Exchange Spreads: The Hidden Cost That Decides Your Arbitrage P&L
NDA-Trade Research · · 8 min read
In two-leg arbitrage, a cross-exchange spread is the price difference between the venue where you open your long leg and the venue where you open your short leg for the same asset. If a perpetual contract trades at $100.05 on the exchange where you sell and $100.00 on the exchange where you buy, the spread works in your favor; if the prices are reversed, you pay the difference the moment you enter. That difference — small, ever-changing, and often ignored — is one of the largest determinants of whether a funding-rate arbitrage trade actually makes money.
Why the Spread Is the Hidden Line Item
Funding-rate arbitrage — explained in depth in our guide to what funding rate arbitrage is — earns money from periodic funding payments while holding a delta-neutral position: long on one venue, short on another. Because the position is market-neutral, the funding income is the visible part of the P&L. Everything else is cost, and the largest costs are the ones that never appear as a line on your fee statement: the price gaps you cross when you open and close each leg.
Traders comparing crypto exchange spreads often look only at the moment of entry. But a delta-neutral position is a round trip by construction: you will eventually unwind both legs, and conditions at exit are rarely the mirror image of those at entry. Arbitrage execution costs are therefore four numbers, not one — entry spread, exit spread, fees on every fill, and slippage on top.
You Pay the Spread Twice: Entry Plus Exit
A delta-neutral position has two transitions, each involving two simultaneous fills:
- Entry: buy the long leg on exchange A, sell (short) on exchange B. The entry spread is the cost — expressed as a percentage of notional — of opening both legs at prevailing prices.
- Exit: sell the long leg on A, buy back the short on B. The exit spread is the cost of closing both legs.
The round-trip spread is simply entry spread plus exit spread. This is the number that matters, because you cannot bank funding income without eventually paying the exit toll. A trade that looks attractive at entry can be underwater on a round-trip basis if the exit spread is habitually wide — for example, when one venue consistently trades at a premium you must cross on the way out. A useful mental model: the spread is rent paid for moving in and out; funding is income earned while holding. The trade works only if cumulative funding exceeds round-trip rent plus fees before you close.
Taker vs Maker: The Cost–Certainty Trade-Off
Every leg can be executed two ways, and the choice changes both the spread you pay and the risk you carry.
Taker (market) execution crosses the order book: you buy at the ask and sell at the bid, immediately. You pay the bid–ask spread on each venue on top of the cross-exchange price gap, plus taker fees, which are typically higher than maker fees. In exchange, you get certainty — both legs fill at once and you are delta-neutral from the first second.
Maker (limit) execution posts resting orders inside or at the spread: you buy at the bid and sell at the ask, so the quoted maker-maker spread is usually tighter — often negative, meaning the market pays you to enter. Maker fees are lower and are sometimes rebated. The price is fill risk (leg risk): one leg may fill while the other does not, leaving you temporarily directional. If the market moves against the unhedged leg, the loss can dwarf the spread you saved.
| Taker (market) | Maker (limit) | |
|---|---|---|
| Effective spread | Wider — you cross both books | Tighter, sometimes favorable |
| Fees | Higher taker fees on every fill | Lower maker fees, possible rebates |
| Fill certainty | Immediate, both legs at once | No guarantee; orders may not fill |
| Leg risk | Minimal | Real — one leg can fill alone |
| Best suited for | Fast-moving opportunities, unwinding under stress | Patient entries when spreads mean-revert |
Neither mode is universally correct. Many desks enter with limit orders when they can afford to wait and reserve market orders for exits that must happen now — for instance, when funding flips sign or margin needs attention.
A Worked Example: How a “Small” Spread Eats a Week of Funding
All numbers below are illustrative, chosen for round arithmetic — they are not live market statistics.
Suppose a pair offers a funding-rate differential worth 15% APR to the delta-neutral holder. That sounds substantial, but per day it is:
15% ÷ 365 ≈ 0.041% per day of notional.
Now suppose execution, done with taker orders, costs:
| Cost component | Illustrative value |
|---|---|
| Entry spread (taker-taker) | 0.05% |
| Exit spread (taker-taker) | 0.05% |
| Taker fees (4 fills × 0.05%) | 0.20% |
| Round-trip execution cost | 0.30% |
Breakeven holding period = round-trip cost ÷ daily funding income:
0.30% ÷ 0.041% ≈ 7.3 days.
For an entire week, every funding payment goes to paying off the entry and exit — only after day eight does the trade earn anything. If funding compresses to 8% APR mid-hold (≈0.022% per day), breakeven stretches to roughly 14 days. And if you were forced to exit early through a temporarily wide 0.15% exit spread, the same trade closes at a loss despite the funding leg performing exactly as advertised.
Run the same trade with maker execution — say a 0.00% combined entry and exit spread and 0.02% maker fees per fill (0.08% total) — and breakeven drops to about two days. That gap between the two rows is why “maker vs taker spread” is not a detail; it is often the whole edge.
Why the Live Spread Is Not Enough
Spreads between exchanges are not static. They widen when one venue leads a move, when liquidity is pulled around news, or when funding itself attracts one-sided flow — and they tend to mean-revert toward a pair-specific typical level afterward. A snapshot of the live spread tells you what execution costs right now; it tells you nothing about whether now is a good time.
Historical distribution statistics fix that:
- Average over a window shows the typical cost you should budget for.
- 10th percentile (p10) shows the favorable tail — the entry level that occurs often enough to realistically wait for.
- 90th percentile (p90) shows the adverse tail — roughly what a forced, badly timed exit could cost.
- Min/max extremes bound the best and worst observed prints, useful for stress-testing assumptions.
If the live entry spread sits near its p90, you are being asked to pay close to the worst observed cost; patience will usually be rewarded. If it sits at or below p10, history says such opportunities are scarce and short-lived — that is when acting quickly is justified.
Timing Entries and Exits Against the Distribution
The practical workflow follows directly:
- Enter when the entry spread is cheap relative to its own history — at or below its average, ideally near p10. An absolute number like “0.03%” means nothing without the pair’s distribution; 0.03% may be a bargain on one venue pair and expensive on another.
- Budget the exit at its average or worse, not its best. Assuming you will exit at the p10 exit spread is planning around luck.
- Watch the round-trip total, not each side alone. A generous entry spread can be a trap if the exit side of the same route is structurally wide.
- Pre-decide your exit mode. If the thesis breaks, you will likely exit as a taker at whatever spread prevails — one more reason the p90 exit figure belongs in your sizing math.
Fees and Slippage: The Other Two Costs
The spread is the largest hidden cost, but two companions travel with it. Fees are explicit: a round trip involves four fills, so per-fill fees multiply by four, and the taker/maker fee difference compounds the spread difference discussed above. Slippage is the gap between the quoted price and your actual fill: large orders walk the book beyond the top-of-book quote, so effective spreads deteriorate with size. Quoted spreads assume top-of-book liquidity; sizing beyond it means your realized round-trip cost will exceed any screen number. (Definitions of these and related terms are collected in our glossary.)
How NDA-Trade Measures Execution Cost
NDA-Trade computes, for every tracked pair and venue combination, the entry spread and exit spread in both taker-taker (market) and maker-maker (limit) variants, plus the round-trip total. Alongside the live values, the platform maintains historical statistics from 1-minute snapshots over a rolling window: the average, the 10th and 90th percentiles, and the observed minimum and maximum. That lets you see at a glance whether the current entry cost is cheap or dear relative to the pair’s own history — the exact comparison the timing rules above require. The live table is at /exchange-spreads.
FAQ
What is a good entry spread?
There is no universal threshold — “good” is relative to the pair’s own distribution. An entry spread at or below the 10th percentile of its recent history is objectively cheap for that route; one near the 90th percentile is expensive regardless of its absolute size. Always evaluate the entry together with the typical exit spread, since only the round-trip total determines profitability.
Maker or taker for arbitrage legs?
Maker (limit) execution is cheaper — tighter effective spreads and lower fees — but carries fill risk: one leg can fill while the other does not, leaving you briefly directional. Taker (market) execution guarantees both legs fill immediately at a higher cost. Many traders use maker orders for patient entries and accept taker costs for exits that cannot wait.
What is a round-trip spread?
It is the sum of the entry spread (cost to open both legs of a delta-neutral position) and the exit spread (cost to close both legs). Because every arbitrage position must eventually be unwound, the round trip — not the entry alone — is the execution cost your funding income has to beat. Comparing round-trip cost to daily funding income gives your breakeven holding period.
Why do spreads differ between exchanges?
Each venue has its own order book, participants, fee schedule, and liquidity depth, so the same contract prices slightly differently on each. Spreads widen when one exchange leads a price move, when market makers pull quotes, or when funding attracts one-sided flow to a venue. These dislocations tend to mean-revert, which is why historical percentiles are more informative than any single live reading.
This article is for educational purposes only and does not constitute financial, investment, or trading advice. Cryptocurrency derivatives involve substantial risk, and all figures above are illustrative hypotheticals, not market data.