterparty risk: exchanges fail — recent history proved it in the most expensive way. Operating capital on the exchange, the rest in self-custody or distributed.
- Scams: crypto is the natural habitat of rug pulls, Ponzis dressed as "30% staking", Telegram signals and fake account managers. The eternal rule, amplified: if the promised return exists, why are they selling access to you instead of using it?
- Taxes: crypto gains are taxed nearly everywhere, with rules that keep changing: track every operation and check current law with a professional — this guide is not tax advice.