Fortress ranked update

Ranked

The position you are assigned no longer decides your RP. Each position has its own bar, because a Center and a Sweep do not produce comparable numbers. That bar used to be the position's population average, which cannot separate how much the position yields from how good the people sitting in it are. Captains put strong players at Sweep and Defense and weak or new players on Wing, so Wing's bar was set low and anyone decent cleared it easily. Measured on players who did both jobs in 2026, all 30 who played Wing and Defense scored higher at Wing, worth about 18 RP a match to the same person. Each position's bar is now read at a skill level shared across all positions, with the position's measured yield subtracted on top.

38 of 38players who scored higher at Wing, before 22 of 39players who scored higher at Wing, now

Every one of those players scored better at Wing than at Sweep. Now it is close to a coin flip, which is what no position advantage looks like.

Wing still has the lowest average score of any position, and that is intended. Its occupants really are weaker on average. What was removed is the part a fixed player gained by moving positions.

NEW Defense Incentive. As a reward for doing your chores and playing the least popular role, def games that gained you RP now earns +2 overall RP on top. Of the 80 players with at least 20 games this season, 18 have never played defense once. Please go play def!

Ranks settle instead of drifting. Every rank has an expected level of play attached to it. Beat your rank's expectation and you gain more and lose less; fall short of it and the reverse. Those expectations were picked by hand and spaced evenly up the ladder, and they were never checked against what players at each rank actually do.

At the bottom that was wrong in the direction that punishes you. Bronze's expected level was set well above what Bronze players really produce, so the average Bronze player was permanently treated as overranked: gains cut, losses amplified, about two to one. 64 of 74 Bronze players finished the season with negative net RP. That is why the bottom of the ladder drained toward the floor instead of settling anywhere. Every rank's expectation is now measured from what players at that rank actually do.

At the top the problem was the opposite. The expectation was one flat value per rank, and there is no rank above Master, so nothing pushed back once a player got there and a strong player climbed all season. Master's expectation is now a line that keeps rising with your RP, fitted on previous seasons rather than the one being scored, so a good run converges on a rank instead of running away. The top of the ladder comes in from Master 917 RP to Master 438 RP.

Season rollover returns 80% of your RP, up from 70%, so a season starts nearer where the last one ended.

The one-match protection immediately after a promotion is gone.

Per position placement matches affect overall RP.

Cuts and holes

Cut and hole detection were rebuilt in a long series of changes, each one replayed over the whole record so its effect could be measured on its own. The tables below are the milestones, on a fixed corpus of 1,920 matches so the rows are comparable.

Cuts. A cut used to hinge on an approach test measured against the zone, which could throw away a real cut because the cutter drifted afterwards. Getting inside the defender's box is now the cut. On its own that change more than doubled the count, and most of the work since has been putting back the requirements a cut actually has.

Holes. Detection used to stop for the rest of the round once a side's defender died, because it was anchored to that defender's wall. A wall put up by whoever steps in now counts as a fort that holes are scored against.

That replacement-defender rule is the largest single change in the batch. It added 2,839 holes and removed none, and it completed 33 holes where the hole had been recorded but the take had been missed because detection went dark in between.

496rounds that previously recorded no hole at all

A hole in the sweep line is now its own event, and the window it was detected in was keeping only 170 of 885, so it has been widened.

Net effect

Every change above was applied to the same season. The charts below compare today's live ladder against that season run through the new system, for all 117 players with at least 10 games this season.

85 of those 117 finished lower, and the average finished 65 RP lower, 545 down to 479. The order held: rank correlation between the two ladders is 0.924, and the average player moved 9 places out of 117.

The 2026 ladder, for players with at least 10 games this season, now reads Bronze 49, Silver 37, Gold 12, Platinum 9, Diamond 5, Master 5.

Legend gets a champion row

The top player in each game mode gets their own highlight reel baked into the leaderboard. The Legend strip at the top of each leaderboard gives number one a full-width row of its own, instead of sharing column width with the rest of the top five.

What counts as a highlight is picked per mode. Fortress looks for cuts and centers. Team Sumo looks for clutches, multikills, duels, and late round chaos. Sumobar has the strip but no reel yet, because that ladder does not have the data behind it.

Past seasons where you finished first will keep your highlight reel for that year. Current season shows all years.

As soon as you get Rank 1, your highlight reel will appear, so do your best to knock Ampz and mina out of their top spots. Good luck!