How ratings work
Each player rating is an Elo score (starting at 1000). Logic lives in rating-elo.ts and is summarized below.
Overview
Map wins and losses drive rank; K/D and damage add a small per-map nudge.
- Elo rating — map W/L vs opponent team strength (primary), plus a capped performance adjustment.
- K/D and avg damage columns — career totals on the leaderboard for context; the map nudge uses per-map ranks.
How a map counts
Each uploaded session is one map played by both teams.
- Players on the winning side have won = true; losers have won = false.
- Teams are inferred from the session — no separate team field is stored.
- All maps in the current year are replayed in date order to compute today’s rating.
Elo formula (outcome)
Before each map, team averages use ratings earned from earlier maps only. Full K is applied to the result.
expected = 1 / (1 + 10^((oppTeamAvg - myTeamAvg) / 400)) actual = 1 if won else 0 outcomeDelta = K × (actual - expected)
Start 1000, K = 24, scale = 400
Performance adjustment (skill mix)
K/D and damage on that map are ranked against everyone who played the map. Default weights: 40% outcome label / 60% skill scale (sum 100%).
perfIndex = 0.6 × kdPercentile + 0.4 × adrPercentile skillDelta = (skillWeightPct / 100) × K × (perfIndex - 0.5) totalDelta = outcomeDelta + skillDelta defaults: outcome 40% / skill 60%
Dashboard sliders (K, scale, outcome/skill %) are session-only testers. Docs show production defaults only.
Example map deltas
Equal 2v2 teams, K = 24, outcome/skill weights 40% / 60%. Outcome Δ is the W/L part; skill Δ is the performance nudge.
| Result | Performance | Outcome Δ | Skill Δ | Total Δ |
|---|---|---|---|---|
| Win | Average on map | +12.0 | +0.0 | +12.0 |
| Win | Best on map | +12.0 | +7.2 | +19.2 |
| Win | Worst on map | +12.0 | -7.2 | +4.8 |
| Loss | Best on map | -12.0 | +7.2 | -4.8 |
| Loss | Worst on map | -12.0 | -7.2 | -19.2 |
Default parameters
Production defaults (not live dashboard slider values).
| Parameter | Default | Production defaults (not live dashboard slider values). |
|---|---|---|
| Start rating | 1000 | New player baseline |
| K factor | 24 | Outcome change speed |
| Elo scale | 400 | Favorite/underdog curve |
| Outcome weight | 40% | Map W/L label (full K applied) |
| Skill weight | 60% | K/D + damage nudge scale |
| perfIndex mix | 60% K/D, 40% damage | Within-map performance rank |
perfIndex illustration
One map, five players — ranks are within that map only.
| Player | K/D rank | Damage rank | perfIndex | Skill Δ (winner) |
|---|---|---|---|---|
| A | 1st | 1st | ~1.0 | +7.2 |
| B | 3rd | 2nd | ~0.5 | +0.0 |
| C | 5th | 5th | ~0.0 | -7.2 |
Skill Δ for a winner at default settings; exact value depends on K and skill weight.
Equal teams
When both team averages match, expected win is 50%. With default K = 24, the outcome part of a win is about +12 before skill adjustment.
Neutral maps
If no player won in a session (all won = false), the map is treated as a scrim and skipped — no rating change for anyone.
Display stats
Career totals on the leaderboard (not the per-map rank used for skill adjustment):
kdRatio = totalKills / totalDeaths (or totalKills if no deaths) averageDamage = totalDamage / totalMaps
These columns help you judge overall form. The skill nudge uses how you ranked on each map only.
Rating change badge
The +/- badge compares your Elo at the end of the last global play day vs the previous play day (365-day window). Shown only if you played on the last day.
Typical range
In a small group, active players usually sit roughly between 900 and 1100. There is no upper cap at 100.