Rarest Items in Dig & Clean: Anomaly Trophy List and Formula
Last updated: 2026-08-20 · Play on Roblox ↗
Update 2 extends the verified trophy-card range to 1 in 191 billion. A MINT Great White Shark card showed 1 in 191,220,577,344 and 3.5B in S2. A Good T-Rex shown as a new discovery in S3 read 1 in 110,274,895,491 and 2.4B. Each card is a single-recording read; the formula check across all accepted rows is the stronger cross-row result.
For the derivation and economy behind the numbers, use the separate item-values guide.
Omega then Anomaly
Update 1 added “New Rarity: Omega”, double-read in the in-game changelog in S8 and S7. Update 2 then added Anomaly. The observed ladder above the original conditions and tiers is therefore Omega (Update 1) → Anomaly (Update 2). Across the footage reviewed through August 20, an item card explicitly labeled Omega was not shown clearly enough for card-level verification.
Update 2 adds Anomaly rarity
The in-game changelog names “New Rarity: Anomaly”, confirmed in two independent recordings, S2 and S5. In a targeted recheck of S2, the word “Anomaly” rendered in green. The new top collection cards in that same recording used a green border with a green-to-black gradient.
Color is not a load-bearing identification rule here. S3 rendered its top-tier card highlight as pink/magenta, so the card-color treatment conflicts between recordings. This guide uses green as the double-checked S2 read while keeping the conflicting S3 rendering visible.
Update 2 trophy table
Every row below is a single-recording card read and also falls within ±5%
of 45 × odds^0.7. S2 supplies twelve collection-card rows; S3 supplies the
T-Rex row.
| Item | Condition | Rarity | Value | Source |
|---|---|---|---|---|
| T-Rex | Good | 1 in 110,274,895,491 | 2.4B | S3 |
| Great White Shark | MINT | 1 in 191,220,577,344 | 3.5B | S2 |
| Longship | MINT | 1 in 17,912,412,190 | 675.9M | S2 |
| Angel | MINT | 1 in 11,583,164,116 | 498.2M | S2 |
| Angel | Great | 1 in 9,500,000,000 | 433.6M | S2 |
| Angel | Good | 1 in 1,870,000,000 | 139.1M | S2 |
| Nuke | MINT | 1 in 508,801,830 | 55.8M | S2 |
| UFO | Common | 1 in 247,000,000 | 33.6M | S2 |
| Nuke | Good | 1 in 131,500,000 | 21.6M | S2 |
| Giant Squid | Common | 1 in 123,500,000 | 20.7M | S2 |
| Krakens Eye | OK | 1 in 90,000,000 | 16.6M | S2 |
| Frozen Mammoth | Great | 1 in 19,515,874 | 5.7M | S2 |
| Midnight Diamond | Perfect | 1 in 15,000,000 | 4.9M | S2 |
The S3 T-Rex card also showed 72,287.6 KG and “(new item discovered)”.
Formula validation now reaches 1 in 191 billion
The existing fitted rule is:
value ≈ 45 × (rarity odds)0.7
It was previously verified against 22 cards up to roughly 1 in 12.6M. The accepted Update 2 cards extend that check by four more orders of magnitude, reaching 1 in 191 BILLION. The predicted values and errors below use the supplied displayed-value comparison exactly.
| Item | Rarity | Actual | Predicted | Error |
|---|---|---|---|---|
| T-Rex | 1 in 110,274,895,491 | 2.4B | 2.415B | +0.6% |
| Great White Shark | 1 in 191,220,577,344 | 3.5B | 3.551B | +1.5% |
| Longship | 1 in 17,912,412,190 | 675.9M | 676.8M | +0.1% |
| Angel MINT | 1 in 11,583,164,116 | 498.2M | 498.7M | +0.1% |
| Nuke MINT | 1 in 508,801,830 | 55.8M | 55.95M | +0.3% |
| UFO | 1 in 247,000,000 | 33.6M | 33.74M | +0.4% |
| Krakens Eye | 1 in 90,000,000 | 16.6M | 16.64M | +0.2% |
| Frozen Mammoth | 1 in 19,515,874 | 5.7M | 5.71M | +0.2% |
| Midnight Diamond | 1 in 15,000,000 | 4.9M | 4.75M | -3.1% |
The same two Update 2 recordings also contained roughly nine readable cards whose numbers we could not reconcile. Several were off the formula by 2×–150×, and three backpack rows showed a value exactly equal to the odds divided by 1,000—a pattern that suggests a transcription error in our video read rather than in-game data. We exclude those rows pending a second view rather than publish numbers we cannot stand behind.
Earlier 22-card independent check
Before Update 2, the Bax recording supplied an out-of-sample test: 22
different item cards from an independent video, spanning rarity odds from 6
to more than 12.5 million. We recalculated every prediction directly from the
displayed 1 in N number.
Error below means (prediction − displayed value) / displayed value. Values
shown as K or M inherit the video's rounding, so those errors have less
precision than the small-value rows.
| Item | Rarity | Actual | Predicted | Error |
|---|---|---|---|---|
| Rusty Can | 1 in 28 | 463 | 464 | +0.15% |
| Duck | 1 in 22 | 391 | 392 | +0.17% |
| Newspaper Hat | 1 in 42 | 615 | 616 | +0.14% |
| Bear | 1 in 34 | 530 | 531 | +0.22% |
| Glass Bottle | 1 in 6 | 157 | 158 | +0.47% |
| Beach Ball | 1 in 507 | 3.5K | 3,521 | +0.61% |
| Vase | 1 in 1,250 | 6.6K | 6,623 | +0.35% |
| Sword | 1 in 4,840 | 17K | 17,085 | +0.50% |
| Guitar | 1 in 10,000 | 28.3K | 28,393 | +0.33% |
| Anchor | 1 in 4,004 | 14.9K | 14,961 | +0.41% |
| Gold Statue | 1 in 170,768 | 206.7K | 206,965 | +0.13% |
| Piano | 1 in 188,500 | 221.5K | 221,784 | +0.13% |
| Dino Skull | 1 in 466,062 | 417.4K | 417,949 | +0.13% |
| Crab | 1 in 6,088 | 20K | 20,061 | +0.30% |
| Gold Bars (Great) | 1 in 1,144,000 | 782.7K | 783,629 | +0.12% |
| Old Tank | 1 in 2,441,496 | 1.3M | 1,332,207 | +2.48% |
| Frozen Mammoth | 1 in 8,645,000 | 3.2M | 3,228,108 | +0.88% |
| Dino Skull (second find) | 1 in 658,221 | 531.6K | 532,197 | +0.11% |
| Nuke (OK) | 1 in 12,575,971 | 4.1M | 4,196,542 | +2.35% |
| Emerald | 1 in 2,794,235 | 1.4M | 1,464,187 | +4.58% |
| Angel Duck | 1 in 165,000 | 201.8K | 202,046 | +0.12% |
| Gold Bars (OK) | 1 in 792,084 | 605.1K | 605,832 | +0.12% |
Treating the displayed, rounded values as exact, all 22 nominal errors land within ±5%. Emerald is the widest nominal miss at +4.58%. Because values such as 1.3M and 1.4M are already rounded in the UI, their unseen full-precision errors could differ. Even with that limit, this remains strong out-of-sample evidence for a rule fitted from a different creator's footage; the evidence comes from video cards rather than a datamined constant.
August 10 trophy list: 1-in-100,000+ cards
These are the high-rarity item cards shown in the Bax footage. Timestamps point to the card in that recording; repeated item types are separate rolls.
| Item | Condition | Rarity | Value | Timestamp |
|---|---|---|---|---|
| Angel Duck | OK | 1 in 165,000 | 201.8K | 16:51 |
| Gold Statue | Perfect | 1 in 170,768 | 206.7K | 10:39 |
| Piano | Good | 1 in 188,500 | 221.5K | 11:58 |
| Dino Skull | OK | 1 in 466,062 | 417.4K | 13:44 |
| Dino Skull | OK | 1 in 658,221 | 531.6K | 16:22 |
| Gold Bars | OK | 1 in 792,084 | 605.1K | 17:05 |
| Gold Bars | Great | 1 in 1,144,000 | 782.7K | 13:59 |
| Old Tank | Good | 1 in 2,441,496 | 1.3M | 16:07 |
| Emerald | Good | 1 in 2,794,235 | 1.4M | 16:45 |
| Frozen Mammoth | Good | 1 in 8,645,000 | 3.2M | 16:17 |
| Nuke | OK | 1 in 12,575,971 | 4.1M | 16:33 |
Polishing is a rarity lever
Polishing did more than change the condition label in the two before-and-after examples. It multiplied the displayed effective rarity, and the same value formula still held after the change.
| Item | Before | After polishing | Rarity multiplier | Mint value vs formula |
|---|---|---|---|---|
| Gold Bars | Great · 1 in 1,144,000 · 782.7K (13:59) | Mint · 1 in 4,940,000 · 2.1M (14:41) | 4.32× | 2.1M actual · 2.18M predicted (+3.90%) |
| Nuke | OK · 1 in 12,575,971 · 4.1M (16:33) | Mint · 1 in 119,471,721 · 20.2M (17:40) | 9.50× | 20.2M actual · 20.29M predicted (+0.45%) |
Two examples are too small a sample to establish one fixed condition
multiplier: their jumps differ. They do show condition acting as a rarity
lever in those two reads. Once the polished 1 in N number is used,
45 × N^0.7 predicts each displayed value within 5%.
Frozen Shores field notes
These observations come from S3 only and are presented as field notes from that recording, read via Gemini video analysis 2026-08-17:
- The player farmed Frozen Shores with Eclipse gear equipped, including an Eclipse Detector with Luck x10500.
- Finds surfaced encased in a blue ice-crystal block that the player broke open, visible at 01:05, 02:43 and 03:22.
- Pink/purple beams appeared most often; yellow, orange and blue appeared occasionally.
Earlier light-beam color table
The Bax recording showed the following visual tiers. This is a single-source visual read; the source notes did not preserve timestamps for these beam shots.
| Beam color | Tier shown |
|---|---|
| White / light yellow | Common |
| Cyan / blue | Rare |
| Pink / magenta | Epic |
| Bright gold / yellow | Legendary |
| Multicolor / rainbow | Secret / Mythic |
Use the table as a field cue rather than a complete official legend. The labels were not independently confirmed in a second recording.
Stack luck before you hunt
Update 2 adds two higher layers to the earlier footage:
- Frozen Shores: 8× Luck. Its arrival banner was confirmed in three independent recordings, S1, S2 and S4.
- Collection Luck Milestones. S3 showed “Find 26/30 Items for +20% Luck”; S2 showed “Find 53/60 Items for +50% Luck”. Each milestone is a single-recording read, and the thresholds and rewards scale as the collection grows. S3 also showed a “+10% Luck” collection-boost popup, currently a single-source read.
The August 10 recording also showed these layers:
- A 2× Luck gamepass active in the UI at 06:31.
- Double Server Luck purchased in the shop at 06:49.
The cited recordings did not show an exact final stacking formula, so this guide leaves the sources separate instead of turning them into an unsupported total. The islands guide covers the passive ladder, and the gear guide records the Eclipse stats and prices.
Related
- Item values — formula derivation and sell versus museum economics
- Islands — the four-island luck ladder
- Gear upgrades — Eclipse stats and gear price evidence
- Codes — the current code list