Lineage 2: Revolution · Gran Kain research
What 652 million Adena
taught us about Gran Kain's Altar.
A rare reward is only half the decision. The chance of landing it matters too.
- Real draws
- 2,172
- Adena spent
- 651.6M
- Research period
- 26–30 July 2026
Our original altar-run research, with a separate offline replay of the same wheels. The replay was a simulation, not a live test of a new strategy.
01 / Read the wheel
Nine slots. Very different chances.
Slot 1 is at the top; the numbers continue clockwise. These are the in-game displayed odds recorded in July 2026. Check the current game display before using them today.
| Slot | Chance | Relative chance |
|---|---|---|
| 1 · top | 0.3% | 0.3% |
| 2 | 0.3% | 0.3% |
| 3 | 5% | 5% |
| 4 | 12.5% | 12.5% |
| 5 | 25% | 25% |
| 6 | 36.6% | 36.6% |
| 7 | 15% | 15% |
| 8 | 5% | 5% |
| 9 | 0.3% | 0.3% |
122× the chanceSlot 6's 36.6% is 122 times a 0.3% slot's chance. That compares probabilities, not reward value. It does not mean you should always choose slot 6.
of actual picks went to slots 1, 2 or 9
Each offered just a 0.3% chance.02 / The original strategy
Rarity pulled us toward the longest odds.
The original July policy strongly favored reward grade and effectively ignored slot probability. Almost half its selection decisions pursued rewards on the three lowest-chance positions. That 48% describes choices made, not rewards won.
A high-grade relic can look irresistible while offering very little chance of acquisition. Grade, enchant and your character's needs still matter, but they need to be weighed against the displayed chance.
This is a historical finding. The Peraphite scoring referenced below includes probability as one additive factor; it is distinct from both this original policy and the experimental replay.
03 / Same wheels, different choices
What the offline replay suggested.
We replayed the same 2,172 recorded wheels, comparing the historical selections with a stronger probability-weighted alternative. For each policy, adding the selected slots' displayed probabilities gives the expected number of winning selections.
Approximately 2.4× in this offline simulation. These are expected values, not live wins or acquired items. This is not a measured improvement from the current Peraphite algorithm.
| Grade | Original | Weighted |
|---|---|---|
| LR | 3.4 | 32.2 |
| UR | 3.4 | 66.3 |
| SR | 29.1 | 52.9 |
| S | 116.2 | 273.0 |
| R | 57.7 | 80.1 |
| Total | 209.7 | 504.4 |
Both columns are simulated expectations. Independently rounded grade values can sum to 0.1 more than the reported total.
The replay used an illustrative grade ranking (LR = 5 through R = 1) and ignored enchant. Each listed grade improved in this dataset, but the comparison does not establish the best strategy for every player's priorities or future wheels.
04 / Put the lesson to use
Choose with a goal and a limit.
Decide what is useful first.
Choose the categories, grades and relic enchant levels you actually want. A high chance at an unwanted item is still an unwanted choice.
Scan slots 4–7, then compare.
They offer stronger displayed chances in the July table. Check the rewards there before chasing rarity elsewhere; a low-chance slot can still be worth considering if its reward matters much more to you.
Think in chance × personal value.
As an illustration, a reward you value at 10 on a 0.3% slot gives 0.003 × 10 = 0.03. A reward valued at 1 on a 25% slot gives 0.25 × 1 = 0.25. These are subjective comparison units, not Adena returns or Peraphite's scoring formula.
Count refresh draws as spending.
If every reward is excluded by your rules, Peraphite can intentionally make a paid Draw to reach a new wheel. It costs 300,000 Adena from the same budget. It is not a free reshuffle and does not improve the next wheel's odds.
05 / Budget before you draw
Give every session a ceiling.
At 300,000 Adena per draw, divide your budget by 300,000 and round down. These examples show the maximum complete draws and spend, not promised results.
Keep Adena for your other goals and stop at the limit you set. A missed reward is not a reason to raise the budget. Paid refresh draws count too.
| Budget | Max draws | Max spend |
|---|---|---|
| 10M | 33 | 9.9M |
| 30M | 100 | 30.0M |
| 50M | 166 | 49.8M |
| 100M | 333 | 99.9M |
06 / How Peraphite helps
Your rules, applied consistently.
Peraphite's Premium altar assistant handles repeated selections within your configured Adena budget, with separate controls for what you want to prioritize.
Reward preferences
Set Relic and Old Relic grades independently, choose Relic enchant levels, and give Relics, Old Relics, Maphr and Crafting Materials their own priorities. Old Relics have no enchant filter.
Visible reasoning
“Why This Slot” explains the selected reward's scoring factors: category, grade, enchant, recognition confidence and displayed slot chance, together with your priorities.
Bounded spending
“Adena Budget” shows maximum draws and spend. An excluded-wheel refresh uses that same limit. Once a Draw has been committed, pressing Stop cannot undo its cost.
Research notes
What this dataset can tell us.
- Historical sample
- The original 26–30 July 2026 report covers 2,172 real altar draws at 300,000 Adena each: exactly 651.6 million Adena and approximately 19,548 wheel reward positions. These were recorded assistant runs, not a randomized live trial of two strategies. The policy stayed largely unchanged, apart from 55 draws with a wider enchant set.
- Measurement uncertainty
- The telemetry recorded 7.43% confirmed HIT outcomes. UNKNOWN animation results occurred more often on higher-chance slots. Treating them as probable wins gives about 216 out of 2,172 draws (roughly 10%), against about 211 expected in the report's outcome analysis. That is an inference, not independent verification of every UNKNOWN or a current global altar win rate.
- Replay limits
- The ~209.7 and ~504.4 comparison is a separate offline expectation calculation on recorded wheels, not observed reward totals. It assumes the July displayed odds, uses an illustrative grade scale and ignores enchant. It cannot show what a changed live sequence would have produced.
- Catalogue and changing conditions
- The first 250 draws contained 107 distinct reward keys; no new keys appeared in the remaining 1,922. This describes this dataset, not the game's complete or future reward catalogue. Displayed odds, rewards and game conditions may change.
Based on Peraphite's preserved July 30, 2026 research report, with product behavior checked separately. Peraphite is an independent project; see the legal disclaimer.
Keep the choices yours
Explore the assistant behind the research.
Peraphite is in Private Beta for invited and approved testers.