Outside the profile on six of seven. Below it on bounty offered, bounty collected, opponent network, LAN wins, activity and win rate.Band is the other 5 Asia-Pacific sides holding a Major slot · dashed is their median
Profile · against the clubs holding a Major slot
Bounty offered8Bounty collected5Opponent network4LAN wins26Activity5Win rate13Head to head51
Valve's factorsOurs: activity, win rate, head to headPercentile among each region's top 40
Asia-Pacific has 5 slots: 1st place enters at Stage 2; 2nd to 5th start in Stage 1. Revenge are 24th, 19 places outside.
<1%Revenge are 24th, 19 places outside the 5, 452 points behind 5starChance of the Major: our simulation to the 2 Nov cut · standing and points: Valve's, 5 Oct 2026
11Revenge28 · 24–3124▲13, 13 places up at the cut<1
In the order of their chance of the Major, our simulation's, as Race to the Major orders them; beside each, its median place at the cut and the range the middle 80% of runs put it in. Today is Valve's standing, and the arrow how far the order by chance moves them.
FlyQuest comes up into the 5, Lynn Vision falls out.
FlyQuest comes up into the 5, Lynn Vision falls out.
How these numbers are made
Today is Valve's Regional Standings. At the cut is our simulation: the clubs in the order of their chance of the Major, as Race to the Major orders them, with each one's median place at the cut and the range the middle 80% of runs fall in. A club further down Valve's standings is listed when its median place reaches the figure.
The whole method: the model, its sources, how well it has done and what it cannot see.
7 matches in Valve's six-month window, 3 won, across 2 events, every figure Valve's own, from the published standings. One event since Valve's 5 Oct release is not in its numbers yet; it comes first, with bo3's results.
Under way· Valve counts an event once it has finished; its next release is due 2 Nov
Record0–2
Finish—
Prize—
2 matches
LPassion AcademyOct 11–2
LCTRLSep 270–2
Valve's head-to-head change per match, and which of its counting slots these results take, come with its release. Our estimate of what the event does to the points follows once it has finished.
Their most valuable event. It holds 6 of the 9 slots that count, more than any other.
Valve counts only the ten best wins in each of these three factors. 9 of the thirty slots are filled. A loss takes no slot: it moves only head to head.
1 Aug
Not linked to an event
Valve counted these matches, but we have not linked them to an event: bo3, where our calendar comes from, holds none for them, or holds it under another name.
Record1–2
Head to head-12.23
Still counts77%
These results are down to 77% of their original weight.
3 matches
A lit factor means the win holds one of its ten counting slots there; #1 is their most valuable result in that factor.
Valve counts only the ten best wins in each of these three factors. 9 of the thirty slots are filled. A loss takes no slot: it moves only head to head.
How these numbers are made
Matches, head-to-head changes, age weights and counting slots are Valve's, from its Regional Standings of 5 October 2026. Which event a match belongs to is our join to bo3's calendar.
The events since that release are bo3's results and prize tables, and the estimate is ours: Valve's standings projected to the event's end with and without it. It is replaced by Valve's own figures once Valve publishes them.
The whole method: the model, its sources, how well it has done and what it cannot see.
// Act three
What the rating is made of
Valve's model averages four factors into a starting value, then adjusts it for head-to-head results. For Revenge that is 656.2 plus 2.7.
The two parts
Valve's · release 5 Oct 2026
Starting value656.2
+
Head to head+2.7
=
Points659
The head-to-head correction is 0.4% of the total. Almost everything about where they sit comes from the four factors, not from whom they have beaten.
The four factors
Valve's · each runs from 0 to 1
Bounty offeredTheir own prize money, their ten best events, aged. Divided by the fifth-highest in the field, capped at 1, then curved.0.244median 0.386
Bounty collectedThe bounty of the sides they beat, their ten best wins. Divided by ten, then curved.0.189median 0.340
Opponent networkHow many sides their opponents had beaten, their ten best wins. Divided by ten.0.003median 0.157
LAN winsWins at LAN events, their ten best, aged. Divided by ten, then by the fifth-highest in the field.0.077median 0.400
RevengeMedian of the 11 chasing the slots in Asia-Pacific
0.800 is the highest average of any roster Valve loaded and 0.000 the lowest, both printed in Valve's file. If the top pulls away, everyone below falls without playing worse.
Valve ranked 380 rosters in this release: every lineup with a win and at least five matches in the six-month window.
Head to head is added last: a Glicko adjustment, match by match in date order.
Valve's file prints 656.2; the factors here are rounded to three decimals.
Revenge: Below the median of the 11 chasing the slots on all four factors.
Our Elo answers who wins a match not yet played; Valve's standing measures results weighted by prize money. How our Elo is scored →
The last 7 matches Valve counted · oldest first
LWLWLWL
Every match by opponent strength
Legion #316 +10.4THE UNIT #225 −10.3+20+100−10−20#1#5#10#20#50#100#200Revenge #236Stronger opponentWeaker opponent
None of the 7 matches was against a side in the world's top 30.
How these numbers are made
Valve prints the points rounded to a whole number.
In the scatter both axes are Valve's: across, the opponent's world standing (unranked at the far right); up, Valve's head-to-head change for the match.
The whole method: the model, its sources, how well it has done and what it cannot see.