Thin rails lie: why ranking by widest spread finds nothing worth trading

We built a table that ranked every payment rail on a market by spread, widest first, on the reasoning we have argued before: the crowded rails are competed flat, and the money is in the quiet ones.

Here is what came back at the top:

• Neteller — 11.08 to buy, 8.72 to sell
• Skrill — 9.80 to buy, 8.00 to sell
• Moneygram — 10.00 to buy, 9.00 to sell

A 27% gap on the first one. Meanwhile the busiest rail on that same board, a bank carrying 358 ads across both sides, quoted 9.77 and 9.73 — a spread of 0.41%.

The 27% number is not an opportunity anybody missed. It is what you get when one or two ads at round-number prices are the only thing on a rail.

Round numbers are the tell

Look at the prices again: 11.08, 8.72, 9.80, 8.00, 10.00, 9.00. When a rail's whole quote is built from a handful of ads, you are not reading a market — you are reading two people's opinions, and often those opinions are placeholders. An ad priced at exactly 10.00 against a market trading near 9.75 is not competing for flow. It is parked.

A rail with hundreds of ads cannot do this. The prices converge because merchants undercut each other continuously, which is exactly why the busy rail's 0.41% is a real number and the thin rail's 27% is not.

Widest-first sorting is a noise amplifier

The failure is structural, not incidental. Sorting by spread ranks rails by how far apart their two most extreme ads are. Fewer ads means more extreme, because there is nothing in between to compress them. So the ranking is, in effect, sorted by *lack of liquidity* — it reliably promotes the rails with the least going on.

That is the opposite of what a screen is supposed to do.

The fix is a floor, not a different sort

We now require at least two ads on each side before a rail is eligible, and order what survives by ad count rather than spread. On the same market, the list becomes:

• The busiest bank rail — 0.41% across 358 ads
• The second bank rail — 0.62% across 321 ads
• A third bank — 0.72% across 133 ads
• A cash rail — 0.20% across 95 ads

Those are spreads you could actually work with, attached to rails where somebody is on the other side.

What to take from it

• Always read a spread next to its ad count. A percentage on its own tells you nothing about whether the market exists.
• Be suspicious of round numbers. Real competitive quotes rarely land on 10.00 or 8.00.
• A screen that sorts by spread without a liquidity filter will show you noise at the top every single time, and it will look like it is doing its job.
• Thin is worth investigating. Empty is not. The line between them is whether anyone is quoting both sides.