# ARC Raiders activity evidence and Frozen Trail comparison method

Prepared by ARCRaiders.gg on September 18, 2026. This is a dated research snapshot and a proposed future measurement plan. It is not a forecast or an automated monitoring service.

## Observed monthly series

The monthly CSV transcribes Steam Charts' completed-month averages and peaks from November 2025 through August 2026. The source HTML was captured at 17:00:59 UTC on September 18, 2026. The chart displays January through August to examine the sustained 2026 decline. The CSV keeps the two earlier completed months for context. September and the overlapping rolling 30-day row are excluded from the completed-month series.

Source: [Steam Charts, app 1808500](https://steamcharts.com/app/1808500).

The provider reports average concurrent Steam players. These are averages of the number online at the same time, not unique users, accounts owned, downloads, copies sold, console players, or people who have permanently left. No claim is made about the provider's undocumented sampling precision or timezone for monthly boundaries.

January to August change is `(23,216.75 / 241,196.68 - 1) * 100`, or approximately −90.37%. The article rounds the decline to 90.4%. Values shown in the article table are rounded to whole players; the CSV preserves the source's two decimal places. Monthly change percentages are transcribed as reported rather than recalculated from already rounded values.

[Tracker Network](https://tracker.gg/population/steam/1808500) reports January and August averages of 241,308 and 23,264. Their decline is approximately 90.36%. This is a cross-check of two trackers drawing on the same underlying Steam platform, not two independent audience samples.

## Provider disagreements

[SteamDB](https://steamdb.info/app/1808500/charts/) lists an all-time peak of 481,966. Steam Charts lists 465,097. Tracker Network's monthly table includes a November peak of 481,966 while its headline record showed 481,318 during research. The provider comparison CSV retains these differences.

We did not identify a documented explanation that resolves every discrepancy. We neither average the records nor splice them into the monthly series. We do not use a missing or failed live counter as evidence of zero players. Steam Charts' directly fetched live counter was zero while other trackers showed activity, so it was excluded.

## Review and commercial data

The [English Steam storefront](https://store.steampowered.com/app/1808500/ARC_Raiders/) showed 49% positive recent reviews from 2,603 reviews and 81% positive lifetime English reviews from 199,381 reviews on the research date. Recent reviews are labelled for the last 30 days. Language and filter scopes are retained as displayed; the figures are not presented as equivalent cohorts or an all-player survey. These scores may change after publication.

[Nexon's Q2 2026 presentation, slide 3](https://investgame.net/wp-content/uploads/2026/08/2026-08-13-nexon_q2_2026_wp.pdf#page=3), reports more than 16.3 million cumulative units sold and ¥18.3 billion in quarterly ARC Raiders revenue. The document is a primary publisher presentation hosted on InvestGame. Cumulative sales and historic quarterly revenue do not measure September activity.

## October scorecard

Frozen Trail's [announced release date](https://arcraiders.com/news/frozen-trail-is-coming-october-8th) is October 8, 2026. All following windows are editorial choices, not benchmarks supplied by the developer.

| Window | Start, inclusive | End, inclusive | Days | Measure |
| :--- | :--- | :--- | ---: | :--- |
| Baseline | September 10 | October 7 | 28 | Mean concurrent Steam players |
| Launch | October 8 | October 8 | 1 | Maximum observed concurrent Steam count |
| First week after launch day | October 9 | October 15 | 7 | Mean concurrent Steam players |
| Longer window after launch day | October 9 | November 5 | 28 | Mean concurrent Steam players |

The baseline is incomplete on September 18. Every result cell is deliberately blank. Blank means unmeasured, not zero.

Before filling the sheet, obtain one provider's archived observations covering the baseline and later windows at the same sampling interval, with timestamped UTC observations. Verify historical access and coverage first. The current CSV supplies a template; it does not include those raw observations. A moving 30-day average cannot substitute for a fixed 28-day window. Monthly aggregates cannot reconstruct these windows.

Use UTC intervals from 00:00 on the start date to 00:00 on the day after the end date. For a complete equally spaced series, use its arithmetic mean. For a complete series with unequal observation spacing, use a time-weighted mean and document the duration represented by each observation. Do not interpolate through an outage or extend an observation across an undocumented gap. Use one sampling method across every window.

Record the provider, interval and observed coverage. If coverage is incomplete, leave the headline comparison blank or clearly label it provisional and provide the gaps. Do not silently compare complete and incomplete windows. Higher-frequency observations may capture higher peaks, which is another reason to keep provider and cadence fixed.

Calculate `activity_change_percent = (post_28_day_mean / baseline_28_day_mean - 1) * 100`. A zero or missing baseline makes this result undefined. Both 28-day windows contain four of every weekday. Launch day is excluded from the longer window so one exceptionally busy day does not dominate the comparison.

This measures concurrent activity, not cohort retention. Cohort retention would require following the same players across time. The scorecard cannot attribute a change solely to a map, progression system, marketing, a sale, or the return of Trials. The Expedition and Trials schedule also changes within the baseline period. Record those events when interpreting results.

The official Expedition announcement described an early-2027 plan. A [September 14 PC Gamer report](https://www.pcgamer.com/games/fps/arc-raiders-expeditions-may-not-come-back-maybe-well-scrap-the-whole-system/) quotes Grøndal's subsequent comments to PCGamesN that removing the system is also an option. That later uncertainty is retained in the article. Its final design and return schedule are not established here.

## Community and media research

Reddit threads were selected to identify questions and reported frustrations. This was not a random sample or sentiment poll. Reddit's research-script endpoint returned HTTP 403; accessible web-indexed thread pages were reviewed instead. Some search results were available only as excerpts and were kept out of detailed claims.

JxShepp's September 8 update video was reviewed using a supplied timestamped transcript, covering 0:00 to 10:25. The relevant passages are [0:54 to 1:24](https://www.youtube.com/watch?v=IYmotqxCm1o&t=54), where he discusses personal fatigue, the remaining audience and whether Frozen Trail can hold players' interest, and [9:28 to 9:44](https://www.youtube.com/watch?v=IYmotqxCm1o&t=568), where he discusses his motivation to make ARC Raiders videos. The transcript contains apparent speech-recognition errors. The article uses clear passages and does not turn uncertain game terms, numbers or patch details into facts. The transcript was read; the video was not independently watched or checked against its audio.

The two Loochy links are livestreams. Their titles, public dates and descriptions were checked, but their full discussions were not reviewed. They supply context about the wording of the debate, not a claim about Loochy's position. Creator views, channel size and video traffic are not player counts. One creator's experience cannot establish audience-wide sentiment or a reason for population decline.

The review did not find a current verified cross-platform daily active total or a public breakdown of queues by region, hour and party size. That is a limit of the reviewed public evidence, not a claim that such private data does not exist.

## Sources and reuse

Original charts, the comparison template and our data arrangement may be reused with credit to ARCRaiders.gg and the named underlying providers. Preserve the research date, metric labels and Steam-only scope. Source data remains attributable to its providers; this permission does not override their terms.

The hero image is official Embark artwork from the [Frozen Trail announcement](https://arcraiders.com/news/frozen-trail-is-coming-october-8th). Original asset: [official 1024 × 576 screenshot](https://assets.arcraiders.com/media/ArcRaiders_ScreenShot_16x9_04-1024x576.jpg). Copyright belongs to Embark Studios. It is excluded from the reusable evidence ZIP.

Two additional screenshots were captured from the [official September 18 teaser](https://x.com/ARCRaidersGame/status/2100978160364146717), at approximately 0:07 and 0:16. They show the wide mechanical structure and the first-look announcement card. Each screenshot is 557 × 314 pixels, cropped to the browser's displayed video frame without generated detail or rescaling. The small player labels count down the time remaining. Copyright belongs to Embark Studios. These images are also excluded from the reusable evidence ZIP and the original-data reuse permission.

The official post confirms the September 23 first-look premiere at 10 a.m. PDT, 1 p.m. EDT or 7 p.m. CEST. The footage is not used to confirm enemy names, combat mechanics, or datamined claims. The Gaming Merchant's reaction transcript supplied by Jim helped locate the source; the screenshot captions describe the directly inspected official clip.

Read the accompanying analysis: [Is ARC Raiders dead? The player decline and Frozen Trail’s October test](https://arcraiders.gg/blog/is-arc-raiders-dead/). Published September 18, 2026.