Skip to content
KOdex
KOdex

Glicko-2 ratings and complete records for NHRL combat robots

A focused, feature-rich combat robot rating and ranking system that updates live with events, developed with builder feedback and built on NHRL's datasets.

Leaderboards
See every bot's ranking in the scopes that are important to you. RD (uncertainty) tells you up front which bots are established in the ranking, and which still need to build history.
Bots
Data-rich bot records. Fun KPIs, data derived badges, rating changes over time in graphs, and a catalog of each bot's fight history, with the JD scorecards included when available.
Badges
Rarity-aware achievements for bots. Golden Dumpsters, podiums, World Finals, statistical anomalies, each linking to every bot that holds it.
Archetypes
Unserious at-a-glance titles associated with statistically significant biases in how a bot tends to win.
Hypothetical match-ups
Simulate fights between any two same-weight bots, using their ratings to predict the outcome and project how ratings would change for either outcomes.
Event recaps
A recap page for every NHRL event: the full bracket, event stats, rating movements, and a sortable competitor table. New events are added automatically.

What is Glicko-2?

Glicko-2 is a skill rating system (an evolution of ELO) that ranks competitors from their match results. As bots fight, their ratings update in a way that factors in the skill and rating-certainty of each competitor. KOdex computes it chronologically from NHRL's datasets and updates in real-time during events, so KOdex always reflects the NHRL competitive roster as fights happen.

Rating

The skill estimate. Movement is weighted by how surprising a result is given each competitor's prior rating.

RD (rating deviation)

How certain we are of that rating. A high RD means less certainty. RD shrinks as a bot develops history and we learn where it really stands.

Volatility

How erratic recent results have been. Streaky, up-and-down performances read as more volatile than steady, predictable ones.

Rating certainty is earned.

Every rating on KOdex carries its RD (rating deviation), so a bot with sparse fight history reads as provisional versus data-rich competitors. A settled bot has low RD, a fresh one has high RD, and that classification travels with the number wherever it appears.

The color and the readout tell you how much to trust a placement at a glance, without having to guess at how many fights are behind it.

  • Settled1842

    Low RD, a seasoned competitor

  • Developing1605

    Medium RD, only few events in

  • Provisional1490

    High RD, very new or very inactive

What makes KOdex different?

NHRL-only scope

A bot's rating here reflects only NHRL competitions where builders are trying hard to win, with no other tournaments blended in. Ratings stay meaningful because they measure one arena.

Real-time during events

Ratings and records update live as fights happen. No waiting for results to be aggregated after the event wraps, you watch the standings move in the moment.

Data-driven awards

Bots earn badges for achievements and milestones, from podiums to statistical anomalies. Each badge links to every bot that holds it, so you can explore the roster through its highlights.

Looking for bot performance across different tournament formats?
Havoc Rundown's Havoc Registry ranks bots across all tournaments, so its all-time rankings are weighted over the entire sport. It is a beautiful, feature-rich platform that gives you control over your bot's profile and record.
KOdex