If you scan a P2P board the obvious way — page through ads sorted by price — you'll find the tightest, most competitive spread on the board every time. That's exactly the problem. The tightest spread is where everyone already is.
Ads are returned price-ordered, not rail-ordered
A board scan capped at a fixed number of pages returns the most competitive ads across every rail combined, sorted by price. It systematically discards rails whose best quote sits deeper in the book — which is precisely where the wide, uncontested spreads live. A fixed page cap doesn't just miss some opportunities at random; it specifically deletes the ones that matter.
A measured example
On one market, a single bank rail carried over 1,011 active ads — crowded, and thin, because a thousand merchants competing for the same flow compresses the spread toward nothing. Meanwhile a rail with three ads and no real competition can carry a spread many times wider. Price-ordered scanning keeps the crowded, thin rail on page one and buries the quiet, fat one — the tail is the opportunity, and a naive scan never reaches it.
Two ways to scan, and when each wins
There are two legitimate strategies: page through the whole board and hope every rail turns up (board paging), or query each rail directly and take its own top ads (per-rail query). Cost differs by board shape — board paging is cheaper exactly when the number of ads per side is less than 20 times the rail count. On a market with 324 ads and 25 rails, board paging costs 34 requests against 50 for querying every rail directly; on a market with a couple thousand ads and 40 rails, per-rail querying wins by a wide margin. The right choice is a property of the specific board, not a fixed rule.
Why this matters for makers
- A rail's low ad count is a signal to look closer, not a reason to skip it — thin doesn't mean illiquid, it can mean uncontested.
- Any tool that caps how many ads it reads per board is implicitly biased toward the crowded rails and against the quiet ones, whether it says so or not.
- The opportunity a maker is actually looking for — a wide, real spread with room to place — structurally cannot show up on the first page of a naive scan.