Are Season Cycles Getting Shorter? The Direction Reverses by Game (as of July 2026)

Play a seasonal game long enough and you form an impression: seasons come around faster than they used to, or they have gone quiet lately. That impression almost always comes from one game — the one you actually play. So is there a shared direction once several games are held to the same measurement?

The answer is no. In ARPG Seasons data as of 19 July 2026, games whose recent cycles tightened and games whose cycles stretched split cleanly apart. There is no data support for the claim that the genre as a whole is moving one way.

One Trap First — a Falling Corpus Statistic Is Not a Trend

Something has to be addressed before the main argument. This update added three new games to our corpus, taking the number of computable intervals from 133 to 216. The corpus-wide median cycle duly fell from 15.0 weeks to 13.1, and the 25th percentile dropped from 13.0 weeks to 8.0 (56 days).

Writing 'seasons are getting dramatically shorter' off those two figures would produce an entirely wrong article. In the very same update, **not one of the ten previously tracked games saw its median change** — 84, 91, 98, 105, 112, 133, and 178 days, all exactly as before. No game shortened its cycle, yet the aggregate statistic fell anyway.

The cause is straightforward. Diablo Immortal joined carrying 50 intervals on its own, 48 of which sit between 26 and 30 days. That is 50 of 216 — a full quarter of the sample, all short, all from one game — which means the bottom quartile is now essentially that game and nothing else. The 25th percentile fell to 8.0 weeks because **who is in the sample** changed, not because the industry did.

That distinction is the discipline governing this entire article. Every trend discussed below is measured strictly as change over time *within a single game*. We do not pool different games into one distribution, watch the distribution move, and call the movement a trend — because in a pooled statistic, whichever game ships seasons most frequently dominates in proportion to its record count.

Method, and What We Excluded

For each game we sorted season intervals newest-first and compared the mean of the **four most recent** against the mean of **everything before them**. Four is the recent window because one or two observations let a single incidental delay flip the conclusion, while a wider window stops being recent.

One discipline matters here. A comparison only exists if the earlier segment also has samples. Path of Exile 2 and Last Epoch have just 4 observed intervals each — remove the recent four and nothing remains to compare against. Undecember's earlier segment holds only 2. Those three games are **in no state to support a trend claim** and are excluded from the table. Attaching a direction to insufficient samples is not analysis; it is invention.

The three newly tracked games — Diablo Immortal, Hero Siege, and Dwarven Realms — are absent from the table too, but for a different reason. Not a shortage of data: they were loaded too recently for us to have computed the recent-versus-prior split. Diablo Immortal in particular holds 50 intervals, which makes it the best-equipped candidate for trend analysis in the whole dataset, and it will join this table at the next update. Grim Dawn and No Rest for the Wicked, meanwhile, carry only expansion and update entries, so no season cycle exists to measure. Of the 13 games tracked, five can currently support a trend claim.

Recent Four vs Everything Prior — the Directions Diverge

In the table below, 'change' is the recent-four mean minus the prior-segment mean. A negative value means the cycle tightened.

Diablo IV — the Most Consistent Tightening

The cleanest signal in the table is Diablo IV: 88 days in the prior segment down to 70 across the recent four, an 18-day tightening. More telling than the delta is that the movement shows up as a year-on-year staircase rather than a single exception.

From 94 days in 2023 down to 56 in 2026, with no year rebounding. The 2026 figure rests on only 2 observed intervals and is provisional on its own. What can be stated firmly is the 2023–2025 run of 94 → 91 → 83 days, and that the 2026 value does not contradict it.

Diablo III — a Rhythm Change in a Long-Running Game

Diablo III carries 34 samples in its prior segment, making it the most reliable comparison here. It moved from 116 days to 96, a 20-day tightening.

Year by year, the inflection is sharp. Intervals exceeding five months — 158 days in 2022, 161 in 2023 — dropped abruptly to 96 in 2024, then settled around the 100-day mark at 105 in 2025 and 91 in 2026. A break that distinct in a game more than a decade into operation is uncommon and suggests something changed in how it is run at a particular point. Our data covers schedules only, so we cannot say from it what that something was.

Path of Exile — the Opposite Direction, and Why

Path of Exile runs the other way. Its prior 36 intervals averaged 109 days; the recent four average 182, a 73-day increase — the largest swing of any game in the table.

That number needs careful handling, though. The recent four have a **mean of 182 days but a median of 140**. A 42-day gap between mean and median says one of the four was exceptionally long and dragged the average up. The game's longest recorded interval is in fact 322 days, the figure identified in the volatility guide.

By year: 107 days in 2022 and 119 in 2023, spiking to 221 in 2024, then falling back to 133 in 2025 and 140 in 2026. The shape is less 'continuing to lengthen' and more **one large dislocation followed by re-stabilisation at a higher level than before**. The 2024 spike coincides with the period when development resources concentrated on Path of Exile 2. The subsequent 130–140 day range is clearly longer than the historical 100s, but it is not the 221-day outlier territory either.

In Summary — There Is No Genre-Wide Trend

Of the five games that can support a trend claim, two tightened, two were effectively flat, and one lengthened substantially. The remaining eight either lack the samples, were loaded too recently, or have no season cycle to measure at all. Writing 'ARPG seasons are trending shorter' from this position is a claim the data does not carry.

It is worth returning to the trap set out at the start. Over this same period our corpus-wide median fell from 15.0 weeks to 13.1, and the 25th percentile from 13.0 weeks to 8.0. Quote those figures alone and 'the genre is shortening' follows almost effortlessly. But that decline was the arrival of a single game in the sample, and no individual game moved in that direction. Whenever an aggregate statistic is cited, the question of when and how its sample composition changed has to be asked alongside it — which is why we publish sample counts with every update.

The practical takeaway is different. Estimating the next season from your game's long-run average will keep failing wherever the trend has shifted. Calculate Diablo IV off its old 88-day figure and you will guess late every time; calculate Path of Exile off its historical 109 and you will guess early every time. Reading the last few actual intervals alongside the average is considerably more accurate.

Every figure here is a 19 July 2026 snapshot computed from intervals between start dates, not end dates. Our data limitations and how inference is handled are documented on the methodology page.