Two exchanges show different prices for the same coin. Buy on the cheap one, sell on the expensive one, pocket the difference. It is the most intuitive trading idea there is, and the reason most people who try it lose money is that the intuitive version leaves out everything that matters.
This guide covers the varieties of crypto arbitrage that still work, the costs that quietly consume the spread, and how to recognise the gaps that exist for a reason.
Why price gaps exist at all
In a market with instant, frictionless transfers, arbitrage would close gaps immediately and none would persist. Crypto is not that market.
Moving an asset between exchanges takes time and costs money. Withdrawals can be paused. Some venues restrict certain jurisdictions. Fiat rails are slow. Each of these frictions allows a price difference to persist, and the size of a persistent gap is usually a measure of how much friction sits between the two venues.
That is the first rule worth internalising: a large spread is not a large opportunity, it is a large obstacle.
Spatial arbitrage
The classic form. Same asset, two exchanges, different prices.
The naive execution is to buy on exchange A, withdraw, deposit to exchange B, and sell. This almost never works. Between the buy and the sell you carry full market risk for however long the transfer takes, which can be minutes or hours. A 2 percent spread means nothing if the asset moves 5 percent while your withdrawal confirms.
The professional execution avoids transfers entirely. You pre-fund balances on both venues, buy on the cheap one and sell on the expensive one simultaneously, and rebalance later when it is convenient. This removes the transfer risk but requires capital sitting idle on multiple exchanges.
Costs to account for before deciding a spread is profitable:
- Taker fees on both sides, which for a retail account can be 0.1 percent each
- Withdrawal fee on whichever leg needs rebalancing
- Slippage, which grows with size and is worse on the thinner venue
- Spread, since you are usually crossing the book on both sides
Add those up and a 0.4 percent gap is frequently a losing trade.
Triangular arbitrage
Within a single exchange, three pairs can be priced inconsistently. Convert asset A to B, B to C, then C back to A, and finish with more than you started.
Because everything happens on one venue, there are no transfers and no transfer risk. The problem is speed. These inconsistencies exist for fractions of a second and are captured by bots with co-located infrastructure. Attempting this manually is not viable, and attempting it with a retail API connection and consumer internet is barely better.
Triangular arbitrage is real and it is profitable for firms operating at that level. For most individuals it is a study topic rather than a strategy.
Funding rate arbitrage
This one is genuinely accessible, and it is where most retail arbitrage capital actually goes.
Perpetual futures use a funding rate to keep their price tethered to spot. When the perpetual trades above spot, longs pay shorts. When it trades below, shorts pay longs.
A delta-neutral funding trade holds spot and shorts the perpetual in equal size. Price movement nets out, and you collect funding while the rate stays positive.
During periods of strong bullish sentiment, funding can run at rates that annualise into the high double digits. That is a real yield for a position with no directional exposure.
What can go wrong:
- Funding flips negative. Sentiment turns and you start paying instead of receiving. The position needs monitoring, not setting and forgetting.
- Liquidation on the short leg. If your spot is on one venue and the short on another, a sharp rally can liquidate the short before you can move collateral. Margin buffer is not optional.
- Exchange risk. You are holding balances on a venue. That has been the largest single source of loss in crypto history, and no funding yield compensates for a failed exchange.
Cash-and-carry arbitrage
The dated-futures version of the same idea. When a quarterly future trades at a premium to spot, you buy spot and short the future. At expiry the two converge and you capture the premium.
The advantage over funding arbitrage is certainty. The premium is known at entry and the convergence date is fixed. The disadvantage is that capital is locked until expiry, and the annualised return is usually lower than peak funding rates.
Basis trades widen when the market is euphoric and compress when it is fearful, which means the best entries appear when everyone is bullish.
How to tell a trap from an opportunity
Before acting on any spread, work through this list.
- Can you actually withdraw the asset from the cheap venue? Suspended withdrawals are the most common explanation for a persistent discount. The coins are cheap because they are stuck.
- Is the order book deep enough? A price on a screen with 200 dollars of depth behind it is not a price you can trade in size.
- Does the venue have a real withdrawal history? Check that other users are moving funds out normally.
- Is the pair quoted in the same asset? A discount against a depegging stablecoin is not a discount.
- What is the all-in cost? Write out every fee before deciding the spread is profitable, not after.
If a gap looks unusually large and none of the above explains it, assume you are missing something. That assumption will save you more money than it costs.
Capital, infrastructure and honest expectations
Arbitrage is a low-margin, high-turnover business. Returns per trade are small, so profitability depends on doing it repeatedly with meaningful size and low costs.
That has several consequences for an individual. You need capital spread across venues, which means exchange risk multiplied. You need fee tiers that retail volume does not reach. And you are competing with firms whose entire operation is optimised for this.
The realistic retail versions are funding rate and cash-and-carry trades, where the edge comes from accepting a specific risk rather than from being faster than someone else. Those are worth learning. Chasing spot price differences across exchanges by hand is generally a way to convert time and fees into a small loss.
The tax and record-keeping side
High trade counts create a record-keeping burden that catches people out. Every leg is potentially a taxable event depending on jurisdiction, and a strategy that nets a modest return can generate thousands of transactions in a year.
Decide how you will track this before you start, not in April. Rules differ substantially by country, and a professional who understands crypto in your jurisdiction is worth the fee if you intend to trade at any scale.