← The Evidence

Exhibit E Investigation · 2026-08-02 · 2 min read

The Two-Exchange Arbitrage "Free Lunch"

Verdict: Overturned
The price gap between exchanges is real — and about fifty times too small to survive the fees.

The charge. It's the first idea everyone has: Bitcoin is one price on Kraken and a slightly different price on Coinbase, so buy on the cheaper venue, sell on the pricier one, and pocket the difference. Free money. This very site grew out of a crypto-arbitrage bot built on exactly that premise. The bot never made a dollar. Here's the autopsy.

The evidence

We pulled live top-of-book prices on both exchanges and computed the best possible cross-exchange edge — buy at the lower ask, sell at the higher bid — before any costs:

Coin Best cross-exchange edge
BTC +0.014%
ETH 0.000%
SOL 0.000%
XRP −0.014%
DOGE −0.006%

The gaps are essentially zero — a rounding error. Modern exchanges are stitched together by thousands of bots that erase any real gap in milliseconds. By the time a human sees it, it's already gone.

Now the fees

To capture even that sliver you have to take liquidity on both sides — realistically 0.4% to 0.9% round-trip once you count taker fees on each leg. So you're trying to harvest a 0.01% edge with a 0.5%+ cost. This isn't picking up pennies in front of a steamroller. The pennies aren't even there, and the steamroller is.

The verdict

Retail two-exchange arbitrage is a myth that dies on contact with the fee schedule — the same wall that flattens the 15-minute markets. The edge is real and far too small to keep. The one honest version — being the bot that provides the liquidity everyone else pays to take — is a latency-and-capital business, not a free lunch.

That failed bot did leave us something, though: the lesson this entire site is built on. Measure before you believe.

← Back to The Evidence · See the graded record