FAQ

Specific tracker questions

Answers about missing loot, ownership rules, historical data and the RNG model.
GEAR & COMBAT

Team gear and DPS

Equipment availability, personal stats, combat assumptions and hypothetical setups.

What do green, orange and red equipment mean?

Green is available to this player after reserving other accounts’ equipment, including a complete potential craft. Orange means acquired but allocated, consumed, restricted by levels or personal to another account. Red has neither available stock nor a complete known crafting route. A Collection Log records lifetime acquisitions, so correct bank quantities under Adjust available stock.

Every slot is searchable. Orange and red choices remain selectable for comparison. If the recalculated team cannot supply the item, its slot and DPS are labelled hypothetical; the supply ledger never creates a missing copy. Manual choices are saved by encounter, raid room, player and phase.

How are the default equipment and DPS chosen?

Defaults compare available equipment using each player’s saved levels and the selected target. The search tests leading weapons, iterates armour choices and includes supported set combinations: all three Void styles and regular/elite mixes, low-HP Dharok, Verac, and Ahrim/Karil with an Amulet of the Damned, alongside crystal and Moon armour. Equal DPS can hide bonuses through max-hit rounding, so damage or Strength, accuracy, prayer and defence break ties. Useful alternatives such as Inquisitor’s plateskirt’s +2 Strength can free Blood Moon tassets for another account.

This is an iterative equipment search rather than an exhaustive global team optimiser. Logged owners get first use of their equipment and crafting materials, including phase switches. Spare copies and unused items can then be borrowed in the displayed loan priority. Big Dog Aura gets first use of a spare Tumeken’s shadow after its logged owner. Personal items stay with their owners; a player may reuse their own equipment across phases.

Which combat settings and assumptions affect the result?

The OSRS Wiki combat engine handles equipment bonuses, attack speeds, weapon effects, NPC defences and scaling. The planner exposes potions, offensive prayers, ammo, spells, raid targets, party size, ToA levels, target HP and successful defence reductions. Each raid has an encounter subfilter for its bosses, rooms and path challenges. Choose an encounter, then a target phase where needed; the equipment and DPS recalculate for that selection. Puzzle-only rooms explain why conventional combat DPS is unavailable. Each result names its actual target and shows the levels and buffs used. Calculation details also show mean damage per attack and the expected attack interval: DPS is mean damage divided by that interval in seconds.

DPS assumes continuous attacking and maintained boosts, charges and repairs. Ammo supplies are assumptions. This group treats all prayer unlocks as permanent, supported by its logged prayer scrolls; each account must still meet the Prayer and Defence levels. Scrolls are not allocated as consumable equipment; movement, downtime, supply duration and every quest or crafting prerequisite are not established by the log. Missing combat levels, incompatible ammo/spells or unavailable target data produce an explicit unavailable result. Engine source and licences.

Where do I change ammo, prayers and Slayer tasks?

Use the Account selector in the DPS & combat settings panel, or Combat settings on a player’s card. Every account has independent potions, ammo, spells, HP, tasks and other buffs. Melee, ranged and magic prayers have separate selectors for that account, so mixed-style comparisons use the intended prayer for each setup. Reset this account’s combat settings affects only that account. Previously saved defaults and overrides are preserved as individual settings. Best eligible prayers are the default, using the group’s confirmed unlocks and each account’s base levels. Cards show their actual applied prayer and boosts. Combat changes immediately recalculate the equipped setups, then search available equipment again; the status identifies the optimisation in progress. Matching Slayer tasks can differ between players. Required-task bosses stay on task; only eligible targets receive task bonuses, and Salve does not stack with Slayer headgear.

Ammo preset supports all nine Seeking arrow variants. The engine uses their +20 bow accuracy and minimum accurate hit of 3, leaving misses at zero. Compatible owned arrows participate in automatic selection. Presets assume maintained consumable supplies; manually choosing an ammo slot overrides the preset for that setup. Blowpipe darts and arrows or bolts stored in a quiver use the panel. Defence reductions describe the shared target: enter landed hits or completed drains there to recalculate everyone’s DPS.

What do the item comparisons, special results and attack-time estimate mean?

An item-picker preview changes one slot for that player against the same target phase, keeping the rest of the setup fixed. It shows bonuses and normal DPS delta; ring previews can also show the reserved special weapon’s damage-only cycle. Previews do not consume stock. Selecting an item recalculates the full allocation, which can change other slots or loans.

Special results show mean damage per spec and energy recovery in the player’s main armour. The cycle estimate replaces normal attacks with one spec per natural recovery; Lightbearer doubles recovery. It excludes debuff benefits, Death Charge, adrenaline and overlapping burns. Defence drains remain separate inputs.

Expected attack time uses the engine’s damage distributions and HP-sensitive effects, accounting for overkill and variable delays. It includes the full killing-attack interval, and excludes movement, downtime, defence regeneration and special rotations. Auto ammo picks ruby or diamond bolts for the selected HP snapshot; its attack-time result holds that bolt rather than simulating swaps. Detailed effects and probability plots show what the engine actually used. Research review and source links.

Do defence-reducing weapons apply their special automatically?

No. Enter landed maul or warhammer hits, BGS damage and other completed drains under Defence reductions. They describe the state before normal attacks, not attempted specials. The engine enforces monster defence floors. A Defence-only drain can improve melee and ranged accuracy without changing ordinary magic accuracy; curses that also reduce Magic can affect magic setups.

Does a Vorkath head make an assembler available, and can the planner use blessings?

Yes. A logged Vorkath head makes Ava’s assembler a potential personal craft without separately confirming an accumulator. Ava accepts a head, 75 mithril arrows, and either an accumulator or 4,999 coins. The planner uses the coin route as a stated condition, consumes one tracked head, and checks 70 Ranged. Another account cannot use your untradeable head. Each account also owns a normal Ava’s accumulator, as confirmed for this group; this does not require the optional standard-gear assumption. Accumulators require 50 Ranged and can be removed by setting that account’s remaining stock to 0.

Owned blessings fill a free ammo slot for melee, magic, crystal bows and blowpipes. Bows and crossbows keep their required ammo unless Dizana’s quiver stores the arrows or bolts, freeing the normal slot for a blessing. Javelins, atlatl darts and bolt racks cannot use this quiver exception. Blowpipe darts are loaded internally. Blessings add Prayer bonus rather than direct DPS; Rada’s diary tier must be confirmed for its owner under Add non-log gear.

Can I add an MA2 cape or other gear outside the Collection Log?

Yes. Select Add non-log gear, choose the account, find the item and enter its remaining quantity. Searching MA2 or Mage Arena II finds the imbued god cape. The three god variants share one stock row. A personal reward becomes available only to the account you confirm; it is not granted to the whole group. Shortcuts also cover Barrows gloves, dragon defender and Ava’s assembler.

Select any equipment slot to search and swap items. Orange and red choices remain hypothetical unless physical stock is available. Confirmed stock and manual equipment choices are saved in your browser.

FAQ TOPIC 01

Collection Log

Classic and Advanced Collection Log, group ownership, value, categories, recent unlocks and recovery.

Can Hypothetical Log rolls change my real Collection Log?

No. Simulated KC, drops and saved rolls stay in a separate browser-only sandbox. They never change the real Temple-backed Collection Log, shared group totals or header achievements.

The simulator uses the same boss and raid item-rate model as RNG Index. Raid uniques are mutually exclusive, while pets, kits and tertiary drops roll separately. Ordinary supplies such as coins, runes and food are outside its Collection Log tables.

How do I find items one player owns and another is missing?

Select the players, then set the owner to Has obtained and the other player to Has not obtained. Every chosen condition must match. Either leaves that player’s ownership unrestricted.

Item search, category and ownership presets narrow the same results. A member without a usable saved Temple log is unknown, rather than counted as owning zero items.

Why can an item be exclusive in one selection and shared in another?

Ownership is calculated for the selected members. Exclusive means exactly one selected member owns the item; shared means at least two do. Changing the selection can therefore change its classification.

Duplicates instead means at least one selected owner has a logged quantity above one. Two players owning one copy each makes an item shared, without either player having a duplicate.

How is Collection Log item value calculated?

Most tradeable items use a price from the shared saved OSRS Wiki GE snapshot, multiplied by their logged quantity. Prices refresh at most once every 12 hours, so they can lag the current market. Some untradeable upgrade components use a derived residual value so their economic contribution is not automatically treated as zero.

Examples include Araxyte fang as the amulet of rancour value minus an amulet of torture, Noxious halberd pieces as one-third of the finished halberd, Soulreaper axe pieces as one-quarter of the finished axe, and DT2 vestiges as the upgraded ring value minus the base ring, 500 blood runes and 3 chromium ingots. Residual values are floored at zero; unpriced items contribute 0 gp.

Why are older Collection Log unlock dates missing?

Recent unlocks are saved from Temple after member Collection Log syncs. New item unlocks can therefore appear with their recorded Temple timing from the period in which the tracker has been saving that information.

For old drops that happened before the tracker had comparable saved baselines, exact historical unlock dates cannot be reconstructed reliably.

Does importing a Collection Log overwrite the shared data?

No. Recovery imports override the displayed Collection Log in your browser only. Importing a full saved log or a compatible member export does not replace the shared baseline for other visitors.

You can export the currently displayed snapshot or reset the browser override to return to the shared data.

FAQ TOPIC 02

Chronicle

Three questions: missing loot, reporting rules and first group unlocks.

Why isn’t my loot showing in the Chronicle?

Chronicle reads saved Temple Collection Log records, rather than a complete loot feed. Open the Collection Log in game and use its refresh function, then press Update Collection Log via Temple here so the tracker receives the newer record.

Check that the player is selected and Item unlocks or All events is active. The drop must meet the reporting rules below. Valuable copies missing from the recent feed are recovered from the full saved logs under Date unknown, with their recorded individual and group totals. Bulk-import timestamps are not treated as drop dates. A failed update keeps the previous saved log.

What does the Chronicle report?

Player unlocks: every new Collection Log item for that player, at any value, including untradeable or unpriced items.

Repeat drops: only when one item is worth strictly more than 500,000 gp. Exactly 500,000 gp does not qualify. Stack value does not qualify: 100 copies worth 10,000 gp each remain a cheap duplicate.

Unit values use the shared saved Wiki price snapshot and the tracker’s component-value rules. An unpriced duplicate cannot meet the threshold. These rules apply to all saved Chronicle item history. Stats milestones also include every gained skill level and recorded XP, total-level and KC milestones.

What does “New group Collection Log unlock” mean?

It marks the first confirmed unlock of that item across all five United Gimps accounts. A new item for one player is not automatically new for the group. Narrowing the member filter does not reset the group’s history.

The saved logs must establish who obtained it first. Missing member logs, another owner’s unknown acquisition date or tied earliest dates leave the event unlabelled. Later owners and repeat copies do not receive the group-first label; personal unlocks still appear.

FAQ TOPIC 03

RNG Index

The complete methodology behind the RNG model. Open a section below, then expand individual questions as needed.

RNG SECTION 01

Foundations

The formula, evidence, impact and the final 1–10 score.

6 questions

What is the basic RNG Index formula?

The starting theory is deliberately simple: meaningful RNG = evidence × impact × context/rank modifier. Evidence asks how unusual the observed drop outcome was. Impact asks how useful that exact item is to a mature Group Ironman. The context/rank layer then asks whether this copy actually matters in the selected group and whether the ranking needs protection against things such as duplicate copies or the look-elsewhere effect.

Evidence×Item impact×Context / rank modifier

That formula is literal for the visible Luckiest meaningful drops and Unluckiest meaningful grinds rankings. The final 1–10 headline score uses the same evidence and impact system but combines all independent evidence as a normalized signed index rather than simply multiplying every row and adding the results.

What exactly is “evidence”?

Evidence is the probability layer. For every calculable item, RNG Index reconstructs the number of eligible opportunities, the relevant drop probability and the observed Collection Log count. It then asks where that observed count sits in the probability distribution for that number of rolls.

The discrete percentile uses the midpoint of the observed outcome: the probability of seeing fewer copies plus half the probability of seeing exactly that many. The percentile is converted to a normal-deviation score, usually written as z or σ. Around 0σ is ordinary. Positive values are lucky and negative values are dry. Item-level z is capped at ±3.50σ so one extreme or slightly misspecified mechanic cannot dominate the whole account indefinitely.

Why is expected count not enough by itself?

Because “0 obtained versus 0.79 expected” does not mean you are 0.79 items unlucky in any meaningful statistical sense. With an expectation below one, getting zero is still a completely ordinary result. RNG Index therefore evaluates the full probability distribution rather than ranking items by the raw difference between actual and expected.

The Dragon claws example exposed this directly: a high-impact item with only 0.79 expected copies should not become the top dry grind merely because claws are important. Importance can change how much a real dry streak matters; it cannot manufacture evidence that the dry streak exists.

What does “impact” mean?

Impact is the practical progression value of owning the item. It is not rarity and it is not GE price. The question is: if this were a useful first copy on a mature/late-game GIM, how much account progression or durable utility would it create?

That gives account-defining weapons such as Twisted bow and Tumeken’s shadow very high base impact, major upgrades such as Rigour, Fang and Torva strong impact, useful stepping-stones moderate impact, and cosmetic or collection-only rewards almost none.

What is the context/rank modifier?

The base impact score is only the starting point. The effective weight can change because the selected group already owns copies, because a permanent personal unlock behaves differently from a shareable weapon, because components only matter when they form a usable item, because complementary gear creates limited synergy, or because the underlying rate is assumption-heavy.

The visible positive-luck ranking also applies a modest source-family breadth penalty. If an activity gives the model a huge number of possible Collection Log outcomes, the best-looking random outlier is less impressive than the same statistical deviation from a narrow set of meaningful targets.

How does the final 1–10 score differ from the item-ranking formula?

The headline score is an aggregate. After correlated entries are collapsed and source families are capped, every independent item contributes approximately z × effective impact weight. CoX, ToA and ToB can also contribute pooled raid-portfolio signals. The numerator is the sum of those signed contributions. The denominator is the square root of the sum of squared effective weights, including the raid portfolios.

index = (Σ item z·w + Σ raid contribution) / √(Σ item w² + Σ raid weight²)

That normalized index is then mapped onto a bounded 1–10 scale with a saturating tanh curve. 5.0 is neutral. Values above 5 represent favourable meaningful RNG; values below 5 represent unfavourable meaningful RNG. The score is an impact index, not a literal p-value or standard-normal test statistic.

RNG SECTION 02

Item impact

How the 885 reviewed item scores separate practical utility from rarity and price.

7 questions

Are items still scored by broad category, such as “clue reward” or “boss unique”?

No. The production impact model now contains an explicit record for every item ID currently eligible for the luck calculation: 885 items at the time of the audit. The runtime model does not normally say “all Treasure Trail rewards are 0.04” or “all PvM uniques are 0.95”. It looks up the reviewed score for the exact item ID.

Mechanically identical variants can share the same researched reasoning, but every item still has its own explicit entry. If a future rate-model item is added without an impact review, the feature falls back to a deliberately tiny 0.05 impact and writes a console warning rather than silently guessing a category weight.

How was the 885-item impact table calibrated?

The table is calibrated to a mature/late-game Group Ironman. That prevents early-game usefulness from being treated as permanent endgame importance. A piece of equipment that is fantastic at 70 combat but almost entirely obsolete by late-game receives some credit, but it does not compete with durable raid upgrades.

The audit separates first-copy utility from everything else. Duplicate value, group coverage, synergy, statistical rarity, GE value and the probability evidence are separate layers. This avoids double-counting the same idea inside the base score and again in the context modifier.

How can two Treasure Trail items have completely different impact?

Because the source category says almost nothing about practical utility. Ranger boots have a high clue-item impact because they are part of the Pegasian and later ranged boot progression chain. Holy sandals have genuine prayer-footwear progression. Spiked manacles are real strength boots, although mostly a stepping-stone. Ham joint has an actual fast-attack niche. God-book pages have modest but real completion utility.

By contrast, a 3rd age range top can be astonishingly rare and extremely valuable while producing almost no mature-account progression. A Rune platebody (h2), demon mask or Piscarilius hood is essentially cosmetic in this model. Their rarity can still appear in Raw RNG; it does not make them meaningful progression luck.

What are some concrete clue-impact examples?

Ranger boots4.5major ranged boot/treads upgrade-chain ingredient
Holy sandals1.8real prayer-boot progression ingredient
Spiked manacles1.0real early/midgame strength boots
Ham joint0.8real 3-tick melee niche
Ancient page0.7god-book component with modest late-game use
3rd age range top0.12rare collector gear with little mature-account utility
Rune platebody (h2)0.08negligible mature-account value
Demon mask / Piscarilius hood0.05cosmetic or collection-only

Does rarity increase an item’s impact score?

No. Rarity belongs in the evidence term. If a 1-in-20,000 cosmetic drops, Raw RNG should correctly recognise that the event was unusual. But the base impact score should still be near zero if owning the item does essentially nothing for the account. Mixing rarity into impact would count the same concept twice.

Does GE value increase impact?

Only slightly, and only after the explicit utility score already says the item is relevant. In the overall model, gameplay-relevant items can receive a small live-GE modifier of roughly ±6%. This can break close ties between otherwise comparable useful items, but it cannot turn a useless expensive collectible into a major progression item.

GE has its own separate analysis lens where price is allowed to matter heavily. The headline RNG Index intentionally does not equal bank value.

Why can an upgrade ingredient have substantial impact even if the dropped item itself is not powerful?

Because some drops are required gates to an endgame item. Ranger boots, Hydra leather, Nihil horn, DT2 vestiges, Bandos/Armadyl pieces and several weapon components matter because they unlock or complete something substantially more useful.

The table gives ingredients discounted credit for the progression they unlock rather than pretending the ingredient itself has the full power of the finished item. Component-completion logic then separately checks whether the selected accounts can actually form the item.

RNG SECTION 03

Lucky drops and dry grinds

Why the positive and negative rankings are deliberately asymmetric.

6 questions

Why are lucky drops and unlucky grinds ranked differently?

Because the selection problem is asymmetric. Players often deliberately grind high-value targets, so going genuinely dry on a major target is meaningful evidence. Positive luck has an additional problem: if the Collection Log contains hundreds of possible outcomes, something obscure is eventually going to look unusually lucky purely because the model had so many places to look.

That is the look-elsewhere or multiple-comparisons problem. RNG Index therefore demands more relevance from a positive “meaningful luck” result than from a negative dry result.

What is the exact evidence term in the meaningful item lists?

For both sides, the visible meaningful ranking uses |z|² as its evidence term. Squaring the deviation makes genuinely extreme RNG separate more strongly from routine variance. A 2σ result has four units of evidence while a 1σ result has one; a 3σ result has nine.

This is a ranking device, not a claim that z² is itself a probability. Raw RNG remains available when the user wants the statistically interpretable percentile/z view without utility ranking.

How is an unlucky grind ranked?

For negative z, the ranking is deliberately evidence-first. The score is approximately −|z|² × impact factor. The impact factor rises smoothly with effective item impact but is bounded to a modest range, so a coveted item can move a real dry streak upward without turning an ordinary result into a dry grind.

dry rank = −|z|² × [0.75 + 1.25·(1 − e^(−w/3))]

There is no look-elsewhere penalty on the dry side. If you deliberately put hundreds of kills into a target and fail to get it, the existence of many other unrelated Collection Log slots does not make that failure less real.

How is a lucky meaningful drop ranked?

For positive z, RNG Index uses the same squared evidence, but relevance matters more and a modest source-family breadth penalty is applied. The score is approximately |z|² × lucky impact factor × search penalty.

lucky rank = |z|² × [0.35 + 1.85·(1 − e^(−w/3))] × searchPenalty

The search penalty becomes smaller when the same source family offered many independent chances for some item to become the apparent outlier. It is intentionally much softer than a strict Bonferroni correction.

Can a useless cosmetic still be the statistically luckiest item?

Yes, in Raw RNG. If the roll really was absurd, the raw probability view should say so. But pure cosmetics, vanity items and generic low-impact collection items are filtered out of the Luckiest meaningful drops list. That list is answering a different question: where did RNG actually give the account something worthwhile?

Why not filter low-impact items out of the dry list too?

Because dryness is usually tied to intentional pursuit. If someone has enough eligible opportunities for a low-impact item to become statistically extreme, that dryness is still real evidence. Impact can keep it below more important grinds, but the model should not pretend the statistical result did not happen. Positive luck needs the stronger filter because the model can discover accidental outliers across many items without the player ever targeting them.

RNG SECTION 04

GIM context

Duplicate value, teammate coverage, synergies and component completion.

6 questions

How are duplicate copies valued?

Useful copies follow diminishing marginal utility instead of a single blunt duplicate penalty. A first major shareable weapon receives full copy utility. Second through fifth copies can still matter substantially in a five-person GIM because several members may want to use the item simultaneously. After useful group capacity is filled, additional copies fall much more sharply.

Permanent personal unlocks such as prayer scrolls follow a different curve because each player can consume one. The first five group unlocks therefore remain meaningful in a five-person group in a way that five copies of a freely shareable niche weapon may not.

How does teammate ownership affect someone who is dry?

The raw probability layer does not change: if one player is statistically dry, that remains true. The overall-impact layer asks how costly that dryness still is to the selected GIM. If another selected member already owns a shareable copy, the practical deficit is smaller, so the negative impact weight is softened.

This is deliberately not all-or-nothing. One existing Tbow does not make a second-Tbow drought irrelevant, but it does mean the group already has access to the weapon. For personal unlocks, group coverage is based on how much of the selected member capacity has already been solved.

Can lucky players compensate for dry teammates?

Yes, but only where the items actually solve the same group problem. That is one of the reasons combinations are scored directly instead of averaging solo scores. If one account is very dry on a major shareable target and another account has supplied useful copies, the selected group portfolio is in a better position than the dry account alone.

Compensation is constrained so one ordinary spoon does not erase broad dryness across unrelated targets. Source caps, item impact, copy utility and raid portfolio structure all limit that effect.

How does synergy work?

Complementary gear can receive a capped synergy bonus when the combination creates more progression value together than separately. Examples include Tbow with Rigour, Shadow with Ancestral, Masori with Tbow, enhanced crystal weapon seed with enough crystal armour, or Lightbearer with strong special-attack weapons.

Synergy is count-aware and split across both sides of the interaction. One Dex does not magically support an unlimited number of Tbows, and duplicate armour pieces do not pretend to be a complete set.

How are untradeable component sets handled?

Components that must belong to the same account are never pooled across members as if Group Storage could assemble them. Noxious halberd and Soulreaper axe completion checks are personal. The same principle applies to other component sets where the game requires the pieces on one account.

The impact table can still recognise that each component is meaningful progression, while the context layer separately checks whether the selected portfolio actually contains completed sets.

Does the model invent a historical acquisition order?

No. Old Temple imports do not reliably tell the model whether the Tbow came before the Rigour scroll, whether a teammate copy existed at the time, or which exact item solved a progression bottleneck first. RNG Index therefore uses the current selected portfolio and allocates copy utility order-neutrally. It measures the progression significance of the recorded RNG, not a fictional reconstruction of every historical decision.

RNG SECTION 05

Raid modelling

How CoX, ToA and ToB are handled as both items and reward portfolios.

7 questions

Why are CoX, ToA and ToB treated as portfolios as well as individual items?

Raid rewards create a portfolio problem. Someone can be lucky on total purple volume but uneven across specific items, or receive an ordinary number of purples with extraordinarily good quality. Looking only at individual items can therefore misdescribe the raid experience.

RNG Index adds a pooled raid signal for CoX, ToA and ToB, while still retaining a reduced item-level contribution so a specific Shadow, Tbow or Scythe remains visible.

What does a raid portfolio measure?

Each raid portfolio has three components: volume asks whether total uniques are above or below expectation; quality asks whether the observed uniques skew toward more useful rewards; coverage asks whether the account hit useful distinct rewards instead of merely stacking one repetitive reward.

The current combined portfolio signal uses weights of 0.70 volume + 0.50 quality + 0.30 coverage, normalized by the length of that weight vector. CoX and ToA portfolios have structural weight 9; ToB has weight 8, before confidence adjustment.

Does the raid portfolio double-count individual raid items?

It is designed not to. Raid items represented in a pooled portfolio retain only a residual fraction of their ordinary item-level weight: the highest-impact megarares retain the largest residual, major items a smaller one, and ordinary raid items less again. That keeps iconic individual outcomes visible without allowing the same purple evidence to vote at full strength twice.

How is Chambers of Xeric modelled?

CoX uses personal points rather than pretending every completion has an identical personal purple chance. The current representative assumptions are approximately 30,000 personal points for ordinary CoX and 62,000 for timed deathless solo CM. Historical KC is split around the 12 August 2026 reward-table rework where the saved WOM history allows that split.

Almost all of this group’s current recorded CoX KC predates that rework, so the historical unique table remains especially important for the present dataset.

What is the Big Dog Aura CoX caveat?

A large share of Big Dog Aura’s ordinary CoX KC was completed in scaled raids, so his real average personal points were probably above the generic 30,000-point baseline. The model uses a higher representative baseline for him, but the exact historical points are not recoverable for every completion.

That uncertainty is not ignored. His affected CoX evidence receives a confidence discount, and the discount applies to his source share rather than automatically reducing another selected member’s cleaner CoX evidence.

How is Tombs of Amascut modelled?

Normal and expert ToA use representative fixed raid conditions because WOM does not preserve the complete historical invocation/points profile for every run. The current normal model is approximately a 150-invocation deathless raid without skull skipping, around a 1/45 purple chance. The expert model is approximately 350 invocation, deathless with skull skipping, around a 1/16 purple chance.

Because those are assumptions rather than exact historical points, ToA evidence is confidence-damped in the headline model. The raw per-item display still shows the result under the stated assumed model rather than hiding the assumption.

How is Theatre of Blood modelled?

ToB requires an assumed personal share because the group KC alone does not reconstruct every historical team size, personal contribution and MVP situation. The model uses the supported share assumption and applies a confidence discount rather than pretending the denominator is exact.

RNG SECTION 06

Confidence and exclusions

What the model trusts, discounts or refuses to score.

10 questions

Where do RNG Index counts and denominators come from?

Temple supplies the saved Collection Log counts. Wise Old Man supplies most kill counts, raid completions, clues and other opportunity denominators. RNG Index combines the two saved snapshots; it does not hit either API simply because the page is opened.

What happens when the Collection Log and WOM snapshots are out of sync?

The page compares their saved timestamps. If Collection Log data is newer, very recent drops can temporarily look slightly luckier because the KC denominator may lag. If WOM is newer, recent KC can temporarily look slightly drier because the item count may lag. The page warns about that mismatch instead of inventing the missing interim data.

What does model confidence do?

Not every denominator is equally trustworthy. Direct accepted mechanics can use full confidence. Fixed CoX and ToA point assumptions, ToB share assumptions, scaled-CoX estimates, group-boss contribution assumptions and distribution approximations are discounted.

Confidence multiplies the evidence contribution in the numerator while the structural denominator retains its role, so uncertain evidence is genuinely pulled toward neutral rather than having the same factor cancel out of both sides of the calculation.

Why are some items in “Impossible to tell”?

Because exclusion is preferable to fake precision. If the saved data cannot reconstruct the eligible opportunities or the mechanic reliably, the item contributes nothing to the headline score. The separate panel explains the reason for each exclusion rather than silently treating the item as average.

Why are capped Collection Log counters excluded?

Some Collection Log counters stop increasing at a known ceiling even if more copies are obtained. If a selected member is exactly on such a cap, the true observed count is censored. RNG Index therefore cannot know the real number of copies and excludes that item completely rather than undercounting it.

Why are guaranteed variable-quantity drops excluded from spoon/dry luck?

A guaranteed drop with a variable stack size should not masquerade as a rare drop. Granite dust exposed this bug: the old calculation could treat an unusually high average stack as “lucky” even though the item itself was guaranteed. Guaranteed variable-stack entries such as Granite dust and Mole skin are now excluded from spoon/dry rarity evidence.

Non-guaranteed variable-quantity drops can still be modelled where both the drop event and quantity distribution are meaningful and supported.

Why are Nex uniques excluded?

Personal Nex unique chance depends on historical contribution and MVP bonuses. Much of this group’s Nex KC was done in teams or masses, and saved KC does not reconstruct each player’s historical share accurately enough. RNG Index therefore excludes Nex rather than pretending every kill was an equal personal roll.

Why are Barrows bolt racks excluded?

Bolt racks depend on Barrows reward potential and quantity rules. WOM provides chest count but not the reward potential of every historical chest. That makes the eligible quantity distribution unreconstructable from the saved data, so bolt racks are excluded.

How are correlated Collection Log entries prevented from creating fake evidence?

Where several imported entries represent the same underlying roll or dependent component structure, the headline index collapses them into one dependency group rather than letting each counter vote independently. They can still be shown individually in item lists because that is useful for inspection, but the aggregate score does not pretend they are independent experiments.

How are huge drop tables stopped from dominating the account score?

After dependent entries are collapsed, weights are also capped by source family. Clue tables have the lowest cap, Barrows has its own cap, raids get more room, and ordinary sources have a general cap. The cap is applied to the combined L2 weight of that source family.

That means Jagex cannot make one activity mathematically more important merely by splitting its rewards into 100 Collection Log slots.

RNG SECTION 07

Other analysis lenses

Raw RNG, GE-weighted RNG, combinations and score interpretation.

5 questions

What is the difference between Overall impact, Raw drop rate and GE impact?

Overall impact is the flagship model: probability evidence weighted by researched item utility, group context, confidence and raid portfolios. Raw drop rate ignores usefulness and price and shows the statistical spoon/dry signal. GE impact asks a different question again: how favourable was the RNG when live market value is allowed to drive the weighting?

Keeping those lenses separate is deliberate. An absurd clue cosmetic can be genuinely rare in Raw RNG, expensive in GE RNG and still nearly irrelevant to account progression in Overall impact.

Why can the Raw RNG result disagree strongly with the 1–10 score?

Because they answer different questions. Raw RNG asks whether the observed item counts were statistically unusual. RNG Index asks whether the unusual outcomes were useful enough to matter to this GIM. A portfolio containing many bizarre cosmetic spoons but dry major upgrades can therefore look lucky in raw probability and mediocre or unlucky in meaningful progression.

What does the GE-weighted lens actually calculate?

Priced entries receive a log-scaled financial weight based on live OSRS Wiki market data, with model confidence still applied. The page also shows an approximate observed-minus-expected GP delta for entries whose expected item count is directly comparable with the observed count. That GP figure is explicitly partial; untradeable or structurally different mechanics do not magically acquire a fake cash expectation.

Why isn’t a group RNG score the average of its players?

The member picker is additive. Every selected member contributes their recorded RNG to the same model. The combination table scores every non-empty subset of the currently selected synced members. With five members selected, that means all 31 possible solo and grouped combinations.

A group combination is recalculated as a portfolio; it is not the arithmetic average of the members’ solo scores. Copies, coverage, synergies, shared deficits and raid outcomes can therefore make the group result differ materially from the mean of its members.

Is a 10/10 literally the luckiest mathematically possible account?

The displayed endpoints are theoretical bounds of the presentation scale, not statements that every possible OSRS history has been enumerated. The tanh mapping intentionally saturates: once an account is already extraordinarily favourable or unfavourable, ever more extreme evidence should move the displayed score less and less rather than allowing a single outlier to blow the scale apart.

FAQ TOPIC 04

Levels & progression

Current WOM stats, gains, charts, boss progression and historical reconstruction through Time Machine.

How are old screenshot checkpoints used?

Older screenshots add dated skill checkpoints to the shared history used by Stat Progress, Time Records, Time Machine and Chronicle. XP uses the minimum required for each visible base level; exact XP and the time of day are unknown. Boss kills, activities and efficiency hours remain unknown.

Skills hidden in an image may retain their last known level, with that assumption disclosed in Time Machine. A visible total can contain levels that cannot be assigned to a particular skill. These checkpoints do not invent exact level-up times or fill gaps with made-up observations.

Imported history: loading dated entries….

    Why are there gaps in progress charts?

    The site only draws from WOM snapshots that actually exist in the saved history. A gap means there is no saved observation covering that part of the timeline; it is not filled with invented intermediate data.

    That is preferable to interpolation because a large amount of XP or KC may have been gained at any point between two snapshots.

    Why do current levels stay the same when I change the date range?

    Date ranges control gains and the historical points shown in skill and boss charts. Current levels, total XP and current KC still use the newest saved WOM snapshot.

    For a past account state, Time Machine uses the closest saved observation to the date you choose. It does not interpolate XP, levels or kills between observations.

    Why can Time Machine show a snapshot before or after my chosen date?

    A WOM snapshot may not exist on that exact day. Time Machine chooses the closest saved observation in either direction, rather than inventing an exact daily value.

    It shows the snapshot timestamp and its distance from your requested date. Historical rankings use the selected members’ saved state at those observations.

    Are suggested goals manually chosen by the players?

    No. Unless a goal is explicitly added as player-authored data elsewhere, the Goals page should be read as tracker-generated suggestions based on nearby milestones and saved progress.

    FAQ TOPIC 05

    Tracker data & syncing

    How WOM, Temple, duplicate detection and shared snapshots work outside the RNG model itself.

    Does opening the site hit the WOM or Temple APIs?

    No. Normal browsing is deliberately separated from external API traffic. Opening a page, switching menus, viewing charts, checking Collection Log Advanced or opening RNG Index loads the already-saved shared group data. Fresh WOM or Temple requests only start when somebody explicitly presses an update button.

    This means repeated browsing does not create repeated bursts of third-party API traffic.

    Why does a WOM refresh take a while?

    The tracker is intentionally polite to the WOM API. It refreshes the five members sequentially and waits between outbound WOM requests. A single member can require the current profile, snapshot-history pages, achievements and period-gain data, so a complete group refresh can take noticeably longer than a normal page load.

    That delay is deliberate. Normal browsing remains fast because it uses the saved baseline.

    Does a failed Temple update erase a player’s progress?

    No. A failed player request keeps the last usable saved Collection Log. Successful player responses can still update their logs; an unavailable account is not replaced with zero progress.

    Normal browsing reads the saved group baseline. Fresh Temple requests start only when Update Collection Log via Temple is pressed.

    What does “WOM accuracy ±…” mean?

    Historical level dates reconstructed from WOM are bounded by the gap between snapshots. If one snapshot still shows level 92 and the next shows 93 eight hours later, the tracker knows the level occurred somewhere inside that window. The displayed ± value is half of the snapshot interval, not a statistical confidence interval.

    Why do the header’s group totals stay the same when I filter members?

    The header summarizes the full five-member group. Selecting fewer players changes the comparisons, Collection Log tables and RNG analysis below, while those group-wide totals and achievements retain all five accounts.

    RNG teammate compensation, copy coverage and synergy use only the members in your selected combination.