osu!Skill Wiki

Complete documentation on how osu!Skill analyzes your gameplay and what all the numbers mean.

Table of Contents

Overview

What is osu!Skill?

osu!Skill is a comprehensive skill analysis system that goes beyond the traditional PP (Performance Points) system. It analyzes your recent gameplay against your peak performance to determine your current skill level and consistency.

Key Philosophy: Your skill is measured by how well you're currently performing compared to your peak ability, not just your highest PP scores.

The system uses three main components: Recent Skill (your current performance), Peak Skill (your best historical performance), and Skill Match (how close you are to your peak).

Quick Start Guide

  1. Check Your Profile - Log in with your osu! account to see your current skill analysis.
  2. Understand Your Verdict - Your verdict tells you how well you're performing compared to your peak.
  3. Review Insights - Get personalized recommendations based on your play patterns.

Core Metrics

Recent Skill
Measures your current skill level based on your recent plays (up to 30 plays, prioritizing newest). This considers accuracy, star rating, mods, temporal decay, and retry patterns to emphasize genuine performance.
Range: 0-200+ (higher is better)
Peak Skill
Measures your historical peak performance based on your top 25 plays by PP. This represents what you're capable of at your best.
Range: 0-200+ (higher is better)
Skill Match
The percentage of your peak skill that you're currently maintaining. This is the primary indicator of your current form.
(Recent Skill / Peak Skill) × 100 × Reliability Factor
Range: 0-100% (higher means closer to peak)
Confidence
Indicates how reliable the analysis is based on data quality, play volume, diversity, and consistency.
Range: 0-100% (20%+ minimum for analysis)

Verdicts Explained

Accurate
(88%+)
You're performing at or very close to your peak skill level. Your current rank accurately reflects your ability.
Slightly Rusty
(78-87%)
You're performing slightly below your peak but still maintaining good form. Minor practice should restore peak performance.
Rusty
(60-77%)
Noticeable decline from peak performance. You may need more practice to return to your best form.
Overranked
(40-59%)
Significant gap between current and peak performance. Your rank may be inflated compared to current skill.
Inactive
(<40%)
Very low recent activity or performance well below peak. Extended practice needed to restore skill.
Insufficient Data
N/A
Not enough recent plays or confidence too low for reliable analysis. Play more to get accurate results.

Calculations

Skill Components

Each play is broken down into three skill components with improved accuracy scaling:

  • Aim Skill (40% weight): Based on star rating, AR (Approach Rate), and accuracy. Uses conservative difficulty scaling and AR adjustments.
  • Speed Skill (40% weight): Based on BPM relative to 180 BPM baseline and star rating. Speed-focused maps are weighted appropriately.
  • Accuracy Skill (20% weight): Heavily rewards high accuracy (95%+) with exponential scaling. Fixed calculation ensures proper accuracy rewards.

Accuracy Scaling

The system uses tiered accuracy scaling for more realistic skill representation:

  • 99%+: 0.99 + (accuracy - 99) × 0.01
  • 96-99%: 0.96 + (accuracy - 96) × 0.01
  • 90-96%: 0.90 + (accuracy - 90) × 0.01
  • 80-90%: 0.80 + (accuracy - 80) × 0.01
  • <80%: Linear scaling × 0.8

Mod Multipliers

Mod multipliers have been refined with proper combination bonuses:

Single Mods:

  • HD: 1.06x
  • HR: 1.12x
  • DT/NC: 1.18x
  • FL: 1.15x
  • EZ: 0.88x
  • HT: 0.82x
  • SO: 0.92x
  • NF: 0.98x

Mod Combinations (override individual mods):

  • DT+HR+HD: 1.32x
  • DT+HR: 1.25x
  • DT+HD: 1.22x
  • HR+HD: 1.16x

Multiplier Range

All mod multipliers are capped between 0.6x and 2.2x to prevent extreme values.

Temporal Decay

Temporal decay has been made more gradual to better reflect skill retention:

  • ≤7 days: 1.0x (full weight)
  • 8-14 days: 0.98x
  • 15-30 days: 0.93x
  • 31-60 days: 0.85x
  • 61-90 days: 0.75x
  • 91-180 days: 0.6x
  • >180 days: Exponential decay with minimum 0.35x

Retry Detection

The system intelligently detects retry attempts to provide fair analysis:

  • Detection Window: 30 minutes between plays on the same map
  • Retry Penalty: 0.9x weight (reduced from previous harsh penalties)
  • Search Range: Checks up to 8 previous plays for retries

Advanced Features

Confidence System

The confidence system has been refined for more accurate reliability assessment:

Volume (40%)

Based on number of recent plays with more generous scaling:

  • 25+ plays: 1.0
  • 15-24 plays: 0.85 + scaling
  • 10-14 plays: 0.7 + scaling
  • 6-9 plays: 0.5 + scaling
  • <6 plays: Minimum 0.3

Diversity (25%)

Based on unique beatmaps played:

  • 15+ unique maps: 1.0
  • 10-14 maps: 0.8 + scaling
  • 6-9 maps: 0.6 + scaling
  • <6 maps: Minimum 0.4

Consistency (35%)

Based on accuracy standard deviation with less harsh penalties:

max(0.3, 1 - (accuracy_std / 30))

Data Requirements

Minimum Requirements

  • Recent Plays: 6 minimum for analysis
  • Top Plays: 8 minimum for peak skill calculation
  • Confidence Threshold: 20% minimum (down from 50%)

Special Cases

  • Inactive Users: No recent plays but sufficient top plays → "inactive" verdict
  • Low Confidence Override: 15%+ confidence with 50%+ skill match still allows analysis

Data Validation

Each play must have valid:

  • Accuracy: 0-100%
  • Star rating: 0-12*
  • AR: 0-11
  • BPM: 30-600
  • Timestamp: Valid ISO format

Insights System

The system generates personalized insights based on your play patterns:

Performance Insights

  • Old Top Plays: Detects when most top plays are >9 months old
  • Difficulty Analysis: Compares recent vs. peak difficulty levels
  • Consistency Evaluation: Analyzes accuracy consistency with contextual feedback

Behavioral Insights

  • Retry Patterns: Analyzes retry frequency (>40% = high, <15% = low)
  • Mod Variety: Evaluates mod usage diversity
  • Map Exploration: Encourages playing different maps

Limitations

Important: osu!Skill is a supplementary tool, not a replacement for official rankings.

System Limitations

  • Data Dependency: Requires sufficient recent play data for accurate analysis
  • Skill Aspects: May not capture all skill nuances (reading, flow, etc.)
  • Mod Complexity: Simplified mod multipliers may not reflect true difficulty
  • Map Variety: Analysis quality depends on map diversity

Best Practices

  • Play regularly for accurate analysis
  • Play diverse maps and difficulties
  • Focus on improvement, not just the numbers
  • Use insights to identify improvement areas
  • Aim for 50%+ confidence for reliable results