Methodology

Overview

SnapPick's recommendation engine is a deterministic, rule-based scoring system. It is not a neural network or a generative AI model. Given identical inputs, it always produces identical outputs — which means every recommendation is auditable and reproducible.

This page explains, in plain terms, how the engine works for each major feature. For the specific data sources it relies on, see the Data Sources page.

Brawl Stars — Brawler Pick Recommendation

When you open the Brawl Stars picker and select a map and mode, the engine scores every brawler you own against a composite formula:

• Win rate on this map and mode — the primary signal. Sourced from aggregated battle data and community-verified statistics. Higher win rate = higher base score.
• Meta presence — how frequently this brawler is picked at high trophy levels on this map. High pick rate at high trophies is a proxy for community-validated strength, weighted less heavily than win rate to avoid a popularity bias.
• Ownership and mastery — brawlers you own at higher power levels receive a positive adjustment. A brawler you have at Power 11 will score higher than one you own at Power 1, reflecting real-world skill and gear advantage.
• Mode-specific role fit — each game mode rewards different roles (e.g. tanks in Brawl Ball, snipers in Bounty). Brawlers whose primary role fits the mode receive a bonus; those with a poor role fit receive a penalty.
• Gadget and Star Power readiness — having the meta-relevant gadget or star power unlocked for a brawler contributes a smaller positive adjustment.

The final score is a weighted sum of these factors. Weights are tuned manually by the team after each major balance patch, with reference to top-trophy win-rate data.

Clash Royale — Deck Recommendation

Deck recommendations combine your card collection with archetype matching and meta performance:

• Archetype selection — the engine first identifies which archetypes (e.g. Hog 2.6, Log Bait, Miner Poison) are strongest in the current meta, ranked by ladder win rate.
• Card ownership fit — each candidate deck is scored against your collection. A deck you can play fully owned scores higher than one requiring cards you don't have. Missing cards trigger an upgrade cost estimate.
• Card level advantage — decks where your cards are closer to the tournament standard level receive a positive adjustment, reflecting a real in-game advantage.
• Archetype preference — if you have expressed a preferred playstyle (e.g. cycle, beatdown, control), decks matching that archetype receive a bonus over equally performing alternatives.
• Synergy score — cards with established synergy pairs (win conditions + support cards) score higher than arbitrary card combinations at the same average cost.

Clash Royale — Upgrade Priority

The upgrade engine answers: "given my current collection and resources, which cards should I upgrade first?"

• Meta weight — cards that appear in S-tier decks score higher. Cards that appear in multiple strong archetypes (versatile win conditions) score highest.
• Current level gap — the difference between your card's current level and the tournament standard level. Cards with a larger gap that are also meta-relevant score highest because upgrading them gives the most competitive return.
• Cost efficiency — cards requiring fewer gold and copies to reach the next meaningful level receive a bonus. The engine shows the exact cost to the next level for every card.
• Deck switch potential — if upgrading a card would unlock a new, stronger deck you cannot currently play, the engine surfaces this as a "deck switch opportunity" above the standard upgrade ranking.

Results are bucketed into Upgrade Now (affordable, high impact), Focus Next (high impact, needs more resources), and Long Term (worthwhile but not urgent).

Tier Lists

Tier lists on SnapPick are not crowd-voted or opinion-based. They are derived programmatically from the same win-rate and meta data that feeds the recommendation engine:

• S tier — brawlers or cards with above-average win rates AND high meta presence at high trophy/ladder levels.
• A tier — strong performers with either high win rate or high meta presence, but not necessarily both.
• B tier — competent options with average win rates; situationally strong in specific modes or maps.
• C tier and below — below-average win rates or negligible meta presence at the current patch level.

Tier cutoffs use absolute thresholds, not relative percentile ranking — so it is possible for an entire tier to be empty after a major buff wave raises most brawlers' win rates above the previous cutoff.

Limitations and Caveats

• Patch lag. Balance patches take 24–72 hours to propagate through our data sources as the new meta stabilises in ladder play. Recommendations during this window reflect the pre-patch meta with a stale-data warning where possible.
• Low-trophy calibration. Win-rate data is most reliable at mid-to-high trophy levels. Recommendations for very low-trophy players may over-index on high-skill brawlers or cards that are harder to execute.
• Playstyle mismatch. The engine optimises for win rate, not enjoyment. A brawler that suits your personal playstyle but ranks B-tier may be the right choice for you. Use recommendations as a starting point, not a mandate.