Writing

The spelling was ours

13 November 2026 · figures measured 13 November 2026

Yesterday’s post said PegasusSwap’s real problem was coverage: seventeen pairs, where the widest providers quote fifty or sixty. Their team wrote back the same day. Their own logs show 75% of requests answered in October, not 35%, and their site lists far more than seventeen pairs. They asked what our figure was based on.

The honest answer turned out to be: mostly a spelling mistake, and it was ours.

What was sent

Their API asks for the coin and its network separately. Their documentation shows four network names, BTC, TRX, ERC20 and XMR, and the list of coins cannot be read without a key. So this site sent ERC20 for anything on Ethereum, and passed every other network through as written: LITECOIN, SOLANA, STELLAR.

They want the chain’s ticker: ETH, LTC, SOL, XLM. USDT on Ethereum is “ETH”, not “ERC20”. Every wrong name came back as a 422 that said “Pair unavailable”, the same words they use for a pair they really do not run. That is why it read as missing coverage.

RequestSentAnswer
ETH to XMRERC20Pair unavailable
ETH to XMRETH1.0987 XMR
LTC to XMRLITECOINNot available
LTC to XMRLTC0.9075 XMR
BTC to USDT on EthereumERC20Pair unavailable
BTC to USDT on EthereumETH706.13 USDT

Probed on 13 November with this site’s own key.

How much of the record it was

Of PegasusSwap’s 4,026 misses in the record, 2,829 were on a network we had misspelt. Those were never really a question put to them, so they are no longer counted as one. They stay in the raw record, because that is what happened.

PegasusSwapBeforeNow
Answered34.9%64.4%
On pairs it trades89.8%89.7%

What is left is real. Zcash is not offered, in any spelling, and some amounts are below their minimum. The second figure will move again as the newly reachable pairs collect answers: until now, ETH, LTC and SOL never counted as pairs it trades. Its page shows the figures as they change.

Why it took a month

A wrong network name is refused in the same words as a pair that does not exist. Nothing on our side separated the two, so a translation bug looked exactly like a narrow catalogue. It was the provider whose figure was disputed, with its own numbers, that made it worth checking.

The 75% in their logs and the figures here still measure different things: theirs counts every partner’s requests, ours only the ones this site sent. But the gap between them was not a difference of method. Most of it was a bug.

Plan an exchange

Every figure above can be checked against the providers yourself — no account, no sign-up, and nothing is held for you at any point.