Evidence and scope
Round 1880047227552134678 uses nonce 15 and reproduces 76.28. The revealed server seed hashes to the recorded commitment. The screenshot shows Under 50 with a displayed 50.00% chance; the result loses.
The wager is dated October 25, 2026 at 18:25:54.547 UTC. The first captured response with its hash completes at 18:26:09.987 UTC, and the first response with its revealed seed completes at 18:27:07.156 UTC. That first capture alone establishes a matching hash before disclosure, after the wager.
Round 1880048288356627997 provides an additional commitment sequence. A settings screenshot at 18:42:28 UTC shows the current client seed, next nonce 1 and a 48-character server hash prefix. Nonce 2 is timestamped 18:42:46.209 UTC and reproduces 35.84, winning on Under 50 with a reported gross multiplier of 1.96x. The full hash is first captured at 18:43:13.269 UTC; the revealed seed follows at 18:43:41.931 UTC. SHA-256 matches that full hash and its saved prefix.
This combined screenshot and network capture evidence supports Verified for ClassicDice98 replay. The screenshot clips the final 16 hash characters; the pre-bet record is a prefix, not a full network hash record. Its timestamp is derived from the user-supplied screenshot filename and converted to UTC. It is not a server-issued timestamp. The original nonce 15 vector retains its post-bet provenance.
Dice98: six additional rounds
The latest network capture identifies a separate game: Dice98. Nonces 2, 3, 4, 6, 7 and 9 replay to 93.62, 11.97, 99.07, 9.25, 37.54 and 13.49. Its settings screenshot at 19:09:54 UTC shows next nonce 0, a 48-character hash prefix and the first 21 client-seed characters. The first captured full hash completes at 19:11:03.939 UTC and disclosure at 19:12:57.224 UTC. The revealed seed matches that full hash and saved prefix. The clipped client-seed suffix is first recorded in the network capture, not in the pre-bet screenshot.
Dice98 uses 10,000 events and results from 0.00 through 99.99. The captured API stores inclusive winning tick ranges: 0–4999 for the UI’s lower half, 5000–9999 for the upper half, 2500–7499 for a central interval, and 884–7499 for the last custom interval. Winning examples report 1.96x for 5,000 ticks and 1.4812x for 6,616 ticks. These are captured observations, not a certified formula for all settings.
Separate algorithm scales
Both captured variants use API version 0. ClassicDice98 uses 10,001 events and Dice98 uses 10,000 events. HMAC-SHA256 uses the UTF-8 server seed as key and clientSeed:nonce as message. The first 15 hexadecimal digits select an offset modulo 47. Read 14 hexadecimal digits at that offset, shift right by three bits, divide by 2⁵³, and multiply by the selected variant’s event count. ClassicDice98 floors to API ticks. Dice98 uses the published JavaScript parseInt conversion. Divide the ticks by 100 for the displayed result.
ClassicDice98 results run from 0.00 through 100.00; Dice98 results run from 0.00 through 99.99. The legacy first-four-byte Classic Dice algorithm is outside this adapter. The current published helper returns a fractional value; the captured API result confirms truncation rather than rounding to nearest.
What is still outside this check?
The first losing round reports zero odds. The second round confirms an announced 1.96x at Under 50. The additional Dice98 capture records winning multipliers at two interval sizes. These observations do not establish a complete payout formula, exact RTP, all target bounds or currency settlement rules. No BC.Game calculator or simulator profile is inferred from this example. BCH.games tools use their own captured rules.
Sources: BC.Game’s Dice verifier, Classic Dice verifier and sliding-window helper, checked October 25, 2026, plus the sanitized disclosed round. Raw network capture files and account credentials are excluded.
