Official results, transparent methods and a dated record of every server-generated selection. Explore patterns, then check what actually happens.
Pink ball: drawn bonus
First-prize winners reported by the official feed: 2. Prize per winner: NOK 6,759,250.
Official archive through 17 Oct 2026 · source fetched 2026-10-22T21:25:01.535166+00:00.
Choose 7 different numbers from 1–34. One bonus number is then drawn from the remaining 27 numbers. You do not choose an eighth number.
Loading registered selections…
Filled lines appear first; empty lines stay at the bottom. Changing or removing a line affects this browser view. Original registered forecasts remain in the results record. For a future draw, a registered line stays fixed even if another result arrives first.
The same personal year, month, day and life-path calculator as the EuroJackpot lab. Repeated values share a number; documented sums fill the remaining positions. Values are mapped cyclically to the game’s number range.
Birthday and name stay in browser memory and are cleared when the page reloads. They are not sent to our server. Personal selections are numerology experiments.
A small regularised logistic model learns how recent frequency, long-run frequency and time since the last appearance relate to earlier results. It converts these signals into number weights and samples a valid line. AI 2 uses the same model with a different fixed random seed.
The date and model version fix the random seed. AI 2 is another sample, not a second independently validated model.
Reduce the target day of the month to 1–9. Among the latest 400 earlier draws, count numbers from days with that same root. Add the equivalent of 20 draws of smoothed full-history frequencies, then sample without replacement.
Sample uniformly, rejecting an exact main-number set if it already appears in the available archive. This changes the filter, not the chance of any remaining individual combination.
Add all digits of the full target date and reduce to 1–9. For earlier dates with that root, find the most common main-number sum root (lowest root breaks a tie). Sample a uniform line until its main-number sum has that root.
Choose every main set with equal probability. This control is recorded alongside the other methods so comparisons include ordinary chance.
Each registered line includes its creation timestamp, target date, method version and last known draw date. Results are attached afterwards. Local edits and personal lines are not claimed as registered forecasts.
Inspect the forecast ledger ↗Select a number to inspect its count.
A digital root reduces a positive sum to 1–9. Example: 48 → 4 + 8 = 12 → 1 + 2 = 3. The date root uses every digit of YYYY-MM-DD. Main, extra and combined views belong to the same draw and are not independent evidence.
Mint: observed · Purple: exact fair-draw expectation. Root probabilities are calculated from all valid combinations, rather than assumed equal.
| Root | Observed | Expected |
|---|
Main sum root equals date root: 53 of 564. Fair-draw expectation: 62.67.
Combined sum root equals date root: 63 of 564. Fair-draw expectation: 62.67.
The most common main sum root is 4 (69 draws). Selecting the largest count after looking at many patterns can produce apparent relationships by chance. These counts do not establish a future advantage.
1: 63 · 2: 64 · 3: 62 · 4: 63 · 5: 69 · 6: 73 · 7: 75 · 8: 48 · 9: 47
The first 451 draws train the model; the last 113 (24 Aug 2024–17 Oct 2026) evaluate fixed learned coefficients. Rolling features use only results known before each prediction. The model is then refitted on available history for new selections.
| Brier score · lower is better | Model | Uniform |
|---|---|---|
| Main inclusion probabilities | 0.16351504 | 0.16349481 |
Brier score averages squared probability errors across numbers. It is not a jackpot hit rate. This is an exploratory retrospective check with no significance claim or independently validated advantage; the forward ledger below is the ongoing test.
C(34,7) equally likely valid combinations.
These are exact jackpot-match odds for distinct valid combinations in one fair draw. Duplicate lines do not add combinations. Smaller prizes are not included in this calculator.
We compare each method on the same future draws and retain misses as well as hits. A single successful line cannot establish a reliable method.
Exact day and month only. Years with no draw on that date are omitted.
| Date | Main numbers | Drawn Bonus |
|---|
0 repeated main-number sets in this archive. A past appearance does not make a set less likely next time.
No exact main-number set repeats in this available history.
| Draw date | Main numbers | Drawn Bonus |
|---|
Weekly official draws from 2 January 2016. This is not the complete history since the game began. 564 weekly draws verified without gaps from 2 Jan 2016 through 17 Oct 2026.
Choose 7 different numbers from 1–34. One bonus number is then drawn from the remaining 27 numbers. You do not choose an eighth number.
Results are checked hourly. Only finalised results are accepted. Conflicting winning numbers, gaps or invalid pools stop the update so the last verified snapshot remains available. Check Norsk Tipping before relying on a prize claim.
It uses the same learned probability weights as AI 1, with a different deterministic sample. Frequent numbers are a little more likely to be selected when the fitted weights favour them. The output remains a guess; two hits do not prove the method works.
The pools, schedules and jackpot odds differ. Lotto’s bonus is drawn, while Vikinglotto’s Viking number is chosen. Historical data and model results stay separate.
This lab treats numerology as an exploratory hypothesis. A fair lottery draw is independent of dates, birthdays and earlier results. The lab reports counts and keeps a forward record so any claim can be checked.