๐Ÿ”™

What every column means

Short answers, one column at a time. Hover any number on the main page for details. Back to the stats ยท Update log ยท Formulas: docs/metrics.md

One row, like on the main page ๐Ÿ‘‰

Type any cached name. Tap a column to see what it means.

The idea in one picture

easy third hard third K/D easy opponents hard opponents โ†’ (how fast they kill)

Most โœจ columns read this line: your K/D against opponents from easy (left) to hard (right). It uses the ~200 players you fight most, not your whole career.

Why Honu and Fisu donโ€™t show these

Why it matters: a player can show 3.96 lifetime K/D and only 1.15 against the people he really fights.

They look up one player. These columns also need the stats of your ~200 opponents, so this site keeps them all cached and refreshed. The ๐Ÿ“Š Census / Fisu / Honu style columns are the same numbers the trackers show.

โœจ Adjusted (the default columns)

โš”๏ธ iVi โ€” your overall rating

416
better than 59% of players

How good you are, once your opponentsโ€™ strength is counted.

  • High: you beat strong players at a good pace.
  • Low: your wins come from weak players, or you play slowly. Below zero shows as 0.
How itโ€™s measured
  • Beating strong players counts more than beating weak ones.
  • Fast killing adds a bonus; slow, safe play costs a little.
  • Infantry only: vehicle and MAX kills above 10% lower it (๐Ÿšœ). Confirmed padders lose 10% (๐Ÿ’ฉ).
  • PlayStation: the 600 baseline is divided by the PC/PlayStation kill-speed gap (rechecked weekly), so top players compare fairly with PC.
  • PC: evened out for ๐Ÿชณ Zerg: low ๐Ÿชณ goes up a little, high ๐Ÿชณ down, the average crowd (๐Ÿชณ 26.8) stays.

Example โ†“

๐Ÿƒ Activity โ€” winning fights, fast

0.45
better than 64% of players

Your โ˜ ๏ธ K/D ร— how fast you kill.

  • High: lots of fights won, quickly.
  • Low: you lose more than you win, or you kill slowly.

๐Ÿฆ Brave โ€” does it hold against strong players?

1.09
better than 66% of players

Your ๐Ÿƒ Activity, checked against the fastest-killing enemies you meet.

  • High: you stay strong when the enemy gets strong.
  • Low with a high iVi: you pick easy fights and avoid even ones. ๐Ÿณ
How itโ€™s measured

PlayStation: enemy kill speeds and your ๐Ÿƒ Activity are multiplied by the PC/PlayStation kill-speed gap (rechecked weekly) before ๐Ÿฆ Brave is worked out, so top players compare fairly with PC.

๐Ÿ›ก๏ธ Resist โ€” K/D, weighted by enemy danger

0.77
better than 69% of players

Your โ˜ ๏ธ K/D ร— how fast your enemies kill. 1.0 K/D against killers is worth more than 1.0 against beginners. โš”๏ธ iVi is built from it.

  • High: you hold your own against fast killers.
  • Low: your K/D comes from slower enemies.

โ˜ ๏ธ K/D โ€” against people you actually fight

0.804
better than 64% of players

Kills รท deaths against your ~200 most-fought players. Not your lifetime K/D. Every โœจ column starts from this.

  • High: you win most of those fights.
  • Low: you die more than you kill.

โšก Const โ€” do you hold up at the hard end?

75%
better than 75% of players

Your K/D in the hard third รท your K/D in the easy third of the picture above (max 1). Shown as a percentile.

  • High (line stays flat): you do as well against hard opponents as easy ones.
  • Low (line drops): you farm easy opponents and struggle against hard ones.
How itโ€™s measured

Catch: someone equally weak everywhere also scores high, so Const is only a colour hint, never a rating.

๐ŸŽฒ Luck

โ€”

Wins more (+) or fewer (โˆ’) of the kills against the same opponents than โš”๏ธ iVi predicts. Not literally luck; vehicle kills are already adjusted.

A lasting positive luck means your โš”๏ธ iVi is understating you โ€” you are growing faster than the rating. The rating catches up, and the luck settles back toward 0.

How itโ€™s measured
  • Formula: for each opponent with 10+ kills between you, take your actual share of those kills minus the share the โš”๏ธ iVi model predicts from the two ratings โ€” then average over all such opponents. +9% means you won 9 points more of the kills than the model expected.
  • Example: YuigahamaYui โ€” gap โˆ’10โ€ฐ = โˆ’1% over 21 opponents (she beat 57% of them). About as expected, slightly unlucky.

Shrunk toward 0 for few opponents, so a player with only a handful of rivals never reads as wildly lucky or unlucky. The table marks it โ€œa bitโ€ past โˆ’48โ€ฐ / +64โ€ฐ and โ€œmuchโ€ past โˆ’71โ€ฐ / +150โ€ฐ.

๐Ÿค Assists

0.60
better than 79% of players

Assists per kill in recent Honu sessions: how often you help teammates get kills.

How itโ€™s measured
  • Formula: assists รท kills in recent Honu sessions. An assist is a kill-assist XP tick (ids 2, 3, 371, 372) โ€” you damaged the enemy and a teammate finished them.
  • Example: YuigahamaYui โ€” 0.60, 60 assists per 100 kills, better than 79% of players.

Shown from 100 kills. Apr 1 โ€“ May 6, 2026 is skipped: a game bug recorded no assists then, so too few kills are left to measure. Higher means you share more kills with teammates โ€” it feeds ๐Ÿชณ Zerg, not โš”๏ธ iVi directly.

๐Ÿชณ Zerg โ€” riding a big group

79
better than 20% of players

The lower of two percentiles: ๐ŸŽˆ Inflation (your K/D lifted by easy fights) and ๐Ÿค Assists (kills shared with teammates). 0โ€“99. Shown from 100 kills in recent Honu sessions (โ€œโ€”โ€ until then).

  • High: easy fights and lots of shared kills, at the same time. Lower is better.
  • Low: you earn your K/D in even fights, or you fight on your own.
How itโ€™s measured

Why the lower of the two: a solo farmer of easy fights has high Inflation but few assists, and a team player in hard fights has many assists but no inflation. Neither is a zerg, so both stay low.

Crowd check: where we have Honu XP detail (50+ deaths, 2+ h), we count per death the revives, heals, shield repairs and ammo you received, plus engineer Hardlight barrier XP (it only ticks inside a big crowd). Zergs die in a crowd, so the crowd picks them up. Revives or the crowd score in the top quarter = Zerg confirmed (hover says so). Both in the bottom half = a high Zerg is softened step by step. Never raised. No ๐Ÿชณ Zerg but a top-quarter crowd score = an estimate โ€œ~Nโ€, shown only, never sorted or marked. No XP detail = no change.

Apr 1 โ€“ May 6, 2026 is skipped for ๐Ÿค Assists and ๐Ÿชณ Zerg: a game bug recorded no assists then (fixed May 6). Too few kills left = โ€œโ€”โ€.

๐Ÿงผ Pure โ€” your true solo ability, cleaned twice

234

What your rating would be if every fight counted as if you fought alone, and your K/D had to hold up against the fastest killers too. Pure iVi takes the crowd out; this takes the crowd and the cheese out.

How itโ€™s measured
  • Formula: ๐Ÿง˜ Pure iVi ร— (1 โˆ’ ๐ŸŒˆ G), where Pure iVi is adjs ร— (1 โˆ’ 0.00436 ร— ๐Ÿชณ). The base is the Pure value โ€” never the displayed (๐Ÿชณ-evened) โš”๏ธ iVi, or ๐Ÿชณ gets charged twice.
  • Chosen, not measured: G is the share of cached players you beat on K/D and Brave at once โ€” a rank share, not a cheating probability. Nothing in the data says 1% of G equals 1% of your score; that rate is a choice. The ๐Ÿชณ half is measured (fitted on 915 duel pairs).
  • Example: Eyfia โ€” Pure 921 at G 9.1% โ†’ 921 ร— (1 โˆ’ 0.091) โ‰ˆ 838. Aekris โ€” Pure 2667 at G 99.3% โ†’ 19.

Not a replacement for โš”๏ธ iVi. ๐Ÿชณ and G are near-independent (r โ‰ˆ โˆ’0.06), so stacking them docks each sin once. PC only; no ๐Ÿชณ score or no G โ†’ โ€œโ€”โ€.

๐ŸŒˆ G โ€” how cheesy is this K/D?

21.44%

The share of all cached players you beat on K/D and Brave at once. A high K/D alone doesn't do it โ€” you also need a collapsed ๐Ÿฆ Brave. ๐Ÿช– Infantry doesn't score anymore: it only lives in the gate โ€” the cheesy set is Brave under 1.7 and Infantry under 50%. The 1.7 is the trimmed mean ๐Ÿฆ Brave of the top 100 by K/D on PC โ€” the typical Brave of a top-K/D player (the raw mean, 6.68, is wrecked by two outliers at 461 and 41). On PlayStation the same figure is 2.9 (raw mean 3.17, top two at 19.24 and 13.05).

How itโ€™s measured
  • Formula: (1 โˆ’ K/D rank) ร— (1 โˆ’ Brave rank), ranks over all cached players. Each factor is the share of players you beat on that one dimension.
  • Example: Tepe8756 โ€” K/D 4.03 (rank 129 of 18,571), Brave 0.024 (rank 226): 0.9931 ร— 0.9878 โ‰ˆ 98.10%.

Most players sit at 10โ€“23% (median 17.21%); 90% are under 28.95%. The top 100 start at 70.09%. โ€” a different world from the rest.

๐Ÿช– Infantry โ€” kills made on foot

54.2%

% of your lifetime kills made with infantry weapons. MAX counts as infantry here; vehicles, turrets and roadkills donโ€™t. Shown from 100 kills (โ€œโ€”โ€ until measured). Hover for your MAX and vehicle shares.

How itโ€™s measured
  • Formula: foot kills รท all lifetime kills. Foot means kills with infantry weapons; MAX counts as infantry, vehicles, turrets and roadkills donโ€™t.
  • Example: YuigahamaYui โ€” 54.2%, about 54 of every 100 kills made on foot.

๐Ÿšœ Vehicle: a ๐Ÿšœ appears next to your name, like ๐Ÿชฆ, when you mostly play in vehicles. It works like ๐Ÿชณ, and every line is taken from the data at each build: your infantry share is in the lowest quarter of all ranked players (below about 71.9%) and your Honu XP shows a lot of vehicle repairs or vehicle kills per minute (top quarter of players with 2+ hours of XP detail); or your infantry share is in the lowest 2% (below about 21.7%), even without XP detail. Hover it for the hint; the Infantry number itself does not change. On ๐ŸŽฅ LIVE the ๐Ÿช– turns into ๐Ÿšœ below that infantry cut and fills: almost empty just under the cut, full (and shiny) at half the cut (about 36%) or less. Between the cut and the median (about 85.1%) the ๐Ÿช– stays empty; above the median it fills up to 100% infantry. Not the same as the ๐Ÿšœ vehicle cut on โš”๏ธ iVi or the ๐Ÿ”€๐Ÿšœ Diff reduction: this mark changes no number.

Marks and cuts on โš”๏ธ iVi

๐Ÿšœ Vehicle + MAX cut on โš”๏ธ iVi

The first 10% of vehicle kills are free; every % above costs 1% of iVi. MAX kills work the same way, separately. The two add up.

  • 45% vehicle + 40% MAX โ†’ โˆ’35% โˆ’30% = โˆ’65%
  • 96% vehicle โ†’ โˆ’86%
  • Pure infantry โ†’ no cut
How itโ€™s measured

Why: farming infantry from a tank, aircraft or MAX is easier than on foot. Hover the iVi value to see the cut. The site owner can exempt a player.

๐Ÿชณโš”๏ธ Zerg evening-out on โš”๏ธ iVi

Fighting inside a big crowd lifts the numbers a little beyond what you win on your own. We measured how much on real fights, take exactly that back, and give it to players who fight with less help.

How itโ€™s measured
  • Measured, not guessed: 103,689 pairs of players who fought each other 30+ times. At the same โš”๏ธ iVi, each ๐Ÿชณ step means about 0.08% fewer kills against the other player.
  • Both ways: ๐Ÿชณ below 26.8 (the average crowd, where players without a ๐Ÿชณ score fight) goes up, above it goes down. At iVi 1,000: ๐Ÿชณ 0 โ†’ +23, ๐Ÿชณ 50 โ†’ โˆ’20, ๐Ÿชณ 99 โ†’ โˆ’62.
  • Checked on players it never saw: fitted on half the players and tested on the other half, the ๐Ÿชณ bias in their fights disappeared and fight predictions got slightly better.
  • Same weight in every tier; most players move by 1% or less. No ๐Ÿชณ score or PlayStation โ†’ unchanged. Vehicles are not counted twice: the ๐Ÿšœ cut already handles them. Away for 60+ days (no recent sessions) โ†’ unchanged until a refresh brings fresh data.

Research and numbers: docs/zerg-adjustment.md ยท discussion: issue #84. Hover โš”๏ธ iVi to see your before โ†’ after.

๐Ÿ’ฉ Confirmed padders: โˆ’10% on โš”๏ธ iVi

Players confirmed farming kills on their own or friendly alt accounts get ๐Ÿ’ฉ and lose 10% of iVi (kills on farm accounts are already left out). Hover ๐Ÿ’ฉ for the reason. Think itโ€™s wrong? Contact the site owner.

What ๐Ÿ›– / โ›‘๏ธ means

๐Ÿ›– next to a name: rarely dies, slow pace, low accuracy. That player's positive Diff is reduced, because it likely comes from safe positions.

How itโ€™s measured

๐Ÿ›– Spawn Room Warrior: rarely dies, plays at a slow pace and aims poorly, yet โš”๏ธ iVi beats IvI by a lot. That positive ๐Ÿ”€ Diff likely comes from safe positions, so it is reduced. The check gets stricter the higher the โš”๏ธ iVi and the longer the play time: newer and weaker players are judged gently, and players who die often are never touched.

โ›‘๏ธ Squad support: the same pattern as ๐Ÿ›– (rarely dies, a high โš”๏ธ iVi for the pace and aim), from a player who mostly heals, revives, resupplies or repairs their squad. Safe play, but outside the spawn room. The ๐Ÿ”€ Diff is lowered the same way.

๐Ÿ“Š Census / Fisu / Honu style (the tracker numbers)

Lifetime numbers from Census, no adjustment. The same ones Honu and Fisu show.

KD
Kills รท deaths over your whole career, against everyone. Flatters players who farm weak opponents.
KPM
Kills per minute over your whole career.
own KPM
Kills per minute with your own weapons. Drives the iVi pace bonus or cut.
Acc %
Share of your shots that hit.
HSR %
Share of your kills that were headshots.
IvI
Censusโ€™s lifetime infantry score, with no opponent weighting. Compare it with โš”๏ธ iVi (see the example).

Example: ๐Ÿ‘‰ YuigahamaYui

What trackers show: KD 1.71 ยท IvI 216, which looks below average.

What the site sees:

  1. Their 200 most-fought opponents kill at 0.96 per minute (a typical playerโ€™s opponents: ~0.35). Very hard opposition.
  2. โ˜ ๏ธ K/D against them: 826 รท 1027 = 0.80
  3. ๐Ÿ›ก๏ธ Resist: 0.80 ร— 0.96 = 0.773
  4. Rating: 600 ร— (1 + logโ‚‚(0.773 รท 0.6)) = 819
  5. Slow pace (own KPM 0.56, under 0.8): ร— 0.86 โ†’ โš”๏ธ iVi 708
  6. ๐Ÿšœ vehicle kills 46%: โˆ’36% โ†’ โš”๏ธ iVi 455
  7. ๐Ÿชณ Zerg 79, evened out to the average crowd (๐Ÿชณ 26.8) โ†’ โš”๏ธ iVi 416

Result: 416, better than 59% of tracked players (median 278). The public IvI said 216. Same sample: ๐Ÿƒ Activity 0.45 ยท ๐Ÿฆ Brave 1.09 ยท โšก Const 0.66.


Debug-only columns

Older or in-between numbers. Tick โ€œshow older debug statsโ€ in the main page footer to see them. Not used to judge anyone.

โš™๏ธ Mech%
A guess of your vehicle / mech share, projected from ๐Ÿ›ก๏ธ Resist. Not measured.
๐ŸŽฏ๐ŸŽˆ ivi
โš”๏ธ iVi before the pace bonus or cut, so slow play still looks good.
eKPM
How fast your opponents kill, on average. Higher = harder opposition.
own KPM
Same as the ๐Ÿ“Š Census / Fisu / Honu style column.
๐Ÿ“Š COI
๐Ÿ›ก๏ธ Resist รท 0.6. 1.0 = soft opposition; higher = harder.
๐Ÿ“‰ Slope
The angle of your line. Negative = your K/D drops against harder enemies. Feeds ๐Ÿฆ Brave.
๐ŸŽˆ Inflation
Lifetime KD รท KD against enemies killing at 0.5+ per minute. Above 1 = easy fights inflate your KD. Lower is better.
โš”๏ธ iVi before ๐Ÿชณ
โš”๏ธ iVi before the ๐Ÿชณ Zerg evening-out, to compare with the โš”๏ธ iVi column.
๐Ÿ‘ฟ Sore %
Share of measured players you beat on revenge kills, as a percentage โ€” the top player reads 100%, a player with no revenge kills reads 0%. The Revenge Kill XP tick. Higher means you avenge your deaths more often; it runs with lower K/D, since revenge needs a death first.

๐Ÿง˜โ€โ™‚๏ธ Pure iVi โ€” your true solo ability

298

What your rating would be if every fight counted as if you fought alone. A high iVi on a zerg surfer says: you have it only because of crowd efforts. Pure iVi says: this is your true ability as a solo player.

How itโ€™s measured
  • Formula: adjs above 0 โ†’ adjs ร— (1 โˆ’ 0.00436 ร— ๐Ÿชณ); adjs at or below 0 โ†’ adjs โˆ’ 0.00436 ร— ๐Ÿชณ ร— 610.6. Each ๐Ÿชณ point costs about 0.44% of the rating; ๐Ÿชณ 100 โ†’ โˆ’43.6%.
  • Measured, not guessed: fitted on 915 duel pairs (723 infantry-only). At the same adjs gap, each ๐Ÿชณ point of gap meant 0.286 fewer kill-gap points against 0.656 per adjs point โ€” 0.286 รท 0.656 โ‰ˆ 0.00436.
  • Example: ahorn, adjs 1225 at ๐Ÿชณ 87 โ†’ 1225 ร— (1 โˆ’ 0.00436 ร— 87) โ‰ˆ 760.

Not a replacement for โš”๏ธ iVi, and not the ๐Ÿชณโš”๏ธ evening-out: that one centers on the average crowd (๐Ÿชณ 26.8) and lifts low-๐Ÿชณ players up; Pure only ever discounts, never boosts, and has no freshness gate. PC only; no ๐Ÿชณ score โ†’ โ€œโ€”โ€.

๐Ÿ”€ Diff โ€” aim on paper vs real results

Your โš”๏ธ iVi percentile minus your public IvI (aim) percentile, from โˆ’99 to +99. Practice minus theory.

  • ๐Ÿคฅ Impostor (โˆ’33 or lower): great aim on paper, but real fights donโ€™t convert it.
  • ๐Ÿ˜ค DieHard (+33 or higher): you win more than your aim predicts โ€” positioning, fight choice, grit. Vehicle play can push it up too.
  • In between (โˆ’32โ€ฆ+32): balanced, aim and results roughly agree.

Neither side is better or worse. It shows a play style, not cheating.

๐Ÿ›– next to a name: rarely dies, slow pace, low accuracy. That player's positive Diff is reduced, because it likely comes from safe positions.

Vehicle damp: above the typical vehicle share (about 15% of kills), a positive Diff counts only in proportion to your kills on foot (46% vehicle โ†’ ร—0.63, 76% โ†’ ร—0.28). If ๐Ÿ›– also applies, only the stronger cut counts. Hover the Diff value to see it.

Newbie softening: a negative Diff is scaled by estimated play time (lifetime kills รท lifetime KPM). About 6 h or less (the newest 5% of players) โ†’ 0; then it rises smoothly to the full value at about 385 h (the median player). New accounts are still learning, so they are not called ๐Ÿคฅ Impostors yet. Positive Diff is untouched. Hover the Diff value to see it.

๐Ÿฃ next to the name: a new player, under about 20 h played (the newest 14% of players), whatever the Diff. Up to that point about 29% of players read as ๐Ÿคฅ Impostors; after it the share drops. ๐Ÿฃ also shows when a negative Diff was softened by 10 or more.