Schedule tracking2026-07-19

Measured Season Cycles Compared — a 28-to-178-Day Median Range (as of July 2026)

We tested the 'seasons run about three months' assumption against 237 records across the 13 games tracked as of 19 July 2026. Among the 11 games where a cycle could be computed, median cycles scatter from 28 days to 178. The corpus-wide median sits at 13.1 weeks (92 days), and every figure is published with its sample size and observation window.

Written by: ARPG Seasons Editorial TeamPublished 2026-07-19Last reviewed 2026-07-19

The editorial team that tracks the season, league, and ladder schedules of major action RPGs (ARPGs) such as Diablo, Path of Exile, and Last Epoch. We combine an automated pipeline that collects publishers' official announcements, patch notes, and RSS feeds with human review to cross-check each schedule against official sources and correct any errors.

See our editorial policy & team

Almost every article about ARPG seasons repeats the same line: 'seasons run on roughly a three-month cycle.' But which games is that based on, and across how many seasons? As of 19 July 2026, ARPG Seasons has accumulated 237 season, expansion, and update records across 13 games, each anchored to an official announcement. This article aggregates the 234 of those confirmed against a studio notice to find out where the conventional wisdom holds and where it breaks.

The short answer: 'three months' is true for some games only. Across the 11 games where a cycle could be computed, median cycles scatter from 28 days to 178 — and only four of them sit anywhere near the 90-day mark. Summarising the whole genre with a single number is not accurate.

Defining What We Actually Measured

Before the numbers, the definition matters. 'Cycle' here does not mean season length (end date minus start date). It means **the interval from one season's start date to the next season's start date**. The distinction exists because of data quality.

In our database, a substantial share of records carry an algorithmically inferred end date rather than an announced one. Studios announce a firm season end date less often than you might expect, so when a next-season announcement is confirmed we treat that moment as the prior season's close and backfill accordingly. Presenting that as 'season length' would mean passing inferred values off as fact. Start dates, by contrast, are almost always read directly from official announcements — which makes intervals between starts a far sturdier metric.

We also limited the aggregate to the 223 records typed as 'season'. Five entries classified as expansions (3 for Grim Dawn, 2 for Diablo Immortal) and six classified as major updates (4 for No Rest for the Wicked, 1 each for Dwarven Realms and Hero Siege) are different in kind and were excluded from cycle maths. That leaves Grim Dawn and No Rest for the Wicked without a single closed interval, so they drop out and 11 games appear in the table.

Median Cycles by Game — 28 Days to 178

Below are the intervals between consecutive season starts for 11 games. Read the median before the mean: occasional stretches where a season runs unusually long ahead of a major expansion drag the average upward. Path of Exile is the clearest case — a 98-day median against a 116-day mean, an 18-day gap.

Season cycle (start to next start) by game — 19 July 2026 snapshot, is_confirmed=true. Sample counts are intervals, so each runs one below that game's season count (the most recent season has no successor and yields no interval). Diablo Immortal's 84-day maximum is not a real hiatus: two seasons held unconfirmed drop out, fusing three 28-day cycles into one measured gap — the predictability guide covers that in detail.
Game	Median cycle	Mean	Shortest–longest	Samples	Observed over
Diablo Immortal	28 days	29 days	26–84 days	50	~4.1 years
Dwarven Realms	70 days	75 days	44–117 days	9	2024-08 to 2026-06
Diablo IV	84 days	83 days	48–112 days	13	~2.9 years
Torchlight: Infinite	91 days	91 days	49–189 days	14	~3.5 years
Hero Siege	120 days	145 days	70–353 days	25	2016-08 to 2026-07
Path of Exile	98 days	116 days	84–322 days	40	~12.7 years
Diablo II: Resurrected	105 days	114 days	78–161 days	13	~4.2 years
Diablo III	112 days	114 days	56–203 days	38	~11.9 years
Undecember	116 days	130 days	63–252 days	6	~2.1 years
Path of Exile 2	133 days	135 days	105–169 days	4	~1.6 years
Last Epoch	178 days	191 days	126–282 days	4	~2.1 years

Four Things Worth Reading Off the Table

First, only the four games in the middle of the table actually match the 'three months' claim. Diablo IV at 84 days, Torchlight: Infinite at 91, Path of Exile at 98, and Diablo II: Resurrected at 105 cluster around the 90-day mark. Both ends of the table sit well outside it: Last Epoch's 178 days is closer to six months, while Diablo Immortal at the opposite end turns over in under a month.

Second, Diablo Immortal's 28 days is a different kind of number. Of its 50 intervals, 48 land between 26 and 30 days — where the other games wobble by tens of days, this one effectively runs on a fixed clock. The 84-day maximum in the table looks like an exception but is not a real hiatus: two seasons held unconfirmed drop out, fusing three 28-day cycles into one. That a mobile live-service cadence differs from a PC seasonal one is not news — being able to demonstrate it from 50 observations rather than impression is.

Third, sample sizes are not comparable. Path of Exile's 40 intervals, Diablo III's 38, Diablo Immortal's 50, and Hero Siege's 25 are reasonably stable. Path of Exile 2 and Last Epoch rest on 4 intervals each, Undecember on 6, and Dwarven Realms on 9. A median drawn from single-digit observations can shift meaningfully on the strength of one or two upcoming seasons, so treat those as provisional readings rather than established rules. That is precisely why we publish sample counts alongside the medians instead of the medians alone.

Fourth, the shortest-to-longest spread reveals character that a median hides. Path of Exile's 322-day maximum is a fingerprint of the period when development resources shifted toward Path of Exile 2, and Torchlight: Infinite's 49-day minimum comes from a stretch of deliberately short back-to-back seasons. Hero Siege sits at a 120-day median until you notice a 353-day interval in the mix pulling its mean up to 145 — that gap is the real service hiatus between Season 18 in early 2023 and the 2.0 rebuild. By median alone these look like '90-to-100-day games', but lived experience is governed by the spread. We turn that spread into a ranked metric in the schedule-predictability guide.

How Far Corpus Statistics Move When Samples Arrive

This table is not fixed truth, and this update made that unusually visible. Before the three newly tracked games were loaded, our corpus held 133 computable intervals with an overall median of 15.0 weeks. Afterwards it holds 216, with a median of 13.1 weeks. Not one per-game median changed — yet the corpus median fell by two full weeks.

The 25th percentile is the more dramatic case: it dropped from 13.0 weeks to 8.0 weeks (56 days). Read alone, that looks like a surge in short seasons. It is not. Diablo Immortal arrived carrying 50 intervals — nearly a quarter of all 216 — and because every one of them lands near 28 days, that single game now occupies the bottom quartile by itself. This is a change in who is in the sample, not a change in the industry.

So we would encourage reading the per-game medians before the corpus-wide one. Merge several games into a single distribution and whichever game ships seasons most often dominates the statistics in proportion to its record count. The broader problem of mistaking a composition shift for a trend is treated at length in our season-cycle trend guide.

Putting the Numbers to Work

The most practical use is estimating roughly when the next season lands. Add a game's median cycle to its current season start: about 84 days for Diablo IV, about 112 for Diablo III. This is an extrapolation from past behaviour, not an announcement — always confirm the real date against the studio's own notice.

The second use is planning across games. If you play several ARPGs, mixing a short-cycle title with a long-cycle one distributes your time better. A game turning over every 28 days like Diablo Immortal rewards steady ongoing engagement; one spaced 178 days apart like Last Epoch lets you take your time without the next cycle crowding in. Diablo IV's 84 days sits between the two, suiting frequent but concentrated bursts.

Finally, these figures are a snapshot. They reflect 19 July 2026 and shift as each new season is confirmed — and, as this update showed, whenever a new game joins the dataset. How we collect the data and which values we flag as inferred is documented on our methodology page.