How is a game's hype measured?

Each game gets an overall score: the sum of views across its counted videos — trailers and clips from official and media channels, plus creator buzz. Views are collected daily on YouTube and kept as history.

What's the difference between the "trailer" and "buzz" scores?

The trailer score sums views from official and media sources (publisher, platform, events, press). The buzz score sums views from creators outside the registry: faithful re-uploads, reactions, analyses. Together they form the overall score but are always shown separately.

How often are the numbers updated?

Once a day: a nightly collection refreshes the views of every tracked video, then the ranking and trends are recomputed.

Do fake videos count?

No. Fakes, fan-made videos, parodies and compilations are detected and excluded from the score. A human check can correct any classification, and that decision takes effect immediately.

Where does the data come from?

From public counters — today, the YouTube view counts of tracked videos. No personal data is collected. Other networks may be added tomorrow.

How do I read the 7-day trend (▲/▼)?

The badge compares a game's current position with its position 7 days ago, under constant rules; "+X views" shows the views gained over the same window. While the history is shorter than 7 days, the comparison starts at the first available measurement, and this is flagged.

How are debate pages produced?

From the public comments of tracked videos. An automated analysis identifies controversies, estimates the weight of each position in the discussion, and selects quotes verified word for word against the corpus. These pages are descriptive: they never influence the scores or the ranking.

Methodology

HypeTracker measures the hype of upcoming video games as the total measured views of the videos that carry each game on YouTube — official trailers and media coverage on one side, creator buzz on the other — collected daily, with full history kept.

Where the data comes from

View counts come from the official YouTube Data API, read from each video's public counter. Collection runs once a day for every tracked video; every reading is stored, never overwritten. No scraping of private data, no estimates: if a number is on the site, it existed on a public YouTube counter at a dated moment.

The two-score model

Each game has two scores, never blended. The trailer score sums the views of videos published by official and media sources: publishers, platforms, event channels, specialist press. The buzz score sums the views of community videos: creators, re-uploads, analyses, reactions. The overall score is their sum. A video is matched to a game by title (tolerant to accents and languages) and counted once; if it appears on both the official and community side, it is counted on the official side only.

What is excluded, and why

Fakes, fan-made trailers, parodies and AI recreations are excluded from all scores — hype for a video that isn't the game is not hype for the game. "All trailers" compilations are excluded to avoid double counting. Reaction and analysis videos never count as trailers; they feed the buzz signal only. Off-topic videos caught by title homonyms are filtered out automatically and on review.

Human corrections

Automatic classification is reviewed. A human verdict can correct any video's status at any time, takes effect immediately, and overrides automation permanently. The per-game audit trail shows every video with its status — counted or excluded — and the reason.

Comment analysis: summaries and debates

HypeTracker doesn't just measure views: the site also reads what the community writes. Every night, public comments from tracked videos are collected via the YouTube API. For each game discussed enough, a weekly summary is produced by a language model from that corpus: what excites people, what worries them, what divides them — illustrated with quotes taken word for word from real comments.

When a controversy grows — around one game or across several — it gets a debate page: the context, the question at stake, the positions involved with their estimated weight in the discussion, and arguments backed by quotes. Every quote is checked automatically: it must exist verbatim in the collected corpus, otherwise it is discarded.

These analyses are descriptive and never enter the scores or the ranking. The percentages on a debate page estimate the split of opinions among the comments analyzed — an honest order of magnitude, not a representative poll. Every page shows its corpus size and last-updated date, and a human review can correct or unpublish a page at any time.

Known limits

YouTube only, for now. Views on other platforms (TikTok, X, Twitch) are not measured; announced figures like "views across all platforms" are publisher statements, not comparable to our counts.
Views are not sales. Hype measures attention, not purchase intent.
View counters are YouTube's. We measure them faithfully; we do not audit how YouTube counts a view.
Coverage starts at tracking time. For games added late, buzz published before tracking is captured progressively, not exhaustively.
Engagement (like rates, comment volume) is shown as descriptive context and never enters the ranking.

Citing HypeTracker

Our figures may be reused with attribution: "Source: HypeTracker" with a link to the cited page. Every number on the site is dated; when quoting, include the date shown. Journalists can request charts, per-game time series or methodological detail via the contact page — usually answered within 24 hours.

Some game artworks are provided by the RAWG.io database.