Two kinds of number get published for Roblox items, and they behave nothing alike.
Between 22 July and 17 August 2026 — 26 days — the figures for 1,062 Pet Simulator 99 pets changed 7,971 times. Over the same 26 days, the figures for 1,051 items across five other games changed 68 times between them. Almost identical sample sizes, and a hundredfold difference in churn.
That gap is not a fact about which game's economy is busier. It is a fact about who publishes the number.
Two different kinds of number
Pet Simulator 99's figure comes from Big Games' own public API. It is RAP — Recent Auction Price — the game's rolling average of recent completed auction sales, in Diamonds. A server recomputes it as trades happen, and no person edits it.
RAP is not what a pet is worth, and not a price anyone is obliged to pay. It is a trailing average of sales that already happened, so a thinly traded pet can carry a RAP that no live offer comes near.
The other five figures come from community wikis on Fandom. Murder Mystery 2's item pages publish a figure in MM2 value points and cite MM2 Values for it. That number changes when a person edits the page. It is a community's written-down claim about the going rate — not a record of any particular trade.
One number is computed, the other is negotiated. The 26 days measure what that difference does.
What changed, and how often
Every item below was first read on 22 July 2026 and followed for the same 26 days. All six systems were read again on 17 August 2026, so none of these is a list that stopped being checked.
| Game | Where the figure comes from | Items | Items that changed | Changes |
|---|---|---|---|---|
| Pet Simulator 99 | Big Games API (RAP) | 1,062 | 870 (81.9%) | 7,971 |
| Murder Mystery 2 | Fandom wiki | 425 | 53 (12.5%) | 64 |
| Grow a Garden | Fandom wiki | 464 | 3 (0.6%) | 3 |
| Steal a Brainrot | Fandom wiki | 136 | 1 (0.7%) | 1 |
| Bee Swarm Simulator | Fandom wiki | 19 | 0 | 0 |
| Fisch | Fandom wiki | 7 | 0 | 0 |
The five wiki-sourced games hold 1,051 items between them and produced 68 changes. The one API-sourced game holds 1,062 and produced 7,971.
Pet Simulator 99: something moves every day
Of the 1,062 pets, 870 recorded at least one change — 81.9% of them. Those changes landed on all 20 days that saw any movement, between 266 and 700 a day. The three busiest days account for 1,562 of the 7,971, so this is a steady grind rather than a few big events.
A pet that moved at all moved a median of six times in the window. The most-changed pet recorded 22.
Direction was not symmetrical: 4,651 changes were downward and 3,320 upward. Over these 26 days more pets drifted down than up.
Angelus shows what daily recomputation looks like. Its RAP was different at all 17 readings:
| Date | RAP (Diamonds) |
|---|---|
| 22 Jul 2026 | 10,174 |
| 31 Jul 2026 | 5,192 |
| 2 Aug 2026 | 2,977 |
| 9 Aug 2026 | 12,595 |
| 13 Aug 2026 | 5,016 |
| 15 Aug 2026 | 3,981 |
From 10,174 down to 2,977, up to 12,595, and back to 3,981 inside three weeks. A figure read on 9 August and quoted a week later would have been wrong by a factor of three. Big Games' API returned 3,981 for Angelus when it was asked again on 17 August 2026.
Murder Mystery 2: the numbers move in batches
The wiki side is not frozen — it is bursty. Murder Mystery 2 recorded 64 changes across 53 items on 12 separate days. But 46 of those 64 fell on three days: 20 on 29 July, 18 on 11 August and 8 on 12 August.
That is the shape of an editing session. A contributor works through the list, revises what the community has agreed has moved, and saves. Between sessions the published figure sits still, whatever is happening in trades.
Chroma Sunset moved more than any other Murder Mystery 2 item in the window, and even that came to four changes:
| Date read | MM2 value points |
|---|---|
| 22 Jul 2026 | 6,500 |
| 29 Jul 2026 | 8,250 |
| 6 Aug 2026 | 8,750 |
| 9 Aug 2026 | 9,000 |
| 17 Aug 2026 | 9,350 |
The last step is datable to its edit. The Chroma Sunset wiki page was revised at 21:50 UTC on 16 August 2026, and the read the next morning picked up 9,350.
What this means if you are trading
The practical consequence is that the two numbers go stale at completely different speeds, and need to be treated differently.
A RAP figure has a short shelf life. Angelus was different at every reading, and the 12,595 recorded on 9 August had become 3,981 by 15 August — a figure six days old was off by more than three times. Quoting one without the date it was read says almost nothing.
A wiki value is the opposite problem. It will look identical for a week or more, which makes it feel dependable, but that stability is the update schedule rather than the market. In the 12 days between the 29 July and 11 August editing sessions, Murder Mystery 2's list recorded 12 changes across 425 items — the trading did not stop, the editing did.
Neither figure is a price. RAP is an average of completed sales, and a wiki value is a community's estimate of a fair trade. What someone will give you in a trade is its own question, and both numbers are at best an anchor for it.
How this was measured
Bloxism reads each value system's source on a daily cycle and writes an observation when a reading differs from the last one, or when a 7-day heartbeat comes due for an unchanged figure. A "change" above means one observation whose figure differs from the previous observation for the same item.
The cohort is every item first read on 22 July 2026, so each has the same 26 days of history behind it. That clause matters more than it looks: 818 of Fisch's 825 items and 292 of Steal a Brainrot's 428 were first read on 13 August, and counting those would have turned "no time to move" into a false "did not move". Only the 22 July cohort appears in the table, which is why Fisch shows 7 items rather than 825.
Two of the eight value systems on this site are left out of every comparison above. Adopt Me and Blox Fruits carry Bloxism's own editorial scale rather than a third-party source, so their zero changes describe a list nobody has revised, not a market. Using them as evidence would be citing ourselves.
Three limits are worth stating.
A reading more than four times away from an item's trusted history is quarantined for review and never published, which catches parser errors but also suppresses genuine four-fold moves. The change counts here are therefore a floor.
Daily reads catch daily net movement, not intraday movement. RAP that rose and fell back between two reads is recorded as no change.
Finally, a wiki figure is a claim, and it is reported here as one. What is measured on this site is when the published number changed — not whether the community was right about the going rate. The Big Games figures were checked against the live API on 17 August 2026 and the Murder Mystery 2 figures against the wiki's own revision history the same day; both matched the stored readings.
