Volcano Treasure: 31-Day Coin Output, Payer Mix and Bonus-State Audit

Executive Summary

• Free players are the participation majority. 14,714Source for 14,714Source: OH live backend events: france_holeTable: france_hole of 17,026Source for 17,026Source: OH live backend events: france_holeTable: france_hole starters (86.4%Source for 86.4%Source: OH live backend events: france_holeTable: france_hole) had no successful purchase recorded by the window end; they generated 70.8%Source for 70.8%Source: OH live backend events: france_holeTable: france_hole of starts and 62.7%Source for 62.7%Source: OH live backend events: france_holeTable: france_hole of Volcano Treasure coins.
• High-frequency and high-value behaviour is not payer-only. Free players were 64.7%Source for 64.7%Source: OH live backend events: france_holeTable: france_hole of players with at least three starts in a rolling 60Source for 60Source: OH live backend events: france_holeTable: france_hole-minute window, and 55.8%Source for 55.8%Source: OH live backend events: france_holeTable: france_hole of players who had at least three starts and at least 3,000Source for 3,000Source: OH live backend events: france_holeTable: france_hole coins on the same day.
• Payouts above 2Source for 2Source: OH live backend events: france_holeTable: france_hole× are a real server-state defect, not a logging mismatch. All 52,038Source for 52,038Source: OH live backend events: france_holeTable: france_hole paired settlement/grant records are reproduced exactly by 5Source for 5Source: OH live backend events: france_holeTable: france_hole-percentage-point bonus tiers. 1,206Source for 1,206Source: OH live backend events: france_holeTable: france_hole runs imply a current bonus above 100%Source for 100%Source: OH live backend events: france_holeTable: france_hole; the maximum is 405%Source for 405%Source: OH live backend events: france_holeTable: france_hole, or about 5.05Source for 5.05Source: OH live backend events: france_holeTable: france_hole×, creating 551,295Source for 551,295Source: OH live backend events: france_holeTable: france_hole coins above the intended cap.
• The root cause is confirmed. The code caps historical maximum bonus add_max at 100%Source for 100%Source: OH live backend events: france_holeTable: france_hole, but does not cap the live settlement state add_cur. The daily reset uses a strict >24h comparison, so consecutive calendar days exactly 24Source for 24Source: OH live backend events: france_holeTable: france_hole hours apart do not reset and the state keeps increasing by 5%Source for 5%Source: OH live backend events: france_holeTable: france_hole.
• The aggregate inflation impact is moderate, but the correctness and fairness risk is urgent. The defect added 1.56%Source for 1.56%Source: OH live backend events: france_holeTable: france_hole of total Volcano Treasure output above the intended cap. The bonus state must be fixed immediately.

Free players drive most participation and coin output

Using successful raw event_id=205 purchases before the window end, deduplicated by order_id, 2,312Source for 2,312Source: OH live backend events: france_holeTable: france_hole starters (13.6%Source for 13.6%Source: OH live backend events: france_holeTable: france_hole) are payers and 14,714Source for 14,714Source: OH live backend events: france_holeTable: france_hole (86.4%Source for 86.4%Source: OH live backend events: france_holeTable: france_hole) have no recorded order. Payers contributed 29.2%Source for 29.2%Source: OH live backend events: france_holeTable: france_hole of starts and 37.3%Source for 37.3%Source: OH live backend events: france_holeTable: france_hole of coins; free/no-order players contributed 70.8%Source for 70.8%Source: OH live backend events: france_holeTable: france_hole of starts and 62.7%Source for 62.7%Source: OH live backend events: france_holeTable: france_hole of coins. The activity is therefore not payer-led. Purchase history is queryable from May 25, 2026; a run-level cross-check against recharge_num found no missing payer classification across the 52,038Source for 52,038Source: OH live backend events: france_holeTable: france_hole paired runs.

Participation declined while coin output per player increased

Across 31Source for 31Source: OH live backend events: france_holeTable: france_hole days, Volcano Treasure recorded 53,595Source for 53,595Source: OH live backend events: france_holeTable: france_hole starts from 17,026Source for 17,026Source: OH live backend events: france_holeTable: france_hole players. 77.3%Source for 77.3%Source: OH live backend events: france_holeTable: france_hole of player-days had one start, P90 was two starts and the maximum was 21Source for 21Source: OH live backend events: france_holeTable: france_hole. From the first seven days to the last seven, average daily starts fell from 1,864 to 1,701Source for 1,701Source: OH live backend events: france_holeTable: france_hole (-8.7%Source for -8.7%Source: OH live backend events: france_holeTable: france_hole) and average daily players fell from 1,376 to 1,190Source for 1,190Source: OH live backend events: france_holeTable: france_hole (-13.5%Source for -13.5%Source: OH live backend events: france_holeTable: france_hole). However, average daily Volcano Treasure output rose from 1,113,991 to 1,211,105Source for 1,211,105Source: OH live backend events: france_holeTable: france_hole (+8.7%Source for +8.7%Source: OH live backend events: france_holeTable: france_hole), and coins per recipient player-day rose from 844 to 1,046Source for 1,046Source: OH live backend events: france_holeTable: france_hole (+23.9%Source for +23.9%Source: OH live backend events: france_holeTable: france_hole).

Free players remain the majority of short-window burst users

There were 927Source for 927Source: OH live backend events: france_holeTable: france_hole, 254Source for 254Source: OH live backend events: france_holeTable: france_hole and 16Source for 16Source: OH live backend events: france_holeTable: france_hole players with at least 3Source for 3Source: OH live backend events: france_holeTable: france_hole, 5Source for 5Source: OH live backend events: france_holeTable: france_hole and 10Source for 10Source: OH live backend events: france_holeTable: france_hole starts respectively in a rolling 60Source for 60Source: OH live backend events: france_holeTable: france_hole-minute window. Free/no-order players were 64.7%Source for 64.7%Source: OH live backend events: france_holeTable: france_hole, 64.2%Source for 64.2%Source: OH live backend events: france_holeTable: france_hole and 50.0%Source for 50.0%Source: OH live backend events: france_holeTable: france_hole of those groups. The observed maximum was 16Source for 16Source: OH live backend events: france_holeTable: france_hole starts in 60 minutes. Of 89Source for 89Source: OH live backend events: france_holeTable: france_hole players with at least 10Source for 10Source: OH live backend events: france_holeTable: france_hole starts in one Beijing calendar day, 50Source for 50Source: OH live backend events: france_holeTable: france_hole (56.2%Source for 56.2%Source: OH live backend events: france_holeTable: france_hole) were free; the 21Source for 21Source: OH live backend events: france_holeTable: france_hole-start daily maximum was also a free player. Burst participation therefore requires independent controls on ticket/revive sources and rapid repeated starts.

The core per-run distribution is moderate, but settlement bonuses widen the tail

There were 52,045Source for 52,045Source: OH live backend events: france_holeTable: france_hole actual coin grants totalling 35,382,249Source for 35,382,249Source: OH live backend events: france_holeTable: france_hole coins; 70%Source for 70%Source: OH live backend events: france_holeTable: france_hole of completed runs paid fewer than 750Source for 750Source: OH live backend events: france_holeTable: france_hole coins. In paired runs, raw collection was 30,543,240Source for 30,543,240Source: OH live backend events: france_holeTable: france_hole and actual grants were 35,377,815Source for 35,377,815Source: OH live backend events: france_holeTable: france_hole, an uplift of 15.8%Source for 15.8%Source: OH live backend events: france_holeTable: france_hole. The uplift rose from 9.6%Source for 9.6%Source: OH live backend events: france_holeTable: france_hole in the first seven days to 21.6%Source for 21.6%Source: OH live backend events: france_holeTable: france_hole in the last seven. All 52,038Source for 52,038Source: OH live backend events: france_holeTable: france_hole paired records are reproduced exactly by the server formula floor(raw × (1 + add_cur/100)), with add_cur moving in 5%Source for 5%Source: OH live backend events: france_holeTable: france_hole tiers, so this is now a code-and-live-behaviour diagnosis rather than a correlation-only observation.

Above-2Source for -2Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua× grants are real: the formula matches every paired run

All 52,038Source for 52,038Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua settlement-to-grant pairs are reproduced exactly by floor(raw × (1 + add_cur/100)), and every inferred add_cur falls on a 5%Source for 5%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua tier. 1,206Source for 1,206Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua runs (2.32%Source for 2.32%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua) imply a current bonus above 100%Source for 100%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua, affecting 94Source for 94Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua players. The maximum inferred bonus is 405%Source for 405%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua, or about 5.05Source for 5.05Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua×. Relative to the intended 100%Source for 100%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua bonus cap, these runs generated 551,295Source for 551,295Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua excess coins, equal to 1.56%Source for 1.56%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua of total Volcano Treasure output. There is no exact 1,000 → 3,400 row in this window; the closest is a free player at 974 → 3,360, implying a 245%Source for 245%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua bonus. The conclusion does not depend on that example: many 3Source for 3Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua×-plus records follow the same mechanism.

Root cause: the live bonus is uncapped and consecutive days miss the reset boundary

The server defines ADD_MAX=100 and ADD_ONCE=5, but AddBonus increments the live settlement value add_cur by five and writes it back without min(...,100). Only historical maximum add_max is capped. Daily refresh resets add_cur only when cur_start - add_start > 24*3600. Consecutive calendar days are exactly 24Source for 24Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua hours apart, so the state is not reset and the next completed run adds another 5%Source for 5%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua. Reward settlement then uses this uncapped add_cur; AddBonus executes after the current payout, producing the chain: consecutive-day participation → no reset → another +5%Source for +5%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua → a larger multiplier on the next run. The online maximum of 405%Source for 405%Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua equals 81Source for 81Source: OH Challenge server logic, cached origin/developTables: ChallengeDefine.lua, Challenge.lua, Challenge_GS.lua, ChallengeCmd_C2S.lua five-point tiers, and every paired run follows the formula. Release engineering should still confirm the deployed image commit/hash against the inspected source snapshot.

Very high single-run payouts are concentrated, but a simple hard cap will not solve the economy

Only 129Source for 129Source: Deterministic aggregated outputsTables: daily_volcano_metrics_bjt.csv, threshold_shares.csv, daily_grant_adjustment_bjt.csv, single_run_cap_sensitivity.csv runs exceeded 3,000Source for 3,000Source: Deterministic aggregated outputsTables: daily_volcano_metrics_bjt.csv, threshold_shares.csv, daily_grant_adjustment_bjt.csv, single_run_cap_sensitivity.csv coins. The largest player generated 31.8%Source for 31.8%Source: Deterministic aggregated outputsTables: daily_volcano_metrics_bjt.csv, threshold_shares.csv, daily_grant_adjustment_bjt.csv, single_run_cap_sensitivity.csv of those runs and the top three generated 59.7%Source for 59.7%Source: Deterministic aggregated outputsTables: daily_volcano_metrics_bjt.csv, threshold_shares.csv, daily_grant_adjustment_bjt.csv, single_run_cap_sensitivity.csv. However, a hypothetical 3,000Source for 3,000Source: Deterministic aggregated outputsTables: daily_volcano_metrics_bjt.csv, threshold_shares.csv, daily_grant_adjustment_bjt.csv, single_run_cap_sensitivity.csv-coin hard cap removes only 56,293Source for 56,293Source: Deterministic aggregated outputsTables: daily_volcano_metrics_bjt.csv, threshold_shares.csv, daily_grant_adjustment_bjt.csv, single_run_cap_sensitivity.csv coins, or 0.16%Source for 0.16%Source: Deterministic aggregated outputsTables: daily_volcano_metrics_bjt.csv, threshold_shares.csv, daily_grant_adjustment_bjt.csv, single_run_cap_sensitivity.csv of Volcano Treasure output, with almost no movement in the global gain/spend ratio. An absolute cap is useful as an anomaly guardrail, but it is not a substitute for fixing the bonus state and monitoring repeated high-value player-days.

Extreme payouts are not payer-only; free players are prominent at the very top

The 129Source for 129Source: OH live backend events: france_holeTable: france_hole runs above 3,000Source for 3,000Source: OH live backend events: france_holeTable: france_hole came from 17Source for 17Source: OH live backend events: france_holeTable: france_hole players. Only three of those players were free, but they generated 53Source for 53Source: OH live backend events: france_holeTable: france_hole runs (41.1%Source for 41.1%Source: OH live backend events: france_holeTable: france_hole) and 41.6%Source for 41.6%Source: OH live backend events: france_holeTable: france_hole of the high-value coins, showing that the tail is driven by a small number of repeat users. Five of the six runs above 4,000Source for 4,000Source: OH live backend events: france_holeTable: france_hole came from free players; the only run above 4,500Source for 4,500Source: OH live backend events: france_holeTable: france_hole, at 4,994Source for 4,994Source: OH live backend events: france_holeTable: france_hole coins, was also a free player. At the more stable player-day grain, 361Source for 361Source: OH live backend events: france_holeTable: france_hole of 664Source for 664Source: OH live backend events: france_holeTable: france_hole players (54.4%Source for 54.4%Source: OH live backend events: france_holeTable: france_hole) above 3,000Source for 3,000Source: OH live backend events: france_holeTable: france_hole daily coins were free; at above 5,000Source for 5,000Source: OH live backend events: france_holeTable: france_hole, 112Source for 112Source: OH live backend events: france_holeTable: france_hole of 249Source for 249Source: OH live backend events: france_holeTable: france_hole (45.0%Source for 45.0%Source: OH live backend events: france_holeTable: france_hole) were free.

Free players are roughly half or more of the high-frequency, high-value intersection

For player-days with at least three starts and at least 3,000Source for 3,000Source: OH live backend events: france_holeTable: france_hole coins, there were 634Source for 634Source: OH live backend events: france_holeTable: france_hole players; 55.8%Source for 55.8%Source: OH live backend events: france_holeTable: france_hole were free and they generated 43.7%Source for 43.7%Source: OH live backend events: france_holeTable: france_hole of the coins. In the more extreme group with at least 10Source for 10Source: OH live backend events: france_holeTable: france_hole starts and at least 5,000Source for 5,000Source: OH live backend events: france_holeTable: france_hole coins in one day, 47Source for 47Source: OH live backend events: france_holeTable: france_hole of 86Source for 86Source: OH live backend events: france_holeTable: france_hole players (54.7%Source for 54.7%Source: OH live backend events: france_holeTable: france_hole) were free and generated 49.2%Source for 49.2%Source: OH live backend events: france_holeTable: france_hole of the coins. High-frequency, high-value behaviour therefore includes a material free-player majority, not a marginal edge case.

Head players can earn 10kSource for 10kSource: OH live backend events: france_holeTable: france_hole–24kSource for 24kSource: OH live backend events: france_holeTable: france_hole coins per day, but do not dominate global output

Player-day coin output was 662Source for 662Source: OH live backend events: france_holeTable: france_hole at P50, 1,866Source for 1,866Source: OH live backend events: france_holeTable: france_hole at P90, 5,157Source for 5,157Source: OH live backend events: france_holeTable: france_hole at P99 and 24,443Source for 24,443Source: OH live backend events: france_holeTable: france_hole at the maximum. 388Source for 388Source: OH live backend events: france_holeTable: france_hole players exceeded 4,000Source for 4,000Source: OH live backend events: france_holeTable: france_hole coins on at least one day; 53Source for 53Source: OH live backend events: france_holeTable: france_hole did so on at least three days, and one player did so on 18Source for 18Source: OH live backend events: france_holeTable: france_hole days. At the same time, the daily top player averaged only 1.20%Source for 1.20%Source: OH live backend events: france_holeTable: france_hole of Volcano Treasure output (maximum 1.97%Source for 1.97%Source: OH live backend events: france_holeTable: france_hole), and the daily top 10Source for 10Source: OH live backend events: france_holeTable: france_hole averaged 7.18%Source for 7.18%Source: OH live backend events: france_holeTable: france_hole (maximum 9.38%Source for 9.38%Source: OH live backend events: france_holeTable: france_hole). The correct control point is persistent high-value player-days and bonus state, not punitive action based on daily rank alone.

Monthly concentration is low, but a free player can still become the period champion

The monthly top 10Source for 10Source: OH live backend events: france_holeTable: france_hole generated 3.50%Source for 3.50%Source: OH live backend events: france_holeTable: france_hole of Volcano Treasure coins. Two were free players and contributed 25.5%Source for 25.5%Source: OH live backend events: france_holeTable: france_hole of top-10Source for -10Source: OH live backend events: france_holeTable: france_hole coins. The last-14Source for -14Source: OH live backend events: france_holeTable: france_hole-day champion, P-5477CD33, was also a free player, earning 116,684Source for 116,684Source: OH live backend events: france_holeTable: france_hole coins across 34Source for 34Source: OH live backend events: france_holeTable: france_hole runs, or 0.71%Source for 0.71%Source: OH live backend events: france_holeTable: france_hole of the period's Volcano Treasure output. Global concentration is not high, but individuals can still receive unusually large gains through repeated participation and the uncapped state defect.

Volcano Treasure is not the largest source, but it is material to net coin creation

Global coin gains were 554,761,934Source for 554,761,934Source: OH live backend events: france_holeTable: france_hole and spends were 431,958,800Source for 431,958,800Source: OH live backend events: france_holeTable: france_hole, a gain/spend ratio of 1.284Source for 1.284Source: OH live backend events: france_holeTable: france_hole. Volcano Treasure granted 35,382,249Source for 35,382,249Source: OH live backend events: france_holeTable: france_hole coins: 6.38%Source for 6.38%Source: OH live backend events: france_holeTable: france_hole of all gains and 28.81%Source for 28.81%Source: OH live backend events: france_holeTable: france_hole of the 122,803,134Source for 122,803,134Source: OH live backend events: france_holeTable: france_hole net increase. Removing Volcano Treasure in a static accounting view leaves a gain/spend ratio of 1.202Source for 1.202Source: OH live backend events: france_holeTable: france_hole. The activity is not the sole source of inflation, but it is large enough to warrant close monitoring after the immediate state fix.

Hot-load the bonus-state fix

This is not a recommendation. It is an immediate action item.

1. P0 — fix both boundaries immediately. Apply add_cur = min(add_cur + 5, 100). Replace the cross-day reset comparison with explicit calendar-day sequencing that cannot miss the exact 24Source for 24Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine-hour boundary. Do not fix only one side.
2. Add pre-release regression coverage. Cover consecutive participation, a one-day gap, a two-day gap, restoring historical maximum bonus, server daily-refresh offset and the 100%Source for 100%Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine boundary. Assert that every settlement multiplier is at most 2Source for 2Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine×.
3. Repair existing bad state. Clamp online and offline records with add_cur>100 to 100Source for 100Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine during the hot-load. Reconcile affected roles for one week after release so stale state cannot continue contaminating payouts.
4. Install two-level monitoring. Record raw_gold, add_cur_before_reward, actual_gold and reset_reason per run. Monitor player-days above 3,000Source for 3,000Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine / 5,000Source for 5,000Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine / 10,000Source for 10,000Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine coins and rolling-60Source for -60Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine-minute start counts of 3Source for 3Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine / 5Source for 5Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine / 10Source for 10Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine, split by payer status.
5. Whether to reduce base rewards remains undecided. The defect generated about 551kSource for 551kSource: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine coins above the intended cap, equal to 1.56%Source for 1.56%Source: Current Volcano Treasure design document and Challenge.xlsmTables: 【251107】火山寻宝.docx, Challenge.xlsm:ChallengeDefine of Volcano Treasure output. Complete the bonus-state repair first and then re-observe actual output.