18+ National Responsible Gambling Programme 0800 006 008. Winners know when to stop. Get help
Beta Prices are read about every four minutes and published with the time we saw them. Real prices, read about every four minutes. Why?

Tell us what is wrong

The prices here are real and they are ours. We read the bookmakers on our board roughly every four minutes and publish what we actually saw, with the time we saw it. Nothing on this site is estimated, modelled or bought in. We hold no data agreement with any bookmaker, so a price can be a few minutes old and the bookmaker's own price is the one that counts. How live is this?

So why call it a demo? Because we are a data company, and data gets better with time. Every day we run, the price record gets longer, the licence register gets more complete and the ranking rests on more evidence. Some pages are still being written and a few bookmakers are not fully covered yet.

We would rather be useful now than perfect later. What is here already answers the question most punters never get answered: who is actually paying most, right now, and who is actually licensed to take your bet. Waiting another six months to publish that would help nobody.

Where we are unsure, we say so. Where we have not checked something, we do not claim it. If you find something wrong, tell us and we will fix it and say we did.

LicensedBets Biggest gap Live odds

UK49s history and statistics

UK49s hot and cold numbers

“Hot” means more frequent in past draws; “cold” means less frequent. Neither predicts the next result.

Showing the last 30 recorded Lunchtime draws. Change a control to inspect a different sample.

Your sample at a glance

30 lunchtime draws, 2026-08-08 to 2026-09-06. Six main balls only; Booster excluded.

Most frequent

23917323346

6 appearances each in 30 draws

Least frequent

414184041

1 appearance each in 30 draws

All tied numbers are shown. These are past counts, not recommended selections.

All 49 numbers, ordered by recorded frequency
RankNumberAppearances% of drawsRecorded draws ago
12620.0%3
13620.0%5
19620.0%1
117620.0%10
132620.0%4
133620.0%4
146620.0%0
85516.7%1
810516.7%0
812516.7%10
820516.7%8
821516.7%0
823516.7%0
830516.7%14
837516.7%3
168413.3%1
1611413.3%3
1613413.3%0
1616413.3%4
1619413.3%0
1624413.3%3
1627413.3%6
1628413.3%7
1638413.3%19
257310.0%18
2515310.0%13
2522310.0%6
2525310.0%9
2526310.0%5
2529310.0%7
2531310.0%9
2534310.0%1
2535310.0%9
2536310.0%2
2539310.0%2
2543310.0%9
2544310.0%4
2545310.0%6
2547310.0%2
2548310.0%7
2549310.0%3
42126.7%11
42626.7%19
424226.7%2
45413.3%8
451413.3%18
451813.3%29
454013.3%9
454113.3%8

Count check: 180 ball appearances across 30 draws. Number 1 appeared in 2 of 30 draws: 6.7% of draws, not a forecast probability. Missing dates are not filled with invented results.

How the counts work, probabilities and sources

By Licensed Bets editorial team · Method published 6 September 2026. Sample dates are shown above.

How to read this table

The percentage is appearances divided by the number of draws in the selected sample. Each draw contributes six main-ball appearances, so the main-ball column adds to 600%, not 100%, before rounding. In Booster mode there is one appearance per draw, so that column adds to 100% before rounding. Displayed percentages can add to slightly different totals in either mode.

“Recorded draws ago” counts backwards from the newest draw in this sample. Zero means the number appeared in that newest draw. “Not in sample” means we did not observe it in the selected window. It does not mean the number has never appeared, and it does not count a missing archive date as a draw.

Changing from 30 draws to 500 can change the ordering substantially. That is why a list of supposedly hot numbers without a sample size and date range is impossible to check. Equal counts share the same rank here; the ball number only determines display order within a tie.

Are cold or overdue numbers due to appear?

No. Under a fair, independent six-from-49 draw, a particular number has a 6/49 chance of appearing among the six main balls, about 12.24%. Its chance of being the single Booster is 1/49, about 2.04%. A long absence does not increase either probability on the next draw.

For example, the chance that a particular number misses ten consecutive main-ball draws is (43/49)10, about 27.1%. That sort of gap is not an extraordinary event requiring a correction. After the gap, its next-draw main-ball chance is still 6/49, assuming the draw is fair and independent.

Can these statistics produce UK49s predictions?

They describe the archive; they do not identify tomorrow’s result. Selecting a pattern after seeing the results and then showing how well it fits those same results is not a prediction test. Any claimed forecasting method needs a rule fixed in advance and results recorded on future draws it has not seen.

We do not sell predictions, rank “sure” picks, generate booking codes or label a number a banker. There is no guaranteed winning combination. If someone asks for a payment to unlock a supposedly certain result, a convincing-looking frequency chart is not proof that they know the draw.

Where the records come from

Our existing results archive supplies the table. Recent draws are compared with additional results publishers; older rows are archive imports. We validate the format, keep the main six separate from the Booster, and exclude conflicting duplicate date-and-draw records from these counts.

The archive’s source records come from LottoNumbers’ Lunchtime archive, its Teatime archive, National-Lottery.com and UK49s Latest Results. Licensed Bets is an independent publisher, not the draw operator. An archive can contain gaps or errors; send a correction if a dated draw is wrong.

For the current numbers, use Lunchtime results or Teatime results. For how fixed-odds number betting differs from buying a lottery ticket, read the South African UK49s guide.

A 500-draw window can cross calendar years. Earlier source archives: 2025 Lunchtime, 2025 Teatime, 2024 Lunchtime and 2024 Teatime. Calendar-valid future dates are excluded. Past dates are not proof of source agreement.