Methodology
V2 Scoring Engine
Every balance change is classified by category and magnitude before being weighted into a per-hero composite score.
Category Base Weights
| Category | Base Weight |
|---|---|
| Ultimate | 4 |
| Damage | 3 |
| Mobility | 3 |
| Cooldown | 2 |
| Health | 2 |
| Ability | 1 |
Magnitude Multipliers
Each change description is classified by magnitude:
- Minor (×1) — small numerical adjustments
- Moderate (×1.5) — meaningful changes
- Major (×2.5) — hero-defining overhauls
Final score: BaseWeight × Magnitude × (+1 buff / -1 nerf). Scores are rounded to one decimal place and capped at ±12 per hero.
Verdict Tiers
Each composite score maps to a verdict tier:
| Tier | Score Range |
|---|---|
| Major Winner | at least +8 |
| Moderate Winner | at least +4 |
| Slight Winner | above +2 |
| Stable Pick | at least -2 |
| Slight Loser | above -4 |
| Moderate Loser | above -8 |
| Major Loser | -8 or below |
Pick Labels
Verdict tiers translate to user-facing recommendation labels:
High Priority Pick — Major Winner
Strong Pick — Moderate Winner
Minor Gain — Slight Winner
Stable Pick — Neutral (±2)
Slight Drop — Slight Loser
Nerfed, Still Playable — Moderate/Major Loser with S/A meta tier
Avoid For Now — Moderate/Major Loser (non S/A)
Confidence Scoring
Every hero verdict carries a confidence level:
- HIGH — at least 3 changes, 3 categories, |score| above 10, not mixed
- MEDIUM — at least 2 changes, 2 categories, |score| above 5; or mixed with strong signal
- LOW — everything else
Mixed heroes (buffed and nerfed in the same patch) are capped at MEDIUM confidence.
Meta-Aware Adjustments
Patch impact contributes 65% of the recommendation. Meta tier context contributes 35%. S-tier and A-tier heroes that receive heavy nerfs are labeled Nerfed, Still Playable rather than Avoid For Now — their strong meta position means they remain viable despite balance changes.
Meta tier scores: S = 8, A = 5, B = 2, C = 0, D = -2, UNKNOWN = 0.
Trend Intelligence
Multi-patch history is analyzed to detect directional trends. A hero with consecutive buffs across 3 patches shows Improving (or Strongly Improving if the cumulative impact exceeds +15). A hero with consecutive nerfs shows Declining (or Strongly Declining below -15). One-patch heroes show Not enough data.
Trend confidence requires 5+ patches with clear direction for HIGH, 3+ for MEDIUM, and fewer for LOW.
Coverage Scoring
Coverage reflects how much data backs a recommendation:
- HIGH — 5+ patches tracked, meta context available
- MEDIUM — 2+ patches with meta or trend data
- LOW — limited or no patch history
Deterministic
All BuffTracker analysis is deterministic. No AI, no LLM calls, no machine learning. Every score, tier, label, and recommendation is computed from structured data using explicit, auditable rules.