Everybody understands that leverage multiplies exposure. Far fewer notice that it multiplies the fee bill by exactly the same factor, on every single trade, whether the trade worked or not.
A fee schedule is a percentage. That percentage is applied to the notional value of a position, which is the full size of the exposure, not to the margin you posted to open it. This single sentence contains everything, and it is almost never stated plainly on the page where the schedule is published.
Open a position with ten thousand of margin at one times leverage and the notional is ten thousand. Open it with the same ten thousand at ten times and the notional is one hundred thousand. The percentage charged has not moved by a hair. The amount charged has been multiplied by ten.
Leverage does not get you a worse rate. It gets you a bigger bill at the same rate, which feels different and costs the same.
Consider two traders with identical accounts, identical strategies, identical numbers of trades in a month. One runs at two times leverage, the other at ten. Their entries are the same, their exits are the same, their win rates are the same.
The second trader pays five times the fees of the first. Not five times more often, five times more per trade, on every trade, in every month, forever. If the strategy has an edge, that edge has to cover a cost five times larger before either trader sees anything. If the edge is thin, the second trader can be running an identical system and still end the year behind.
The standard way traders compare venues is to line up the published taker rates and pick the smaller one. That comparison is sound only if the two accounts would be run at the same leverage, and it quietly ignores the far larger variable sitting in their own hands.
A trader who halves their leverage halves their fee bill immediately, on any venue, without changing anything else and without asking anyone's permission. No difference between two published schedules comes close to that. The lever with the largest effect on the number is not the one being compared.
This is not an argument for or against leverage. It is an argument for knowing which line of the cost actually moves when you change something.
There is a second effect, and it is the one that surprises people who have already understood the first. Higher leverage does not only multiply the fee per trade, it usually increases the number of trades.
A position at ten times leverage has a liquidation price close enough that ordinary noise reaches it. That forces tighter stops, tighter stops get hit more often, and every stop that gets hit is a round trip that gets charged twice, at the taker rate, because a stop enters the book as a market order. The trader has not decided to trade more; the position size has decided for them.
So the fee bill at high leverage is multiplied twice over: once by the notional, and once by the frequency that the notional forces. The second multiplication is invisible in any fee schedule, and it is the reason a backtest run without fees looks nothing like the same system run with them.
Three numbers, and they take a minute to write down.
Not your account size. The sum of the notional value of every position you opened last month. Most traders have never calculated this and are surprised by it, often by a factor of several, because it counts the size and not the money at risk.
Twice, because a round trip is charged on the way in and on the way out. Use the taker rate for both legs rather than the maker rate: a market order always is one, a limit order can become one, and a stop always is one. Estimating with the maker rate produces a number that is comfortably wrong in the direction you would prefer.
The difference between the two results is what a single decision is worth, per month, before anything about your strategy changes. The calculator on this site does this with your own inputs, and it deliberately uses the taker rate throughout for the reason above.
The mechanics described above are universal: notional times rate, everywhere, on every asset class. What is specific to crypto is how much further the numbers go, and three features of the market push them there.
Regulated equity markets cap retail margin in the low single digits. Crypto venues routinely offer twenty, fifty, a hundred and more. A trader who would be structurally limited to twice their capital elsewhere can carry a hundred times here, and every fee they pay is multiplied accordingly. The ceiling is not the same, so the bill is not the same.
A daily range that would be an event on a large equity index is an ordinary Tuesday on most crypto pairs. Combine that volatility with high leverage and the distance between an entry and a liquidation price shrinks to something ordinary noise can cross. That is what produces the second multiplication described above, the one on frequency, and it is much stronger here than in a market that moves a fraction of a percent a day.
Outside crypto, holding a leveraged position costs interest on borrowed money, calculated and charged in ways that are usually annual and usually visible. In crypto the equivalent is funding on perpetuals, charged every few hours on the notional, and therefore multiplied by leverage exactly as fees are. A leveraged position in crypto pays two costs that both scale with the same number, and most comparisons only count one of them.
None of this argues that crypto leverage is bad or that any particular level is right. It argues that the arithmetic of cost is steeper here, that it steepens fastest in the direction most people move by default, and that the number worth calculating is the total of both costs across a realistic holding period rather than the headline rate on a schedule.
A liquidation is not a neutral event on the fee side. The engine that closes the position has to remove liquidity from the book to do it, so the close is charged at the taker rate on the full notional, exactly like any other market order. That happens on the trade that has already gone furthest against you, and on the largest notional you were carrying.
Some venues add a separate liquidation charge on top, described in their documentation rather than in the fee schedule, and it is one of the few numbers worth reading before opening an account rather than after. The point here is narrower and applies everywhere: the cost of being liquidated includes a taker fee on the entire position, and high leverage both increases that notional and shortens the distance to the price that triggers it.
The two effects of leverage on cost are not independent. It enlarges every fee, and it makes the most expensive fill of all more likely to happen.
Two accounts of twenty thousand. Both open ten positions a month. The first uses three times leverage, so each position carries sixty thousand of notional and the monthly notional is six hundred thousand. The second uses fifteen times, so each position carries three hundred thousand and the monthly notional is three million.
Five times the notional, five times the fees, at an identical published rate and for an identical number of decisions. Over twelve months the gap is not a detail on a statement; on a thin edge it is the difference between a system that clears its costs and one that does not. And the second account is also the one whose stops sit closest to the noise, which tends to push its ten trades a month upward without anybody deciding to trade more.
It does not say that low leverage is better. Leverage is a risk decision that belongs to whoever is taking the risk, and there is no responsible way for a venue to have an opinion on the right number for somebody else's account. Plenty of sound strategies require leverage to be worth running at all.
What it says is narrower and, we think, more useful: whatever leverage you choose, the fee consequence of that choice is mechanical, immediate, and larger than any difference between two published fee schedules. It deserves to be a number you calculated rather than a surprise at the end of the month.
No. The rate on the schedule is the same at one times and at fifty times. What changes is the notional value the rate is applied to. At ten times leverage, ten thousand of margin controls one hundred thousand of notional, and the fee is charged on the hundred thousand.
On the position size, which is the notional. This is why two traders posting the same margin can pay very different fees for the same number of trades. It is also why reducing leverage reduces the fee bill immediately and proportionally.
Yes, on perpetual contracts. Funding is also charged on notional, so it scales with leverage exactly as trading fees do. The difference is that funding is paid to the other side of the book rather than to the venue, and it is only paid if the position is held across a payment timestamp.
For the same total notional, the fee is the same: fees are charged per fill on the notional filled, not per position. What changes the bill is the total notional traded and the number of round trips, not how that notional is split up at a given moment.
That is not a question a venue can answer for you, and any venue that tries is telling you something about itself rather than about your account. Leverage is a risk decision belonging to whoever carries the risk. What this article claims is narrower: whatever level you choose, the fee and funding consequence of that choice is mechanical and calculable in advance, and it deserves to be a number rather than a surprise.
Not proportionally, no. Fees scale linearly with leverage; the probability of being liquidated does not, because it depends on the distance between your entry and your liquidation price relative to the volatility of the pair. Halving leverage halves the fee bill exactly and improves the liquidation distance by more than half. The two effects are different in kind, which is why they are worth looking at separately.
Neither is cheaper in fees: the rate and the notional are the same in both. What changes is which balance backs the position and therefore when a liquidation happens, so the choice affects the probability of paying the liquidation fee rather than the size of any ordinary fee.
Because they usually assume the maker rate, or no fees at all, and because they do not model the fact that higher leverage forces tighter stops and therefore more round trips. Both errors point the same way, which makes the result comfortable and wrong.