Forecast review · 2026-10-20

The result.
Keep the full record.

The two original AI lines returned €0 in ordinary number prizes. Rewriting a rule after seeing the answer would not turn it into a successful prediction.

Official result · 2026-10-20

1 jackpot-winning line(s) · total jackpot paid €19,286,658

516323438

Our lines and actual matches

Method / evidence	Main numbers	Euro numbers	Matches	Gross prize
AI selection 1 Registered forecast log	1330384142	59	1+1	€0
AI selection 2 Registered forecast log	710114349	35	0+0	€0
Random control 1 Registered forecast log	3671142	28	0+1	€0
Random control 2 Registered forecast log	17152443	12	0+0	€0
History filter Reconstructed from the pre-draw rule and history; not a separately timestamped forecast	122293542	45	0+0	€0
Calendar hypothesis Reconstructed from the pre-draw rule and history; not a separately timestamped forecast	1112172130	16	0+0	€0

Outlined numbers matched. Prize amounts use the official euro table, not a guessed NOK exchange rate. Zero euro is also zero NOK. Personal and manually edited lines are unavailable without the numbers actually chosen. Reconstructed lines are not counted as registered forecasts.

What did the model select?

The model used history through 2026-10-13, as recorded when these lines were registered. Frequency and absence features gave small adjustments around equal weights. Weighted random sampling then selected the lines; it did not reliably identify the winning set.

The actual winning numbers had these model weights before the draw. Ranking them after the result is an explanation, not information we could use to know the answer in advance.

Winning number	Model weight	Rank	Uniform baseline
Main 5	9.962%	33 / 50	10.000%
Main 16	10.012%	24 / 50	10.000%
Main 32	10.149%	2 / 50	10.000%
Main 34	9.857%	47 / 50	10.000%
Main 38	10.119%	7 / 50	10.000%
Euro 8	16.071%	8 / 12	16.667%
Euro 9	14.850%	10 / 12	16.667%

Choosing precisely 5, 16, 32, 34, 38 + 8, 9 would have won. We have no pre-draw evidence that could reliably distinguish that one combination from the other valid combinations.

Would another rule have helped?

We replayed five fixed rules on the 200 draws before this result (2024-11-19 to 2026-10-16). Every replay uses only earlier information; the calendar rule also keeps its seven-day lag. These tests were designed after the archive was known and remain exploratory.

Replayed rule	Main + Euro hits	Mean hits / draw	This draw
Uniform replay	95 + 79	0.870	0+0
Highest last-50 frequency	97 + 76	0.865	0+0
Longest absence	89 + 66	0.775	0+0
Calendar day-root rule	112 + 72	0.920	0+0
Unseen main-set filter	104 + 66	0.850	0+0

The fair-draw expected mean is 0.833 individual-number matches per line. This is not a prize rate. Picking whichever historical rule looks best after many trials can fit chance patterns; these results do not establish an advantage for the next draw.

Current next selections, registered before the draw

Existing AI 1 and AI 2 remain exactly as originally registered. The two new uniform comparisons use a fresh recorded seed, sample five and two distinct numbers, and reject complete combinations already registered for the date. They do not favour the previous winners, hot numbers or supposedly overdue numbers.

Each valid line still has a jackpot chance of 1 in 139,838,160. More distinct lines cover more combinations and also cost more. These are optional comparison lines, not a recommendation to increase spending.

2026-10-23

AI selection 1

Registered 2026-10-16T18:15:03.463Z · history through 2026-10-16

2328424748

610

AI selection 2

Registered 2026-10-16T18:15:03.463Z · history through 2026-10-16

214151939

New uniform comparison 1

Registered 2026-10-20T22:28:36.440Z · history through 2026-10-20

112263436

New uniform comparison 2

Registered 2026-10-20T22:28:36.440Z · history through 2026-10-20

926314345

Frozen inverse quadratic

Registered 2026-10-20T22:47:53.610Z · history through 2026-10-16

534384248

2026-10-27

AI selection 1

Registered 2026-10-20T19:15:03.386Z · history through 2026-10-20

36161838

310

AI selection 2

Registered 2026-10-20T19:15:03.386Z · history through 2026-10-20

429364147

612

Open the current ticket →

All registered forward results

Date	Registered method	Matches	Hypothetical gross EUR
2026-10-20	AI selection 1	1+1	€0
2026-10-20	AI selection 2	0+0	€0
2026-10-20	Random control 1	0+1	€0
2026-10-20	Random control 2	0+0	€0
2026-10-16	AI selection 1	0+0	€0
2026-10-16	AI selection 2	0+1	€0
2026-10-16	Random control 1	2+0	€0
2026-10-16	Random control 2	0+1	€0
2026-10-13	AI selection 1	0+1	€0
2026-10-13	AI selection 2	1+2	€10
2026-10-13	Random control 1	1+0	€0
2026-10-13	Random control 2	0+0	€0

Only timestamped original forecasts are included. Retrospective reconstructions and personal lines are excluded. Amounts are hypothetical gross winnings before any ticket costs, not payouts received.

Official prize table

Rank	Winning lines	EUR per winning line
1	1	€19,286,658
2	0	No winning line
3	7	€112,637.4
4	21	€6,193.1
5	541	€300.4
6	966	€185.1
7	1,036	€125.5
8	13,916	€29.7
9	23,383	€19.8
10	48,792	€17.9
11	75,398	€14.5
12	341,775	€9.6

WestLotto official result files · ONCE dated result

Inspect the records

This analysis (JSON) · Original AI and random forecast ledger · New uniform comparison registration · Official results and prize-completion log

Registrations are self-published timestamps and source history, not independent third-party certification. Completing previously unknown official prize fields is logged; winning numbers and already-known prize values remain protected from silent changes.

The result.Keep the full record.

Official result · 2026-10-20

Our lines and actual matches

What did the model select?

Would another rule have helped?

Current next selections, registered before the draw

2026-10-23

AI selection 1

AI selection 2

New uniform comparison 1

New uniform comparison 2

Frozen inverse quadratic

2026-10-27

AI selection 1

AI selection 2

All registered forward results

Official prize table

Inspect the records

The result.
Keep the full record.