Loka · Conquest
← Leaderboard
2022–2025 was the golden age. 2026 has been the slow year: fewer fights, shorter fights, and one side that always seems to be the stacked one.
The complaint about this year is specific. There are fewer real fights. The ones that happen are over in ten minutes. The other side folds halfway through the month. And whichever group of towns is winning, it wasn’t the same group last season and won’t be next.
This looks at every conquest fight of 20-a-side or larger on the four continents: Kalros, Ascalon, Garama and Balak; from 2018 to now. 3,244 of them.
The golden-age memory checks out: 2022–2025 was the peak for scale, and 2025 was the busiest of all. But 2026 is down about a quarter on fights, its fights are as short as they have ever been and the most one-sided in the record, and the two sides no longer keep their shape from one month to the next. Some of that, the two clear sides dissolving, was already creeping in during the good years. 2026 is where it started to cost something.
From 2021 to 2023 the number of big conquest fights roughly tripled. A full year on the four continents ran to 550–610 fights in 2022, 2023 and 2024; 2025 was just behind. That volume is what the golden age felt like. There was something to log on for in Loka.
2026 broke the pattern. The first seven months carried 293 fights against 388 in the same stretch of 2025: down a quarter, and the thinnest since 2021.
The fights that do happen are still enormous though. A median of about 90 players a side, and 2026’s median is the biggest of any year. Whatever “slow” means, it doesn’t mean small.
| Year | Fights | Median size | Median length | 30 min+ | Under 10 min |
|---|
The fights people miss are the long ones. The hour-long grind where both sides keep warping in and neither side gives up. In 2019 the median conquest fight ran 26 minutes and a third of them went past the half-hour. 2022 was the last strong year for it: median 20 minutes, one in four still going at 30.
Since 2023 the median has sat at 14–15 minutes. Roughly one fight in ten now runs 30 minutes or more, and one in five is over inside eight. A side gets wiped before the fight warms up. The full 60-minute timer was always rare, in every era; what’s gone is the middle-length conquest wars.
Whether that’s people committing less to mid and dedicating themselves more to fighting in the back or just people focussing more on taking charges, the felt result is the same: a fight is half as long as it used to be.
Each side's strength is the sum of its players' ratings, the way the fight predictor gauges it: a side that is bigger and better counts on both. Turn the two sums into a pre-fight win chance for the favoured side, sized to how many players are on the field, and you have a clean read on how lopsided a fight was before anyone swung. 0.5 is a genuine toss-up.
Of the years with enough fights to measure, 2022 was the most balanced. The favoured side went in with about a 78% chance, close enough that the result was usually in doubt. That has climbed since 2023, and 2025 and 2026 are the least balanced years yet: the favourite now starts around 85 to 87%.
And it holds up on the night. The favoured side has won about 83% of fights every year from 2021 to now, flat regardless of the balance question. The fights did not get more chaotic. They got more decided before they start.
The comparison is favourite sum against underdog sum inside each fight, so it is not a rating-inflation effect: ratings drift upward over the years, but that lifts both sides together.
Conquest is two sides. The real question is whether the same towns keep to the same side: because a side with a new name but the same core towns is still that side, and a side whose towns scatter every month isn’t one at all.
In 2018–19 there were two clean sides: Silverhand, Valentia, Eldritch and Gotham on one; Jotunheimr, Sovngarde, Stromgarde and Firsthold on the other. If two towns fought together once, they fought together about nine times in ten. 2022 still held that shape loosely: Southern Valyria, Astro and Amanita against Liberty City, Asgard, Riverside and Euphoria, same-side about three times in four.
From 2023 on, the ten-odd towns that turn up to every fight (Riverside, Silverhand, Dawnstar, Concord, Matrix, Sunspear, Asgard, Proxima, Arak) rearrange every season. In 2025, Riverside, Silverhand, Asgard and Matrix were one side, against Dawnstar and Concord. In 2026, Silverhand and Matrix crossed over to fight beside Sunspear; Dawnstar and Concord, enemies a year ago, are now together. Whether any two of these towns are allies is close to a coin toss.
| Year | One side | The other | Same-side* |
|---|
*Roughly how often a pair of the year’s main towns that fought together kept fighting together. 1.00 would be two rigid sides, 0.50 a free-for-all. Alliance names change constantly and that’s fine; what these numbers say is that the towns underneath the names don’t hold a line any more.
Split a conquest month into thirds and count how many players the eventual losing side fields in the first third versus the last. In 2021 that ratio still bounced around 1.0. The side that was behind kept turning out. From 2022 on it has mostly sat between 0.3 and 0.7, and 2026 is the lowest run of all: eight of its nine months under 0.42, most under 0.30. By the final stretch the trailing side is fielding a quarter of what it started with.
It doesn’t read as getting swiped in any one log. The fights that happen stay roughly even in numbers. One side just slowly stops turning up, and the last week or two of the month goes dead. That’s what “no close months” usually looks like: they realize its unwinnable, and the rest of the month is boring.
The fight count alone doesn’t explain it. A boring stretch of the golden age wasn’t far off this year’s pace. Three other things changed together.
Direction. The golden age was four years of the number going up. 2026 is the first sustained year below a peak everyone just lived through, and a level that felt like momentum on the way up feels like a shortfall on the way down.
The fights themselves. A 2022 fight ran twenty minutes, one in four past the half-hour, between two firm and evenly-matched sides. A 2026 fight runs twelve, one side is routinely the stacked one, and the sides scatter and re-form every month.
Not a headcount problem. As many distinct players are turning up to conquest as ever. They are just spread thinner, across fewer fights, that end sooner, between sides that don’t hold up.
Not for lack of fighting. For lack of a fair one that lasts.
A conquest month in 2022 was a month you could follow: two alliances everyone knows, matched closely enough that the result was in doubt, fights that ran long enough to swing. The scale was smaller than what came after, but the shape was right.
2026 has the crowd and the huge rosters and none of the shape. The fights are short, the sides are mismatched, and by week three whoever fell behind has drifted off. Nothing carries from one month to the next, because the sides that would carry it don’t survive until the next month.
The fighting isn’t the problem. There’s just less worth staying online for.
The 3,244 conquest fights the rating system scores as real: 20 or more players present per side, on Kalros, Ascalon, Garama or Balak, 2018 to September 2026. Fight length, winner and each side’s average skill come from that same scoring pass; monthly buckets are by the fight’s start date.
Player ratings drift upward over the years as more games are played, so a rising absolute skill level is partly mechanical and isn’t leaned on here. The gap between the two sides of a single fight, and how often the stronger side wins, are not affected by that drift.
The August 2026 archive has a gap. Roughly the 9th to the 19th are nearly empty, a collection failure rather than a quiet server, so August and September 2026 counts are understated and every year-on-year figure uses January to July. Sides are inferred from who fought beside whom (the logs carry no alliance names); the co-side grid and the “same-side” column are direct counts and don’t depend on that inference. 2018–19 has very few 20-a-side fights, so its side grid uses a looser 8-a-side threshold, and should be read as a rough picture, not a precise one.
Compiled 2026-09-09 from the local fight-log archive · descriptive analysis, not affiliated with Loka staff.