Data sources
Rolimons item API — values, RAP, ratings
Item names, community values, RAP, demand and trend ratings, and the projected and rare flags come from the public Rolimons item API. We fetch the full catalogue at most once per hour and serve every page from that shared snapshot — pages state when they are running on a cached copy instead of a live fetch.
Coverage limits, stated plainly: only roughly a fifth of catalogue items carry an assigned community value at any given time; the rest are priced on RAP alone and are labelled “RAP only” wherever they appear. A majority of items likewise carry no demand or trend rating; the model treats unrated as neutral, never as negative. Rolimons' values are themselves editorial estimates by their value team — informed judgements, not exchange prices.
Roblox economy API — price history
Price and sales-volume history charts on item pages come from Roblox's own public marketplace endpoint, along with the number of copies of each item in existence. Three honest caveats.
First, the charts are a periodically refreshed copy, not a live read. Roblox rate-limits this data to 50 requests a minute, and a build that renders every item page at once would blow through that and bake errors into pages for an hour, so the series is captured in one deliberate pass and committed. Every chart therefore prints the last day it covers, and that date — not today — is what the change figures are measured to.
Second, the feed occasionally contains glitched recorded days. Verified examples on a single item: a 1-Robux average, and a 3,847,400 spike on a day whose neighbours sat near 297,000. Such days are excluded from charts and statistics, with the count disclosed on the page. The test is deliberately a local one — a day is judged against the days either side of it, not against the item's whole history — because those are different questions. An item that genuinely tripled over four years has legitimate prices well above its all-time median, while a bad record stands alone. Comparing against the neighbourhood keeps the first and drops the second; on current data it removes between 0.1% and 1.2% of days per item.
Third, most limiteds barely sell, and the pages say so rather than dressing it up. Items with no recorded sales get an explanation instead of an empty chart; items with only a handful get their sale days listed individually, because a line drawn through three points two years apart would imply a trend that is not there. Only items with at least eight recorded days are charted.
One correction worth recording, since this page exists to be checkable: until September 2026 this site read Roblox's legacy economy/v1 resale route and told visitors the feed had stopped in early 2025. It had not. Roblox had re-keyed limiteds onto a new identifier and moved the live series with it, leaving the old route answering successfully with frozen data. The charts now come from the current route, and reach yesterday rather than last year.
Static reference data
The currency converter uses fixed reference rates (Robux retail and DevEx rates, USD/EUR/GBP), with the last-updated date printed on the page. The XP predictor uses approximate community-calibrated curves, labelled as such on the page — they are starting points to override with your own measured rate, not published game formulas.
The trade model
The calculator scores each side of a trade in four steps. All constants are listed below exactly as the code uses them; there are no hidden adjustments.
- Base price. Each item is priced at its community value where one exists, otherwise its RAP. The verdict's confidence rating drops whenever RAP-only items are involved.
- Demand and trend weighting. Rated items are adjusted by the multipliers in the table below. The spans are deliberately modest — ratings refine a price rather than replace it.
- Projected penalty. Items flagged as manipulated are priced at 75% of list (-25%), and their presence forces confidence to low.
- Bulk discount. Each item beyond the first on a side costs 3.5%, capped at 25% — the mechanical form of the trader's rule that a pile is harder to move than a single clean item.
| Rating | Demand multiplier | Trend multiplier |
|---|---|---|
| Terrible / Lowering | -15% | -5% |
| Low / Unstable | -7% | -2% |
| Normal / Stable | 0% | 0% |
| High / Raising | +8% | +5% |
| Amazing / Fluctuating | +15% | -2% |
| Unrated | 0% | 0% |
| Rare flag | +2% | |
The two adjusted sides are compared, and anything inside ±5% is reported as fair — value estimates are approximations, and pretending to finer resolution than a few percent would be false precision. A trade with either side empty gets no verdict at all.
How the constants were chosen — honestly
The constants encode widely held trading heuristics — demand moves prices a little, projection ruins them, bulk trades at a discount — at magnitudes set by judgement, not fitted to data. They have not been backtested against resale history, and until they are, that is the model's biggest limitation. We would rather state that than imply a rigour the model does not have. The direction of every adjustment is defensible; the exact size of each is an estimate.
What this site never does
- Invent a value, rating or price. Every figure traces to a source named on this page.
- Present a verdict without disclosing which numbers it stands on.
- Guarantee an outcome. The model is a sanity check, not an oracle — trades are decisions.
Found a number this page cannot explain? That is a bug. Tell us.