Methodology & Data Sources

Sources, lottery by lottery

No draw on this site is entered by hand. Each one is read from the source below by an automated collector and stored in a single database, which every page reads from. Where a source is an independent results site rather than the operator, it says so — the operator's own published result is always the authority for a prize claim.

Powerball

Draws from January 2016 to June 10, 2026 were imported from the New York State Gaming Commission's public Powerball dataset on data.ny.gov (New York is a Powerball member lottery). From June 13, 2026 each drawing is read from powerball.com, the game's official site. Jackpot amounts are only published there for recent drawings, so older rows carry none.

Mega Millions

Every draw is read from megamillions.com, the game's official site, through the same data endpoint its own results page uses. The site only publishes a jackpot figure for the latest drawing, so most rows carry none.

UK Lotto

Draws are read from lottery.co.uk, an independent results site that also publishes the full archive, and from the official National Lottery results feed on national-lottery.co.uk, which replaces the independent figure as soon as it publishes (usually a few hours after the draw). The official feed only ever carries the latest draw, which is why the history comes from lottery.co.uk.

EuroMillions

Every draw is read from euro-millions.com, an independent results site — not an official operator. Its results page shows the two newest draws in a separate block above the table, and our collector reads both so a draw is never skipped. A jackpot figure is published for every draw.

EuroJackpot

Every draw is read from euro-jackpot.net, an independent results site (the older eurojackpot.org address no longer serves results): its yearly archive pages for the history, and its live results page for the newest draws. Only the live page prints a jackpot, so recent draws carry one and older rows do not.

SuperEnalotto

Every draw is read from superenalotto.net, an independent results site. Its results list prints no year, so our collector takes the year from each draw's detail link. Jackpots appear only on individual draw pages, so they are collected for recent draws only.

Collection and updates

The collector runs twice a day, late in the evening and early in the morning (Paris time). European evening draws are normally collected the same night; the US games, drawn at 11 p.m. Eastern Time, are collected the following morning. Each run re-reads the most recent draws of every game, so a draw that a source published late, or corrected, is picked up on the next run.

Pages read the database through a cache that is refreshed every 15 minutes, and every statistics cache is keyed to the newest draw on record, so a new draw updates the results, history and statistics pages together. Each page states the newest draw it was built from ("Latest draw"). If a copy of a page is older than the data, the page says so and links to the current version, and if a game falls behind its own draw schedule, its pages say which draw is missing instead of presenting an old date as the latest.

If the database cannot be reached, the site falls back to a saved copy of the full archive and shows a notice at the top of every page saying how old that copy is.

Checks every draw must pass

• The right number of main and bonus numbers for the game.
• Every number inside the game's range, and no main number repeated within a draw.
• A plausible date: not before the archive begins, not in the future.
• One row per draw: the database keys every draw on lottery, date and round, so a draw collected twice is updated, never duplicated.

A draw that fails any check is skipped and logged, never guessed at. The database adds its own constraints on top — well-formed number lists, a valid round, one row per draw. Beyond that, we compare sources where two exist: UK Lotto's official feed replaces the independent figure for any draw it carries, and the imported Powerball archive was matched against powerball.com for the dates both publish.

Main numbers are displayed in ascending order, as operators publish them, whatever order a source listed them in. Bonus numbers keep their published order, which matters for SuperEnalotto, where the Jolly and the SuperStar share one field.

Format changes inside the archive

Four of the six games changed their number pools during the period we cover. Mixing formats would make every expectation wrong, so the statistics treat each change explicitly:

• Mega Millions — white balls 1–75 → 1–70 and Mega Ball 1–15 → 1–25 (October 31, 2017); Mega Ball 1–25 → 1–24 (April 8, 2025).
• EuroMillions — Lucky Stars 1–11 → 1–12 (September 27, 2016).
• EuroJackpot — Euro Numbers 1–10 → 1–12, and a second weekly draw (March 2022).
• UK Lotto — two result sets per draw night, stored as Round 1 and Round 2 (from June 10, 2026).

All-time frequency tables keep every draw, because they are a record of what happened. Numbers that can no longer be drawn (Mega Millions white balls 71–75, Mega Ball 25) stay in those tables but are left out of hot and cold rankings. Expected counts and randomness tests use only draws from the current format. Hot and cold rankings for UK Lotto use Round 1 only.

How each statistic is calculated

Throughout, d is the number of draws in the period, k the numbers drawn from a pool each time and n the size of that pool. Main numbers and each bonus pool are always counted separately.

Appearances and percentage of draws

Appearances are the number of draws a number came up in. Percentage of draws is appearances ÷ d. The first ranks numbers within one period; the second lets you compare periods of different length. In a fair draw every number has the same chance, p = k ÷ n, per draw — 5 ÷ 69 = 7.25% for a Powerball white ball, 6 ÷ 90 = 6.67% for a SuperEnalotto number.

Expected appearances and standard deviation

Over d draws a number is expected d × p times, with a standard deviation of √(d × p × (1 − p)). About two thirds of numbers should sit within one standard deviation of the expectation and about 95% within two. How far a number sits from its expectation, divided by the standard deviation, is its z-score.

Randomness test

To judge a whole pool at once we use a chi-square goodness-of-fit test. Because a draw takes k different numbers at once, counts are not independent of each other, so the usual statistic is adjusted: χ² = Σ(appearances − d p)² ÷ (d p (1 − p)) × (n − 1) ÷ n, compared with a chi-square distribution with n − 1 degrees of freedom. The p-value (computed with the Wilson–Hilferty approximation) is the chance a perfectly fair draw would look at least this uneven. A second test checks independence between draws: how many numbers each draw shares with the one before, against the hypergeometric distribution a memoryless draw produces.

Hot and cold numbers

Hot numbers are those with the most appearances in a window of the most recent draws (20, 50, 100 or 200); cold numbers have the fewest, or have gone the most draws since they last appeared. Ties are listed in number order. The chance of a particular number being absent for g draws in a row is (1 − p)^g, and a number is expected, on average, every 1 ÷ p draws.

The hot and cold backtest

We replay each archive in date order. Before every draw we rank the pool on the previous 20, 50 or 100 draws only — never on anything that came later — take the ten hottest and ten coldest numbers, and count how many the draw produced. Any fixed set of ten numbers is expected to match 10 × k ÷ n per draw; the difference is measured in standard errors of the hypergeometric distribution.

A draw in context

For the latest draw, the Results pages show the sum of its main numbers as a percentile of all current-format draws, how often its odd/even split has occurred, and how many numbers it repeated from the previous draw.

Jackpot odds

Computed from the game matrix: the number of ways to choose the main numbers, C(n, k), multiplied by the ways to choose the numbers of any second pool that is part of the jackpot. The six games, worked through.

Limitations

• Three of the six games are collected from independent results sites. We validate every draw, but a source error that still passes the checks (a wrong but valid number) could reach the site until it is reported or corrected at the source.
• Jackpot figures are incomplete for the US games, EuroJackpot and SuperEnalotto, whose sources only publish them for recent draws (the table above shows the share of rows that carry one). Prize tiers, winner counts and add-on games (Power Play, the Mega Millions multiplier, EuroMillions add-ons) are not collected.
• Dates are draw dates in the game's own time zone. A US draw at 11 p.m. Eastern Time is already the next day in Europe.
• Statistics describe past draws. No statistic on this site can predict a future draw, and none is presented as a way to improve your odds.

Found something that looks wrong? Email contact@lotterydatahub.com. Who we are and why the site exists: About LotteryDataHub.