Maker, taker, and what you actually pay
You compare two venues by their fee schedules, pick the cheaper one, and your realised costs are higher. Nothing was hidden and the schedule was accurate.
The fee is one line in a bill with several. Working out what a round trip costs means adding all of them, and the fee is frequently not the largest.
Why the two rates exist
A venue with no resting orders is useless — there is nothing to trade against. Depth in the book is the product, and it has to be induced.
So venues price the two roles asymmetrically. Makers post orders that rest and add depth; takers send orders that cross and consume it. Maker fees are typically lower than taker fees, and on some venues and some tiers the maker fee is negative — a rebate, where the venue pays you to provide liquidity.
The rebate isn’t generosity. It’s funded by the taker fee, and the venue keeps the difference. The whole model is a transfer from participants who demand immediacy to participants who supply it, with the venue taking a cut in the middle.
Rates, tier structures and whether rebates exist at all differ everywhere and change, so this site doesn’t quote them. What’s durable is the shape: taking costs more than making, sometimes much more in relative terms.
Which one you are
Determined by what your order did on arrival, not by your intention or your direction.
Crossed the book on arrival → taker. Includes every market order and every limit order priced aggressively enough to match immediately.
Rested in the book → maker, whenever it eventually fills. A limit order that sits for an hour and is then hit by someone else’s market order was a maker order.
Direction is irrelevant. A buy can be either; so can a sell.
The consequence for anyone who cares about the distinction is that it’s decided by pricing, and post-only is the order type that guarantees the maker side by refusing to execute at all rather than crossing.
The four costs of a round trip
Buy and then sell, or the reverse. Illustrative, with synthetic figures throughout.
Suppose the book shows a best bid of 40,000 and a best ask of 40,004 — a spread of 4. Suppose an illustrative fee schedule of 0.02% maker and 0.05% taker. Suppose you trade 1 unit.
Cost one: the spread. Cross to buy at 40,004, cross back to sell at 40,000. That’s 4 lost, before anything else, purely from crossing twice. On a 40,000 notional, 4 is 0.01% — already comparable to the fee itself.
Cost two: fees. Taking twice at 0.05% of about 40,000 is roughly 20 on the buy and 20 on the sell: about 40 total.
Cost three: impact. Zero for 1 unit if the touch holds more than 1 unit. Not zero for larger orders, and it grows faster than linearly with size.
Cost four: the fee differential you didn’t collect. If you’d rested both sides instead, you’d have paid 0.02% twice — about 16 total instead of 40 — and you’d have earned the spread rather than paid it, since you’d be buying at the bid and selling at the ask.
So the taker round trip in this illustration costs about 44 (spread plus fees), and the maker round trip has a gross of about +4 from the spread minus about 16 in fees, for about −12. A difference of roughly 56 on a 40,000 notional, of which the headline fee difference accounts for less than half.
The rest is the spread, and the spread flips sign depending on which side of the book you’re on.
The mechanism
THE MECHANISM — the cost of a round trip
· Order crosses on arrival
→ taker fee, and you pay the
spread by buying the ask or
selling the bid
· Order rests and is later matched
→ maker fee or rebate, and you
earn the spread rather than
paying it
· Comparing venues on headline fee alone
→ INCOMPLETE. Spread and depth
often dominate the fee.
· Resting instead of crossing
→ cheaper per trade, and NO
GUARANTEE of a fill. The saving
is paid for in execution
uncertainty.
· Order size beyond the touch
→ impact adds on top of both fee
and spread, and scales faster
than linearly
· Rates, tiers, rebates, and whether fees
settle in the quote or a native asset
→ VENUE-SPECIFIC. All of it varies
and all of it changes.Tiers and where the schedule bends
Most venues tier fee rates by activity — typically trailing volume, sometimes holdings of a native asset, sometimes both. Rates fall as you move up.
Two structural points that outlast any particular schedule.
Tiers are usually computed on trailing volume over a rolling window, which means a rate can change without you doing anything, and comparisons based on your current rate may not describe next month.
The maker–taker gap usually narrows at higher tiers, and in some schedules the maker side turns into a rebate before the taker side reaches zero. Where the two rates sit relative to each other matters more than either alone for anyone whose costs are dominated by which side they’re on.
What a fee is charged on
A detail that produces confusion out of proportion to its size.
Fees are generally a percentage of notional value — quantity times price — not of profit and not of the margin posted. On a leveraged position this is the source of a common surprise: the fee scales with the full position size, so leverage multiplies the fee against your capital in the same way it multiplies everything else.
What the fee is denominated in also varies. Some venues charge in the quote currency, some in a native asset at a discount, some in the asset received. This affects your accounting and, if it’s a native asset, gives you a small additional exposure you may not have intended.
The failure mode
Optimising for the fee line is the wrong optimisation for most people, and the mistake it produces is concrete: choosing a venue with a lower published rate and a thinner book, then paying more in spread and impact than the rate saved.
The complete cost is spread plus fee plus impact, and their relative sizes depend on your order size against available depth. For small orders on a deep book, fees dominate. For larger orders, or on a thin book, impact dominates and the fee is a rounding error.
Neither the fee schedule nor the spread is visible in the other’s units, so comparing them requires doing the arithmetic on your own typical size. That’s the only way the comparison means anything — and this site takes no position on which venue anyone should reach as a result.