Atlas Future Season Structure DRAFT Proposal: Supremacy Fortnights

In this post we continue to share our latest thinking on how to improve Atlas in the first chunk of this year. We have been working with Early Birds as well as carefully reading feedback shared on these forums and have concluded that stagnation is the primary issue to tackle, and that it largely arises from:

  1. Motivation to Fight - Atlas prizes don’t motivate / justify serious conflict
  2. Accessibility - it’s hard to attack some enemies
    • Castle Accessibility - many castles are many hops deep; this makes them too difficult to attack
    • Player Accessibility - players can attack with impunity; it’s hard to take revenge
    • Note: Currently, nearly impossible-to-counter sniping is a necessity for many less strong players. With better mechanics, it should no longer be a necessity.
  3. Mega Alliances - teams far beyond the scale of 5TAs collaborate in massive blocs, sometimes under coercion. This curtails the opportunity for enjoyable combat between closely matched adversaries.

This proposal seeks to comprehensively tackle stagnation via:

  1. Rewards Redesign
    • Competitiveness - high-stakes, close-matched fights should be far more rewarding than swaps
  2. Map Redesign - shuffle castles to new positions and cluster each team’s castles together
    • Layout - castles should not be impossible to get to (greatly reduce max castle depth)
    • Sharding - split Atlas into independent worlds (with comparably strong teams populating each)
  3. Conflict Refocus - the current conflict mechanics foster too many degenerate cases; it needs a rework

Public DRAFT #1

If we moved forward with some version of this proposal, then the rollout would coincide with the beginning of a future season. We look forward to hearing and incorporating your feedback on this proposal.

Summary: The time has come to shake up the Atlas map… into multiple maps!

  • Every 2 weeks (fortnight), teams are put onto a new map with 8 opposing teams to fight for Supremacy
  • Fight to control Shrines to earn season prizes and steal enemy castles

For a deeper dive into the details of how we hope the accomplish the above, read on!

Table of Contents

  • Key Details

    • Supremacy Phases
    • The Map
    • Shrines
      • Victory Points
      • Guards & Control
    • Castles: Conquering & Upgrading
    • Rewards
  • Extra Details

    • Math
    • Collusion Prevention Explained

Key Details

  • Teams keep their current castles (they’ll be clustered together on each new map)
  • Season prizes are earned primarily based on your team’s results
  • The 2 weeks are split into a long, prep phase (no shrine combat) and a short, intense combat phase
  • Surrendering, alliances, and safe passage are removed from the game
  • Conquering is removed from the game as it is now (the new conquering mechanic is explained below)
  • Every team sees a personalized set of 8 comparable opponents

Supremacy Phases

  • Preparation Phase is first
    • Main game major events occur during this phase; shields are up while major events run, as now
    • Can start sending guards to Shrines
  • Combat Phase is the the last 3-4 days
    • Main game major events don’t take place during this phase
    • Shrines can be taken over and generate VP

The Map

  • Each team’s castles are clustered into one area
  • Lower level castles will tend to guard the team’s borders (and thus be more exposed)
  • Teams with too few castles will be given up to nine level 2 castles
    • Your weakest castle is your Capital; it is always safe (shielded)
    • Each of your opponents will see 1 or more of your castles
    • No two opponents will see the same castle
  • One Shrine occupies the space between your team and each opposing team
  • If you switch teams, your Primarchs remain on the same map but you cannot attack or send guards to Shrines until the next preparation phase starts.

Shrines: Victory Points

  • Shrines generate VP for your team when you control them
    • VP is banked to your team’s score
    • VP is also added to the Shrine’s Bounty
      • When a team takes over the Shrine, the bounty is given to them (consumed)
      • At the end of the event, the bounty is given to the controlling team
    • VP is generated at a rate proportional to the opposing team’s elo (so Shrines contested by a stronger opponent are more valuable to control)
      • VP (and bounty) generation are less if too few guards are present
  • Bubble your enemy’s castles and control the shrine to:
    • Earn to +100% VP generated by the Shrine (doesn’t affect bounty)
    • Your opponent loses 1.5x the bonus VP you earn
    • The amount is relative to the % of your enemy’s castles (that you can see) that you bubble
      • weighted by their infrastructure score

Shrines: Guards & Control

  • Shrines are controlled by guards; send guards for your team by:
    • Attacking beasts (the guards will go to the shrine targeted by your officers)
    • Hiring guards with team glory
    • Hiring guards with gold
    • Sending troops from your barracks
    • Sending castle guards from your castle (Marshal)
  • Guards show up on the map and take time to walk to the Shrine
  • If you send guards to a shrine controlled by an enemy, then your guards will fight their guards (killing each other). The last team with guards on a shrine controls it.
  • If the Shrine is controlled by an enemy, then you can attack the shrine with a Primarch to kill guards
    • Attack power versus Shrine guards depends on only primarch tier and level … a max level sieger will do just as well as a max level rusher

Castles: Conquering & Upgrading

  • Higher level castles will have shorter bubble times than lower level castles (making them more vulnerable, but also harder to keep bubbled)

  • Removed from the game:

    • bubble retriggering cooldown
    • ability to disable shields
  • Infrastructure and Castles now require Nails (a team currency) to upgrade

    • Nails come in tiers (bronze, silver, gold, platinum and sapphire). Bronze nails are used for castle level 2 upgrades, and so on.
    • Higher tier nails are won from higher tier maps. They can be converted into lower tier nails (but not vice versa). Maximum nail storage depends on your team banks.
  • At the end of a Supremacy fortnight, do this for every team ranked in the top 3 on their map (from highest to lowest VP globally):

    1. Castle X = the best castle out of those owned by the opponents on your map you beat who have not lost more than the maximum number of castles this fortnight:

      • Rank 1 won’t lose any castles
      • Rank 2-4 can’t lose more than 1 castle
      • Rank 5-8 can’t lose more than 2 castles
    2. Castle Y = if you own 25 or more castles, then this is the worst castle you own; otherwise this is an unupgraded level 2 castle

    3. Takeover Nail Value = how many nails it would take to get from Y to X

    4. Nails Reward = a function of your rank on the map and the difficulty of your opponents

    5. If X is better than Y, then you takeover castle X. Otherwise, you earn the Nails Reward.

    6. If you have more than 25 castles, your worst castle(s) are destroyed and converted into nails (unupgraded level 2 castles are worth 0 nails).

      • Nail storage will be unlimited for the first 30 days to give teams time to use up their nails.


  • When the map ends (every 2 weeks), Atlas season tokens will be earned by each member:
    • Team award, based on how your team ranks in VP on your map + difficulty of your opponents
    • Personal bonus, based on how you rank on your team in glory earned during the fortnight
      • The bonus is up to 20% of the team award the member earns
    • Top Killers award, given to the top 1000 (edited; was 100 which was incorrect) glory earners each fortnight; the award is proportional to how much glory they earned
  • Season prizes will be increased to reflect the removal of the Kill Troops event
  • Atlas season tokens are only awarded if you finish on the same team as when the map started

Extra Details

One does not need to understand or even be aware of this section in order to excel at Supremacy. These details are important to prevent collusion, etc. but are only important under the hood to keep things running well, and are not required (or even helpful) to play!


  • Atlas Team Elo
    • Initial Elo = roughly based on Atlas team rank; hand-tune the top 100
    • End of Fortnight = Treat the map like 8 separate win-loss matches. Compute the elo change for each win/loss using the initial elos. Apply the sum of the elo changes to the initial elo to get the team’s new elo
  • VP/sec = 10 * min(3, 1 / sqrt(Win Chance))
    • win chance is based on the standard elo calculation
    • 10/sec for a 50% win chance
      • up to 30/sec for 11.1% win chance or less
      • down to 10/sec for 100% win chance
    • Max score ~ 250 million
      • 4 days = 345,600 seconds
      • 8 shrines
      • 10 VP/sec
        • 3x if super hard opponent
        • 3x if you get the generated VP, bounty VP, and bubbled enemy castle VP

Collusion Prevention Explained

Atlas is a team-focused endeavor. We want that focus to be primarily on team-versus-team – not teams-versus-team. This section explains how the above mechanics take pains to better achieve this focus. Note that it’s not ironclad, but we will do much better with these mechanics than in the current Atlas season.

  • Each of your opponents will see you as an opponent.
    • They may not see your other opponents: their other opponents are probably different!
    • You will only see and be able to interact with Primarchs from the opposing teams on your map

Reason 1: Ensure each team has a competitive, fair set of opponents; nobody is stuck with opponents which are all stronger.

Reason 2: Reduce the opportunity for collusion. (Collusion is further minimized by each Shrine only being visible by the two teams competing over it, and each opponent only being able to see a distinct subset of your castles.)

Note: To further minimize excessive collusion, no more than 3-4 teams will all be on each others leaderboards. This minimizes even the theoretical opportunity to coordinate and gang up (certainly far less than today).

  • Each of your opponents can only see a unique subset of your castles. This reduces the opportunity for The collusion (e.g., multiple opponents cannot dogpile on one of your castles)…
  • The VP loss hurts more than the bonus the other team gains. So it’s inefficient to trade bubbling. It’s much, much better to bubble your opponent the entire time and try to avoid being bubbled yourself.

Here’s a version where each team gets 4 opponents and no opponents overlap (i.e., if I fighting A, B and C then A, B and C are not fighting each other). This graph shows just 8 teams:


Please discuss over in the discussion thread, keeping in mind that this is a draft proposal and nothing is final.