Cluster analysis splits Powerball numbers into groups with similar behavior — by draw frequency, depth, and average interval. Current "Hot stable" group: 3, 13, 14, 16, 17, 21, 26, 31, 32, 38, 44, 48 and 9 more (averages: frequency ×3, depth 5.7). Calculated for field 1. Data includes draw #1970 of 09.07.2026.
At the opposite pole — "Cold stable": 4, 7, 8, 9, 11, 15, 18, 22, 23, 33, 36, 40 and 3 more (averages: frequency ×0, depth 20). The largest group is "Warming up": 22 of 69 numbers.
Belonging to a group does not raise a number's odds — every draw is independent and any combination is equally likely. But as a selection system clusters do work: instead of comparing three tables you pick a ready-made group with the profile you want. Clicking a number in the table adds it to the method's generator.
A cluster is a group of numbers with similar statistical behavior. The k-means algorithm compares numbers by three features — draw frequency, depth (draws since the last appearance), and average interval between hits — and assigns each number to the nearest group center. The inputs are the same values as in the frequency table and hit intervals table.
d² = (f − f̄)² + (g − ḡ)² + (i − ī)² → min
A number joins the group whose center lies at the minimum distance d. All features are normalized to the 0–1 range:
- f — the number's draw frequency, f̄ — the group center's average frequency;
- g — depth (draws since the last appearance), ḡ — the center's average depth;
- i — average interval between hits, ī — the center's average interval.
Group names come from the center averages: frequency above 0.6 with depth below 0.4 — "hot stable"; frequency below 0.4 with depth above 0.6 — "cold stable"; rare but recent — "warming up"; frequent but long absent — "cooling down"; the rest — "medium".
Where to go next
Number Frequency
The first cluster feature: how many times each Powerball number was drawn — with a depth column.
OpenHit Intervals
The second and third features: each number's typical interval and its current depth vs the median.
OpenMoving Averages
Whether a number is warming up or cooling down over time — the same labels as clusters, draw by draw.
OpenHot Numbers
A ready-made list of the most frequent Powerball numbers — no cluster setup needed.
OpenCluster Analysis FAQ — Powerball
What is lottery number cluster analysis?
It is automatic grouping of numbers by similar behavior. The k-means algorithm compares every Powerball number by three features — draw frequency, depth, and average interval — and gathers similar numbers into groups. Unlike tables sorted by a single metric, a cluster reflects all three features at once.
What do the group names — "hot stable", "warming up" — mean?
The name comes from the group's averages. "Hot stable" — frequent and recently drawn numbers; "cold stable" — rare and long absent; "warming up" — rare overall but drawn recently; "cooling down" — frequent but long absent; "medium" — no pronounced profile. "Hot" and "cold" here are the same frequency-based notions as on the hot and cold numbers pages.
How many clusters should I choose?
Start with 4 — enough for most lotteries: two poles plus transitional groups. For fields with many balls (6 out of 45, keno with 80 numbers) try 5–6 for tighter, more uniform groups. The result also depends on the sample: over a short period the group composition can change noticeably from draw to draw — that is normal.
Can clusters predict the winning Powerball numbers?
No. A cluster describes a number's past behavior and has no effect on the next draw — balls do not remember which group they were assigned to. But the method works well as a selection system: instead of juggling three tables you pick a group with the profile you want — say, "hot stable" — and generate combinations from it right on this page.
How is a cluster different from a hot numbers list?
A hot numbers list is a sort by a single feature — frequency. A cluster reflects three features at once, so it reveals what a single table cannot: for example, a "warming up" number — rare overall but drawn recently. And the draw-by-draw dynamics of warming up are shown by moving averages.