Crypto Trading Fees Explained: What a Trade Actually Costs

Every exchange publishes a fee page, and almost nobody reads past the first number on it. That number — usually 0.10% — is real, but it is one of four separate costs attached to a trade, and for most accounts it is not the largest.

This guide takes each of them apart: what you pay to open and close a position, what the spread quietly takes, what a futures position is charged every eight hours, and what it costs to move coins off the platform. None of it needs maths beyond multiplication, and knowing it changes which exchange is genuinely cheap for the way you trade.

The four costs attached to one trade

The trading fee is the advertised one: a percentage of the order value, charged when you open and charged again when you close. A round trip therefore costs double the headline rate, which is the single most common miscalculation in crypto.

The spread is not a fee and nobody collects it as one, but it is money you lose all the same. It is the gap between the best buy order and the best sell order on the book. On BTC/USDT it is a rounding error; on a small-cap pair that trades a few thousand dollars a day it can cost several times the trading fee, which is why a low-fee venue with a thin book is not automatically cheap.

Funding exists only on perpetual futures. Every eight hours one side of the market pays the other to keep the contract price anchored to spot. The exchange is not the counterparty — you are paying other traders, or being paid by them.

The withdrawal fee is a flat amount in the coin you are moving, charged when you take it off the exchange. It mostly reflects what the blockchain costs, which means the network you select matters far more than the logo at the top of the page.

Maker and taker: the same trade at two prices

You are a taker when your order removes liquidity that was already sitting on the book — a market order always does this, and so does a limit order priced aggressively enough to fill immediately. You are a maker when your order rests on the book and someone else trades against it.

Exchanges pay for depth, so they charge the two sides differently. On spot the gap is small: most of the six sit at 0.10% either way, with OKX shaving the maker side to 0.08% and MEXC listing 0% maker and 0.05% taker. On futures the gap is wide — 0.02% to make against 0.05% to 0.06% to take. The same position entered with a market order costs roughly three times what it costs entered with a resting limit order.

The practical version: if you are not in a hurry, place a limit order at or just inside the top of the book and let it fill. If it crosses the spread and fills instantly, you paid the taker rate anyway, whatever the order type was called.

What 0.10% actually costs over a year

Work it through once and the number stops being abstract. A $1,000 order at 0.10% costs $1 to open and $1 to close: $2 for the round trip. Ten round trips a month is $20, or $240 a year, on an account that never grew.

Now set that against a welcome bonus. A headline package is a one-off, unlocked in tranches by deposits and volume, and most of it is never claimed. The fee is charged on every order, forever, and it compounds against you in the most literal sense — it comes out of the capital that would otherwise be compounding. Over a year of ordinary activity, the fee line is usually the bigger number.

The heatmap above prices the same behaviour across venues: it multiplies each exchange standard spot taker rate by a month of volume, so you can see the point where the difference stops being noise.

Futures: the taker rate, then a fee that never stops

Futures rates look cheaper than spot, and per dollar of your own money they are not. The fee is charged on the position, not on your deposit. Put up $1,000, open at ten times leverage, and you are holding $10,000 of exposure: at 0.05% that entry costs $5 before the market has moved a cent, and closing costs another $5. Leverage multiplies the fee exactly as reliably as it multiplies the risk.

Then there is funding. Every eight hours the contract settles a payment between longs and shorts, sized by how far the perpetual has drifted from spot. When everyone is long — which is most of the time in a rally — longs pay. A position held through a crowded week can pay more in funding than it ever paid in trading fees, and nothing on the order ticket warns you about it.

Funding is published in advance on every exchange, usually as a rate per eight-hour window next to the contract name. Reading it before you open a multi-day position takes ten seconds, and it is the difference between an informed carry and a slow leak.

VIP tiers, and why yours will not move

Every large exchange runs a volume ladder: trade more inside a rolling thirty-day window and your rate drops, sometimes with a second condition attached, such as holding a quantity of the exchange own token.

The ladder is real, and for almost everyone it is also decoration. The volume needed to reach the first meaningful step is far above what a normal account turns over in a month, and the thirty-day window rolls continuously — a good month does not carry. Treating a VIP tier as your plan for lowering costs is planning around something you will not reach.

The curve above shows the shape rather than any single exchange table: steep where the largest accounts sit, flat across the range where everyone else trades. Which is the point — if your rate is going to change, volume is not what will change it.

The two levers that do move your rate

The first lever is the referral discount. Accounts opened through a referral link on Bybit, OKX, Bitget and Gate.io carry a 20% reduction on standard trading fees — 0.10% becomes 0.08% — and it applies from the first order rather than after a qualifying period. The catch is the one nobody mentions until it is too late: it is set at registration. Adding a code to an existing account works only in narrow windows (MEXC within 30 days, Bybit within 14 days for affiliate codes), OKX says it cannot be done, and for Bitget and Gate.io we found no documented way. Binance runs a fee-rebate programme instead of a standing percentage discount.

The second lever is the venue itself. MEXC's fee page lists a 0% maker fee and a 0.05% taker fee on spot — its standard published rate rather than a promotional window, though MEXC says rates vary with platform events and region and, like any fee schedule, it can be revised. For someone who mostly places resting limit orders that removes the trading fee, and taker orders cost half the 0.10% standard rate.

People who think about this seriously tend to use both levers rather than choose between them: spot orders routed to the lowest-fee venue, derivatives on whichever futures book is cheapest for the pair they trade, and every account opened through a referral link the first time — because that door only opens once.

The fee you only notice on the way out

Withdrawal fees are flat rather than proportional, which makes them brutal on small amounts and irrelevant on large ones. 1.5 USDT on a 2,000 USDT transfer is a rounding error. The same 1.5 on a 200 USDT transfer is 0.75 percent — seven and a half times the 0.10% taker fee you paid to open the position.

The bigger variable is not the exchange, it is the network — and the ranking is the opposite of the folklore. On 2 September 2026, sending USDT over Tron cost 1.5 USDT at Binance, OKX and Bitget and 0.5 at MEXC, while the same transfer over Ethereum cost 0.3, 0.38, 0.8 and 0.094. The dropdown that selects the network sits one line above the address field, and it is the most expensive dropdown in crypto — both for what it costs and because choosing a network the receiving side does not support is how deposits go missing.

Two habits remove most of this cost: consolidate withdrawals instead of moving money in dribs, and confirm which networks both ends support before you start rather than after.