MatMetrics Features

How Glicko-2 Ratings Work

The advanced rating system behind our rankings and predictions—explained simply.

📖 10 min read•Last updated: February 2026

⚡ Quick Summary

  • Glicko-2 rating = A number representing a wrestler's skill level
  • Rating Deviation (RD) = How confident the system is in that rating
  • Starting rating = 1500 (average) with high RD (uncertain)
  • More matches = Lower RD = More stable, confident rating
  • Inactivity = RD increases = Rating becomes less certain

What is Glicko-2?

Glicko-2 is an advanced rating system developed by Professor Mark Glickman at Boston University. It's an improvement over the classic Elo system (used in chess since the 1960s) and is now used by many competitive platforms including online gaming, esports, and—thanks to MatMetrics—wrestling.

The core idea remains the same: your rating reflects your skill level, and it changes based on your results against other rated competitors. But Glicko-2 adds a crucial element: confidence.

Why Glicko-2 is Better for Wrestling

  • ✓Tracks confidence: Knows when a rating is reliable vs. uncertain
  • ✓Handles inactivity: Accounts for wrestlers who haven't competed recently
  • ✓Faster calibration: New wrestlers' ratings stabilize more quickly
  • ✓Better predictions: Uses confidence levels for more accurate win probabilities

Why Not Traditional Elo?

If you're familiar with Elo ratings from chess or video games, here's what makes Glicko-2 different:

FeatureEloGlicko-2
Rating number✓✓
Confidence measure (RD)✗✓
Volatility tracking✗✓
Handles inactivity✗✓
New wrestler calibrationSlowFast

The Key Difference

Elo gives you just a number (like 1650). Glicko-2 gives you a number plus how confident the system is in that number. A 1650 rating with low RD (lots of matches) is very different from a 1650 with high RD (only a few matches).

How It Works

Every wrestler starts with a rating of 1500 and a high Rating Deviation (RD). After each match, both the rating and the RD are updated based on the result and the opponent's rating.

Example: Expected Win

1650
Wrestler A
Favorite (RD: 45)
vs
A wins
1450
Wrestler B
Underdog (RD: 80)
A was expected to win, gains +5 rating, RD decreases slightly
B loses -12 rating (higher RD = bigger swing)

Example: Upset by New Wrestler

1650
Wrestler A
Established (RD: 45)
vs
C wins! 🎉
1500
Wrestler C
New (RD: 200)
C's high RD means their rating moves a lot: +85 rating
A's low RD means less movement: -15 rating

Rating Deviation (RD)

Rating Deviation is the key innovation in Glicko-2. It measures how confident the system is in a wrestler's rating.

Low RD (30-75)

The system is confident in this rating. The wrestler has competed recently and consistently.

  • • Rating changes are smaller per match
  • • Predictions using this rating are reliable
  • • Established wrestler with match history

High RD (150-350)

The system is uncertain about this rating. New wrestler or long period of inactivity.

  • • Rating changes are larger per match
  • • Predictions have wider uncertainty
  • • Needs more matches to calibrate

How RD Changes

↓Competing regularly: RD decreases (more confident)
↑Not competing: RD slowly increases over time (less confident)
↔After upset: RD may increase slightly (system recalibrating)

Reading Your Rating

Here's what different rating ranges generally mean:

RatingLevelDescription
2000+EliteState champion caliber, nationally competitive
1800-1999ExcellentState placer, top of weight class
1600-1799StrongCompetitive varsity, regional contender
1400-1599AverageSolid competitor, wins some/loses some
1200-1399DevelopingBuilding skills, limited experience
<1200BeginnerNew to competition, learning fundamentals

⚠️ Important: Always consider RD alongside the rating. A 1600 with RD of 50 is a confident rating. A 1600 with RD of 200 could easily be anywhere from 1400 to 1800—the system just doesn't have enough data yet.

Using Ratings for Predictions

Glicko-2 rating differences translate to win probabilities. The RD values also affect prediction confidence. This is exactly how MatchIQ works.

Rating Difference → Win Probability

Equal ratings (0 difference)50% - 50%
+100 point advantage64% - 36%
+200 point advantage76% - 24%
+300 point advantage85% - 15%
+400 point advantage91% - 9%

MatchIQ combines Glicko-2 ratings with other factors like archetypes, head-to-head history, and recent form to generate even more accurate predictions.

MatMetrics Wrestling Adaptations

We've tuned Glicko-2 specifically for wrestling:

🏆 Bonus for Dominant Wins

Pins, tech falls, and major decisions have slightly more impact than close decisions. A 15-0 tech fall shows more dominance than a 3-2 decision.

⚖️ Weight Class Context

Ratings are comparable across weight classes. A 1600 rating at 145 lbs means roughly the same skill level as a 1600 at 190 lbs.

📅 Season-Aware

RD increases during off-season to account for wrestler development. A returning wrestler's rating may be the same, but confidence is lower until they compete again.

🚫 Forfeits Excluded

Forfeits, defaults, and byes don't affect ratings. These aren't competitive results and would skew the data.

See Glicko-2 in Action

Related Articles