HOW SCORING WORKS

S2 used a straightforward Elo-style formula applied per set, carried on players.current_elo. It had two known flaws found by later backtesting: it rewarded playing a lot of sets over playing well, and it didn't account for how strong the rest of the lobby was — beating a stacked lobby paid the same as beating a quiet one.

First season on Glicko-1. Each player has a rating and an uncertainty (RD) — the ladder number is the conservative estimate (rating − 1.5×RD), so a hot first night can't jump ahead of a season of real play. Your first 5 sets are placement sets, where your rating is provisional and can swing heavily. A points tie in a set is broken by the official standing (damage, then average placement, then kills, then coin flip) rather than scored as a draw.

Same Glicko-1 engine and settings as dpl1. Mid-season, set-tie pricing was tightened: a points tie broken by the rules' own tiebreak order (damage > average placement > kills > coin flip) is priced as a win/loss, not a 0.5 draw — closing a gap where moderators had been nudging damage totals by hand to force a tiebreak result the engine would otherwise call a draw.

Same Glicko-1 engine, with two additions made mid-season after player feedback:

• A per-set cap: once a player is past their first 5 placement sets, one set can move their rating by at most ±150. Before this, a single bad set could wipe out several good ones.
• A podium minimum: finishing 1st, 2nd or 3rd always moves the number shown on the ladder up by at least +5 / +3 / +1, even if the underlying math would call it a smaller gain or a loss. A podium finish never costs rating, full stop.

Trophies were minted for the season's top 1 / top 10 / top 25.

Two changes from how DPL S3 was played, both aimed at making a single set matter less relative to a season of play:

• Before, a set counted as 9 separate head-to-head results (every pair in a 10-player lobby). Now it counts as about 3 — each result is scaled down, so one set still moves your rating, just by less, win or lose.
• The number on the ladder can't go up or down by more than 40 in a single set. The cap applies to the SHOWN number this time (not the underlying rating), and starts after your first 5 sets, same as before.

There are no per-finish loss caps this season (a finish-based cap was tested and rejected). The podium minimums (+5 / +3 / +1) and the inactivity re-widening of uncertainty while you're away are unchanged from DPL S3.

• Every DPL season starts everyone fresh at the same seed rating (1000) — ratings never carry across a season cutover.
• The first 5 sets of a season are placement sets for every player, every season.
• Sitting out never moves your number on its own; it widens your uncertainty, so your first set back after a break moves you further than usual. See the penalty system for inactivity decay, which is separate from this.