Every number,accounted for.
Clustermelt publishes a price only when the market supports it. Here is how each figure is built: where the data comes from, how it is weighed, and what we refuse to publish.
Computed, never invented
This is the one rule everything else serves.
Every price on this site is computed from observed secondary-market data, not estimated or guessed. As of July 17, 2026, the live catalog reading draws on 166 observations across 28 vendor sources spanning 12 models. When the data does not support a number, we say so rather than publish one.
What feeds a price
Three kinds of observation, in ascending order of how much they are trusted.
Dealer and marketplace listings
The backbone today: live asking prices from established secondary-market dealers and marketplaces, captured with their date and source so each observation is traceable.
Supplier quotes
As suppliers join the network and quote real hardware, those numbers enter the read at a higher reliability than an open listing, because they are a price someone is prepared to transact at.
Closed deals
The strongest signal of all: a price at which hardware actually changed hands. As Clustermelt closes deals, each one deepens the read and carries the most weight.
How the observations combine
Sources do not vote equally, and the arithmetic is fixed.
Reliability
Not every observation counts the same. A price someone actually transacted at outweighs an open listing, and a listing from a documented dealer outweighs an anonymous one. Higher-confidence sources move the number more.
Recency
This market moves. Recent observations count for more than old ones, and observations past a certain age drop out of the read entirely rather than anchor it to a stale market.
Determinism
The aggregation is arithmetic, not opinion. The same inputs always produce the same number: there is no human thumb on the scale, and no figure is ever estimated to fill a gap.
What we will and will not publish
A number that cannot be stood behind does not ship.
A SKU gets a published price only when there are enough independent, recent observations to stand behind it. Where a model is thin, we flag it rather than fill the gap with a guess. That discipline is the whole point: the number is only useful if it is real.
The precise source weights, the exact recency window, the per-SKU observation counts, and the deeper per-condition read are part of the monthly market-intelligence issue. This page is the principle, not the full recipe.
The published band is the envelope over used and refurbished listings that clear our checks. It tells you what the market is doing. It is not a quote: your actual price depends on condition, quantity, timing, and provenance.
The rules we hold to
- 01
Refreshed monthly, so the read tracks the market rather than a launch-day snapshot.
- 02
Each monthly issue freezes its own numbers: a published figure is never quietly changed after the fact.
- 03
The live catalog reading moves as new observations land; every published issue stays fixed as a point-in-time record.
- 04
No number is ever invented. Where the data is too thin to stand behind, we flag the gap instead of filling it.
The monthly read is where this methodology is applied in full. See the current market-intelligence issue.
Want the number for your build?
Tell us what you need or what you have. We handle the rest, from first conversation to final payment.