Why Retail Crypto Arbitrage Usually Fails
Arbitrage looks like free money on paper. In practice it is dominated by HFT firms running on co-located infrastructure. Here is what actually works for retail.
Crypto arbitrage is one of the most-promised, least-delivered strategies in retail trading. The pitch is irresistible — buy low on Exchange A, sell high on Exchange B, capture the spread. The reality involves microseconds, deep order books, and infrastructure costs that make most retail attempts unprofitable before they start. But there are versions of arbitrage that still work.
The myth: simple price difference capture
The classic pitch — "BTC is $60,000 on one exchange and $60,050 on another, just buy and sell" — has not worked at retail scale for years. Those spreads close in milliseconds, captured by HFT firms with co-located servers, low-latency network paths, and pre-funded accounts on every venue.
By the time your VPS-hosted Python script sees the spread, places the buy, and tries to place the sell, the spread is gone — and you are now long an asset on one exchange with no offsetting hedge. Congratulations, you are speculating.
The reality: latency, withdrawal delays, and fees
Three things kill retail spatial arbitrage. Latency — pro firms beat you by 100x on order placement. Withdrawal delays — moving the asset between exchanges can take minutes to hours, by which time the spread has reversed. Fees — taker fees on both legs plus withdrawal costs often exceed the spread you were capturing.
Even pre-funding both exchanges (so you do not need to move funds for each trade) does not fix the latency problem. You are still racing firms with infrastructure budgets in the millions.
What actually works for retail
Triangular arbitrage on a single exchange is more accessible. The latency game is easier (one venue, one API), and the spreads are smaller but more consistent. Funding-rate arbitrage on perpetual futures is another viable path — capture the spread between funding payments on long vs short positions across venues.
Cross-venue arbitrage between a CEX and a DEX is also worth a look — the latency profile is different (block-time matters more than ms-level networking), so retail is not racing the same firms. Returns are still modest after gas, but the playing field is closer to level.
- →Triangular arbitrage — single exchange, three pairs.
- →Funding-rate arbitrage — perpetual futures across venues.
- →CEX-DEX arbitrage — block-time games, not microsecond games.
The opportunity cost of trying
For most retail traders, the time spent fighting arbitrage is better spent on a strategy with a real edge — momentum, trend-following, mean-reversion in defined regimes. Our crypto trading bots cover all of those. The marginal expected return on retail spatial arb is usually negative once you honestly account for fees, infrastructure, and time.
If you do want exposure to spread-capture profits, the realistic path is depositing into market-maker funds or trading the funding spread on perpetuals — not running a homebrew arb bot. For one-screen execution across 9+ venues, our multi-exchange trading terminal is the practical alternative.
Frequently asked questions
Why does retail crypto arbitrage usually fail?
Three reasons. Latency — HFT firms with co-located servers beat retail by ~100x on order placement. Withdrawal delays — moving assets between exchanges takes minutes to hours, by which time the spread has reversed. Fees — taker fees on both legs plus withdrawal costs typically exceed the spread you were trying to capture.
Can retail traders make money with crypto arbitrage?
Sometimes — but not via the classic "buy low on A, sell high on B" pitch. That game belongs to HFT firms. Retail-viable variants are triangular arbitrage on a single exchange, funding-rate arbitrage on perpetual futures, and CEX-DEX arbitrage where block-time matters more than microsecond networking.
What types of crypto arbitrage actually work for retail?
Triangular arbitrage on one exchange (no inter-venue latency), funding-rate arbitrage between perpetuals across venues (capture the funding spread, not the price spread), and CEX-DEX arbitrage where the latency profile is block-time-bound. Returns are modest but the playing field is closer to level than spatial CEX arb.
Try it on Tr8de.ai
Use our multi-exchange terminal for spreads you can act on