Pricing an item with no recent sales

Published 17 October 2026 by CompsAPI

You search the sold listings for something genuinely scarce and eBay shows you nothing, or two sales, one of which was a mistake. The item clearly has a value. The window you are looking through just does not contain it.

This is the most common pricing failure in collectibles, and it has a specific cause: eBay's public sold filter reaches back about 90 days. That is plenty for an iPhone and close to useless for a 1968 redline, a Panzer Dragoon Saga, or a card with a population under 200.

What not to do

Do not treat an empty window as a low price. Zero sales in 90 days means zero sales in 90 days. It carries no information about value, only about frequency.

Do not loosen the query until something appears. This is the reflex, and it is the worst option, because what appears is a different item. Drop the grade and you are pricing a raw card. Drop the model suffix and you are pricing a different lens. You will get a number, it will feel like progress, and it will be a comp for something you do not own.

Do not average in the active listings. Asking prices on a thin item are aspirational by definition, since the ones that were priced correctly already sold.

Four things that do work

1. Widen the window before you loosen the query. This is the whole trick and the order matters. Keep every filter exactly as it is and go from 90 days to 365, then to 1,095. You are looking for the same item, just over more time. A card with two sales in a quarter often has thirty in three years, which is a real distribution.

curl -G https://api.compsapi.com/v1/comps \
  -H "Authorization: Bearer YOUR_KEY" \
  --data-urlencode "q=panzer dragoon saga saturn" \
  --data-urlencode "days=1095"

2. Adjust for when the sale happened. A three year window buys you sample size and costs you currency. Look at whether the oldest sales sit clearly below or above the newest ones before you take a flat median across the whole period. For anything that has moved with a market, weight the recent end.

3. Price the band, not the point. On a thin item the 25th to 75th percentile is more honest than the median, because the median of nine sales moves if one of them was a bad day. Quote yourself a range and decide where in it you want to sit.

4. Use the nearest honest comparable, and say that is what you did. If the exact item truly has no history, step out one notch in a way you can describe: the same card one grade lower, the same console without the box, the same lens in a different mount. Record the step you took. A comp with a stated adjustment is defensible. A comp from a query you quietly broadened is not.

Accepted offers matter more here than anywhere

Thin markets run on offers. Expensive, slow moving items are exactly where buyers negotiate, and on an accepted Best Offer eBay keeps displaying the asking price rather than what was paid.

We measured the gap over 5,859 accepted offer sales: 99.0% of them displayed higher than the real figure, by a median of 14.9%, with the top tenth running over 40%. The full numbers are in the Best Offer gap study. On a thin item, where you may be working from four sales, two of them being asking prices rather than settled prices is not a rounding error, it is the whole answer.

How thin is too thin

A working rule, and it is a rule of thumb rather than statistics:

Our stats return count and used_for_stats on every response precisely so this is visible rather than buried. If you are building a pricing product, surface the count to your users too. The honest move when the data is thin is to say the data is thin.

The API reaches 1,095 days on every graded and raw search, and the free key needs no card.