Guide and methodology

Reading results and win rates

Search for your profile and separate “Last 20 matches” from “Recorded career.” Career records cover the matches stored by Deadlock API, not necessarily every game you have played. Missing matches should not be interpreted as losses or zero-valued performances.

A 12–8 record is evidence, not a settled skill rating

In a hypothetical example, 12 wins in 20 games gives 12 ÷ 20 × 100 = 60%. One more loss changes that to 12 ÷ 21 ≈ 57.1%. A single result moves a small sample substantially. Comparing 60% over 20 games with 55% over 200 games solely by rate overlooks that uncertainty. These are illustrative numbers, not a real player's record.

To review form, compare the same hero, a similar period and the same mode. Open individual losses and inspect economy and purchases as well. K/D/A alone cannot establish objective contribution or explain why a team lost.

The regional leaderboard covers the selected region. Rankings and profiles are saved daily; we show the latest saved snapshot from before today in Japan Standard Time. The displayed timestamp is the actual capture time. Failed updates keep the previous snapshot, and the first snapshot is shown once its capture day has ended. Hero top-player lists do not apply a region filter, so they describe a different population. An unavailable rank does not mean a low rank.

Comparing top players and builds

The hero directory selects players with at least 20 games on that hero in the last 10 days. It fetches up to 40 candidates from the API's win-rate ordering, then uses rank and other metrics to display 20. This is a selected candidate list, not an exhaustive world ranking.

Individual purchase paths sample up to 20 matches. The pooled view reads up to 8 games each from 20 players. Missing matches or purchase records reduce the actual sample. Branch shares refer to games following the same preceding purchase sequence; small branches are folded away.

Purchase rate is not a recommendation score

Popularity does not prove that an item causes wins. Expensive items may be affordable mainly in games already going well, and defensive purchases depend on opponents. Compare two early purchase paths, inspect the number of matches supporting each fork, and use match details to check economy and opposition at the time.

Trends compare the latest 7 days with the preceding 7, using a floor of 30 matches per window when identifying movement. Patches and changes in the sampled population can shift these figures. A rising share alone does not prove an item became stronger.

Reviewing featured matches

Featured matches are automatically selected from public high-rank games in a UTC calendar day, not a local-time day. The query is capped at 10,000 matches. It is neither exhaustive coverage nor a human review of replays.

Metrics are evaluated relative to that day's sample. The combined score weights rank at 30%, kills at 25%, closeness of final souls at 20%, duration at 15% and evenness of souls across players at 10%. “Best” means selected by that metric, not a guarantee of competitive quality. A close final economy can still follow a large lead earlier in the game.

Open a match, locate where the economy gap widened, and compare the items held at that moment. Coinciding changes do not establish causation. Positioning and decision quality are not fully visible in these numbers; compare with an in-game replay when available.

Damage example and model limits

In the damage calculator, choose a hero, attack type and level, then set target health and defenses before adding items. Keep the target fixed while changing one item or buff at a time. Enable conditional effects only for a situation where they are active.

A comparison that includes defenses

Set base damage to 100, flat bonus and scaling to zero, weapon damage to +20%, headshot multiplier to 1, amplification to 0%, target resistance to 25%, resistance reduction to 10 points, other reduction to 0% and barrier to 0. This model gives effective resistance of 25 − 10 = 15% and damage of 100 × 1.20 × 0.85 = 102 per hit. A 500-health target takes 5 hits, rounding up. Raising weapon damage to +30% gives 110.5 damage but still takes 5 hits. More damage does not always reduce the hit count.

This is an example of this site's model, not a guarantee of game behavior. Resistance reductions are added, with effective resistance clamped to −100% through 95%. DPS is damage per hit multiplied by effective hits per second; it does not automatically simulate reloads, accuracy, range falloff or healing during combat. Melee is treated as a single hit. Hits to kill assumes constant damage against health plus barrier.

Some item effects are unsupported or only partially modeled. Read the effect warnings and calculation breakdown, and verify special interactions or post-update changes in game. An upstream asset update does not mean this site's calculation code has been updated at the same time.

Sources, updates and corrections

Profiles, matches and assets come from Deadlock API. This site derives display statistics, organizes purchase paths, scores featured matches and calculates damage. A fetch timestamp is not the match completion time: upstream processing and caches introduce delays. If data is missing, check the period, identifier and region, then try again later.

The sections above explain the selection and calculation conditions. Send corrections through contact with the page URL, input conditions, expected value and actual value.

As explained in our site information, this is not an official Valve service. Worked examples are illustrative, not claims of measured in-game results.