Season Cycle Volatility, Ranked — Which Schedules Are Easier to Predict (as of July 2026)

'How many months is this game's season cycle?' is the common question, but when you are actually planning around one, a different question matters more: how much can you trust that cycle? Two games with identical median cycles are entirely different propositions if one returns within a few days of schedule every time and the other slips by two months.

This article measures how widely each game's season intervals scatter, drawing on the 216 intervals that can be derived from the 13 games and 237 rows held in ARPG Seasons as of 19 July 2026, and ranks those games by schedule predictability.

One thing to flag before the table: this ranking has a different game at the top than our previous edition. Expanding from 10 tracked games to 13 introduced a title whose regularity has no counterpart in the earlier list.

Coefficient of Variation — Comparing Unlike Cycles Fairly

Comparing raw standard deviations is a trap, because longer cycles naturally produce larger deviations. Setting a 10.38-week deviation on a 27.3-week average beside a 2.73-week deviation on an 11.8-week average and declaring the first 'less stable' ignores the difference in scale.

So we use the coefficient of variation (CV): standard deviation divided by mean. It expresses wobble relative to the average, letting games with very different cycle lengths be judged on one scale. A CV of 0.20 means cycles typically move within roughly ±20% of the average. Lower is more regular.

As in the cycle-comparison guide, the measured quantity is the interval from one season start to the next. Each game's most recent season has no successor and therefore yields no interval, so it drops out naturally; the two titles with no season cycle at all (Grim Dawn and No Rest for the Wicked) are excluded from this work entirely.

The Predictability Ranking — a Runaway First Place

The table runs from lowest CV, meaning the most regular schedule, downward.

Why First Place Belongs in a Category of Its Own

Read the table literally and Diablo Immortal sits fifth at 0.272. Unfold its actual intervals, though, and a different picture appears: **48 of its 50 intervals fall between 26 and 30 days.** Nothing else on this list clusters that tightly.

So why does the CV climb to 0.272? One cause, and only one: **a single 84-day interval**. And that 84 is not a delayed season. It is exactly three times 28 — the run jumps from Season 49 to Season 52, with Seasons 50 and 51 missing from between. More precisely: both rows exist in our database, but we could not pin their start dates to a primary source, so they are **held unconfirmed and withheld from display.** This is a verification gate we did not clear, not a collection failure. Three 28-day cycles collapsed into one measured gap. The figure is registering a hole in our records, not irregularity in the game.

This is a structural limitation of the coefficient of variation as applied here. Because an interval is measured from one season start to the next, a single missing record fuses two cycles into one and manufactures a long gap that never happened. The tighter a game's real cadence, the more damage one hole does to its standard deviation.

So what we can state confidently about this game is not its CV but its modal band: **48 of 50 intervals between 26 and 30 days**. That is the fact, and it is also the useful part for planning. The conclusion that this is the one game on the table you can estimate to the day still holds — but it rests on that clustering, not on a rank.

Ignore the Sample Column and You Will Misread This Table

Do not read first-placed Path of Exile 2 as 'the most dependable game'. It rests on 4 intervals. With only four observations there has been limited opportunity to encounter a large disruption, and one heavily delayed season would send that CV climbing immediately. Last Epoch's 0.380 in eighth is provisional on the same grounds — also four samples — so its rank should not be treated as settled.

Weighted for sample size, the genuinely trustworthy conclusions sit mid-table. **Diablo III holds a CV of 0.254 across 38 intervals.** Sustaining that regularity over a long operating history is difficult to attribute to chance and reads instead as an established rhythm. Diablo IV's 0.231, drawn from 13 intervals, is likewise reasonably solid.

Conversely, it is worth noting that Path of Exile — holder of a large sample at 40 intervals — sits at a CV of 0.367. That is not a figure produced by a small-sample accident; league spacing genuinely has swung widely.

Undecember, last in the table at 0.492, comes with a story. While building this ranking we found a row in our own data whose start and end dates were identical — a recorded length of zero days. At first that looked like nothing worse than a polluted statistic. It was worse. Because the end date sat in the past, a season that was actually still running was being displayed on this site as **ended**. The cause: end dates for this game are derived from the following season's start, and the newest season has no follower, so its own start date got written into the end field. We have reset that row's end date to undetermined, the season now displays correctly, and the figures above are recomputed after the fix. Before the correction this game showed a CV of 0.684 — a number that described a defect in our pipeline, not a property of the game.

What Sits Behind the Numbers

Path of Exile's 0.367 connects directly to the 322-day maximum interval identified in the cycle guide. The shape is a game that holds a 98-day median and then opens up dramatically once, in a stretch that coincides with development resources concentrating on Path of Exile 2. Its volatility is therefore not the kind that wobbles a little each cycle, but the kind that stays regular and then slips heavily at a major transition.

The two newly listed games diverge in character. Dwarven Realms lands mid-table at 0.304, holding a 70-day median across 9 intervals but with a measured range running from 44 to 117 days. Hero Siege, at 0.483, is the most variable entry here — though that figure needs its context. The game ran no seasons for roughly a year after Season 18 in early 2023, resuming only with the 2.0 rebuild, and that single 353-day hiatus does most of the work in its standard deviation. What produced the gap was a suspension of service, not a failure of schedule discipline.

Torchlight: Infinite's 0.389 spans a similarly wide range (49–189 days) but comes from 14 intervals. Whether its volatility is concentrated in the past or persists into the present is something the cycle-trend guide separates out by period.

Applying This When You Plan

For low-CV games, median-based estimation is practically workable. Diablo Immortal is the only entry here precise enough to estimate to a specific date; for Diablo IV, adding about 84 days to the last start and treating a three-week window either side as the candidate range will usually serve.

For games above roughly 0.35, change the approach. Waiting on the studio's official announcement beats extrapolating a date from past cycles, and matching your schedule to the announcement costs less than clearing time speculatively in advance.

For the four-sample games — Path of Exile 2 and Last Epoch — do not use the CV itself as a planning input. That the sample is small is more informative than where the game happens to sit in the ranking.

Finally, to be explicit: this ranking is not an assessment of a game's quality or of how diligently it is run. Intervals widening while a major expansion is prepared is a normal part of development, and in several cases substantial content followed exactly such a gap. Nor does an extremely low CV imply that each season carries a lot of content. This table measures one thing only — how easy the schedule is to predict.