How we estimate player counts

Last updated 2026-11-05. The longer story of why player counts are so hard to come by is in How we count MMO players when publishers won't say. This page is the reference for what the numbers on the site mean.

The short version

• One headline number per game:our central estimate of how many people play on a typical day (“~45K daily players”). The same number appears in the page title, the hero, the rankings and as the last point of the history chart.
• A range:roughly a 50% interval around the central estimate — about an even chance the true number is inside it. A lower bound shown as “<X” means the model can't rule out a very small population.
• A confidence label: high, medium or low, depending on what the estimate rests on (below).
• A date:“Last measured” is the day of the data the estimate came from, not the day the page was generated.

What we collect

Every day, for every game we track, we record what is publicly visible:

• Steam:the official peak concurrent player count for the day (Valve's API) and the number of reviews. When a game is on Steam this is the closest thing to ground truth that exists outside the studio.
• Discord: members and members online in the official server.
• Reddit: subreddit members and people online.
• Twitch:live viewers and streams in the game's category.
• YouTube and app stores: official channel subscribers and rating counts.

A reading of zero, or a scrape that failed, is treated as missing data — not as an empty game. X (Twitter) is no longer used: it stopped returning data in May 2026.

How the estimate is built: the basis

Each estimate says what it is anchored on:

• Steam.A full day of Steam's peak concurrent count is converted to daily players — over a day, several times as many people play as are ever online at once. For a game played mainly through Steam, this is close to a measurement. Many MMOs also have their own launcher, console or mobile versions; for those, the Steam reading is treated as a lower bound rather than the whole population, and community signals fill in the rest.
• Community.For games with no usable Steam data, the estimate is modelled from community size. We fit how community signals (Discord, Reddit, Twitch, YouTube, app stores) relate to measured Steam numbers across hundreds of games, then apply that relationship to games we can't measure directly. Several independent signals are combined, with the more reliable ones weighted more. This is much looser than a measurement, which is why these estimates carry low confidence.
• None.Not enough data to say anything. The page shows “No estimate yet” instead of a guess.

Confidence levels

• High: measured — a full day of Steam data for a game that is played mainly on Steam.
• Medium: Steam data exists but covers only part of the player base, or several independent community signals agree.
• Low: modelled from community size alone. Useful for comparing games and following trends; treat the absolute number as an order of magnitude.

Freshness

• Every number shows the date of its data (“Last measured”). Titles and descriptions use that date too, never today's.
• An estimate whose data is more than 30 days old is withdrawn: the page shows no number rather than an old one, and the game drops out of the rankings until fresh data arrives.
• “Updated daily” appears only while the data behind a number is at most 2 days old.

Weekly and monthly changes

• The weekly change compares the daily estimate with the one from about a week earlier, using only the signals present on both days — so a source coming back online is never reported as players arriving. Without a valid week-old baseline it shows “—”, never “0%”. Changes are capped at ±999%.
• Community trends on game pages are labelled “past week” only when the comparison point is 5–10 days old, and “past month” only when it is 21–45 days old.
• The monthly table averages the daily estimate per calendar month, and compares months using only signals present in both. It appears once a game has two calendar months of history.

History

We restarted data collection on Nov 5, 2026, after an audit found old numbers presented as current and a four-month gap in the record. Earlier history and estimates were discarded rather than mixed with the new series. Charts start on that date and break wherever a day is missing instead of drawing a line across it.

Only the latest estimate is stored, so the history chart re-runs the model on each day's signals and is scaled so that its last point equals the published estimate. It shows how the estimate moved, not a second competing figure.

Is it dead? Release status

Release dates come from IGDB, and a past release date says nothing about whether a game is still running. So:

• We say a game has shut down only when that has been verified by hand, with the date we checked and, where we have one, a link to the announcement.
• We say a game is online only when it has a current estimate of at least 100daily players. Otherwise we say what we know — “released”, “release status unknown” — without claiming it is alive or dead.
• “Upcoming” is used only for games whose announced release date is still in the future.

Known limitations

• Regional gameswhose communities live on platforms we don't read (Naver, QQ, KakaoTalk) look smaller than they are.
• Very watchable games can look bigger during Twitch drop campaigns and streamer events.
• Private and emulator servers are invisible to us.
• Games sharing a community (editions, regional versions, remasters on one Discord) can share signals; rankings show each shared community once.

Because every game is measured the same way every day, trends and comparisons between similar games are more reliable than any single absolute number.

Changes to the method

• Nov 5, 2026: fresh start. One central estimate per game with a range and confidence label; Steam treated as a lower bound for games not played mainly on Steam; estimates expire after 30 days; release status verified by hand for shutdowns; history restarted.
• August 2026: estimates computed from collected data replaced the earlier AI-written estimates.

Spotted something wrong? See our corrections policy. Or start with the most played MMOs.