Before each assessed draw, nine fixed rules select the same number of main numbers using only earlier results. We then count the overlap with the real result. These are explicit selection rules based on analyses, not a claim that every statistic predicts numbers.
The target draw and all later draws are excluded from every calculation. No future selection or recommended numbers are generated. Additional pools are assessed separately below where available.
Multi Multi pilot — recorded before the draw
One frozen method and window, selecting 10 numbers for each of the next 100 recorded draws. We compare them with a separate random set of 10 numbers recorded at the same time. The target result is excluded from selection.
How do I choose a saved set and check its hits?
The server pilot does not require you to choose numbers or press a save button. It records the method’s set and a random set before the draw. Changing the analysis form does not change this pilot’s settings.
Expand “Recorded selections and subsequent hits”. Choose the card for the date and time you want to check. Times use Polish local time; the 14:00 and 22:00 draws are separate results.
Check the pre-draw recording time and the dates of the data used for selection. The record must precede the assessed draw, and the last result used for calculation must also be earlier. Note the ten method numbers; the second row is a separate random control set.
Once the result has been published, reopen the page and check the server’s latest check time. Green balls show matches. In the method/random hits row, the first count belongs to the method and the second to the random set. A pending result or missing data does not mean zero hits.
To check manually, open the Multi Multi result for the same date and time. Count how many of the ten recorded numbers appear among the twenty drawn numbers. Count each number once, with no extra hit for Plus. Repeat separately for the random set. Both counts should match the card.
On the next draw’s card, check the new recording time and the updated data period. The method, window length and number of selections must remain the same. The selected numbers may change because the rule now uses more recent results.
Below the chosen row, click “Add the method set to My selections” or “Add the random set to My selections”. A form opens with those ten numbers. Name the set and confirm saving. After the draw, open My selections: the page automatically compares your saved numbers with the latest available Multi Multi result. Check the result date and time — the assessment for a specific pilot draw remains in its card here.
A saved selection keeps fixed numbers in this browser; it does not adopt new pilot sets or check results in the background while the page is closed. The pilot evaluates its own records on the server. Saving a selection does not purchase a lottery entry.
The model average of 2.50 is the expected number of matches for a random set of 10 numbers in a 20-from-80 draw. It is not the score of a particular set. One better result does not establish an advantage; compare all assessed steps. Hits are not profit — this panel does not calculate costs or payouts.
How do I save my own method to check later?
This is the separate “Track a saved method” feature below the pilot. It stores settings in this browser and does not start a server task.
In the analysis settings, choose the rule, training draws per step, number of selections and session. Click “Prepare tracking with these settings”. Alternatively, use the automatic method and window selection button to compare available settings on historical data.
After the page reloads, review the prepared selection: method, window and numbers. Before a future draw, click “Save fixed method”. This saves the rule and its parameters; it does not mean using the same numbers forever or purchasing a ticket.
After the draw, return in the same browser and click “Open and check subsequent draws” for your observation. Check the date, green matches and record type. Steps between visits may be labelled as reconstructed afterwards; they are not proof that those numbers were saved before the draw.
Do not change a saved observation’s parameters to improve its past score. Save different settings as a separate observation. Here the random comparison is a model distribution; the server pilot above shows an actual random set recorded before each draw.
Method / training draws / selected numbers / session: Number intervals / 50 / 10 / Both / all
Saved at: · Last server check: 18.11.2026 01:11:02
Assessed draws: 11 / 100 · Draws without a prior record: 0
Total hits: method / random set: 30 / 26. Average hits: method / random set: 2.73 / 2.36. Random model: an average of 2.50 hits for a set of 10 numbers.
Recorded selections and subsequent hits
18.11.2026 14:00
Saved before the draw: 17.11.2026 22:07:17 · Data used for selection: 24.10.2026 14:00 — 17.11.2026 22:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
5
9
16
32
49
52
53
55
59
71
Waiting for a draw later than the time this method was saved.
17.11.2026 22:00
Saved before the draw: 17.11.2026 14:12:03 · Data used for selection: 23.10.2026 22:00 — 17.11.2026 14:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
9
13
32
42
48
63
67
70
75
79
Hits: method / random set: 4 / 10 · 1 / 10
17.11.2026 14:00 · #17152
Saved before the draw: 16.11.2026 22:07:05 · Data used for selection: 23.10.2026 14:00 — 16.11.2026 22:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
1
22
27
45
48
49
57
63
65
66
Hits: method / random set: 3 / 10 · 2 / 10
16.11.2026 22:00 · #17151
Saved before the draw: 16.11.2026 14:12:10 · Data used for selection: 22.10.2026 22:00 — 16.11.2026 14:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
1
3
5
6
18
46
47
49
55
56
Hits: method / random set: 4 / 10 · 2 / 10
16.11.2026 14:00 · #17150
Saved before the draw: 15.11.2026 22:07:03 · Data used for selection: 22.10.2026 14:00 — 15.11.2026 22:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
4
9
15
20
35
55
66
70
75
80
Hits: method / random set: 2 / 10 · 3 / 10
15.11.2026 22:00 · #17149
Saved before the draw: 15.11.2026 14:11:59 · Data used for selection: 21.10.2026 22:00 — 15.11.2026 14:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
12
18
24
28
41
47
66
71
75
78
Hits: method / random set: 4 / 10 · 0 / 10
15.11.2026 14:00 · #17148
Saved before the draw: 14.11.2026 22:08:16 · Data used for selection: 21.10.2026 14:00 — 14.11.2026 22:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
2
13
16
23
29
31
32
46
53
60
Hits: method / random set: 4 / 10 · 5 / 10
14.11.2026 22:00 · #17147
Saved before the draw: 14.11.2026 14:12:05 · Data used for selection: 20.10.2026 22:00 — 14.11.2026 14:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
8
12
27
31
34
35
37
38
73
80
Hits: method / random set: 1 / 10 · 2 / 10
14.11.2026 14:00 · #17146
Saved before the draw: 13.11.2026 22:07:02 · Data used for selection: 20.10.2026 14:00 — 13.11.2026 22:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
2
8
9
14
22
53
57
58
66
77
Hits: method / random set: 5 / 10 · 2 / 10
13.11.2026 22:00 · #17145
Saved before the draw: 13.11.2026 14:12:03 · Data used for selection: 19.10.2026 22:00 — 13.11.2026 14:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
16
29
32
33
34
62
65
71
73
78
Hits: method / random set: 0 / 10 · 2 / 10
13.11.2026 14:00 · #17144
Saved before the draw: 12.11.2026 22:07:32 · Data used for selection: 19.10.2026 14:00 — 12.11.2026 22:00
Method numbers
31
32
33
34
35
36
37
38
39
40
Random control set
1
15
17
23
26
37
44
61
66
79
Hits: method / random set: 1 / 10 · 5 / 10
12.11.2026 22:00 · #17143
Saved before the draw: 12.11.2026 16:42:03 · Data used for selection: 18.10.2026 22:00 — 12.11.2026 14:00
A saved selection keeps fixed numbers in this browser; it does not adopt new pilot sets or check results in the background while the page is closed. The pilot evaluates its own records on the server. Saving a selection does not purchase a lottery entry.
The method and window are selected once: 20 rules × windows of 7, 14, 25, 50, 100 and 250 draws, evaluated on the same 20 earlier draws. Ties use the shorter window, then the fixed rule order. The rule then stays unchanged while its window moves with new results. Recording closes five minutes before 14:00 or 22:00 Warsaw time. Missed draws are never backfilled as predictions. Green balls indicate hits, which are also counted in text. Only main numbers are evaluated, excluding Plus, costs and prizes. This is a descriptive comparison, not evidence of an advantage or a promise of winning. Timestamps come from our server, not an independent timestamping authority. The page may be cached; its last server check is shown above.
Track a saved method
Save the rule, training window, number of selections and draw session. These parameters stay fixed. Return and open the observation to assess subsequent available draws.
This record stays in this browser. It is not a certified timestamp or a purchased ticket. The rule selects new numbers from the preceding window for each assessed draw. Steps between visits are reconstructed afterwards using the saved rule. Only main-number hits are assessed, not payouts. This is not evidence of an advantage or a prediction.
Automatic selection compares 20 rules on the same 20 earlier draws. Windows: 7, 14, 25, 50, 100, 250 — only those fully available. Ties: shorter window, then fixed rule order. Selection may fit past noise; evaluation starts only after saving.
Tracking covers available published results. Missing publication is not a zero-hit result. It does not run while the page is closed. Changing settings creates a separate observation.
Actual history: 50 · 24.10.2026 14:00 — 17.11.2026 22:00. Assessed steps: 25 (exploration 15, later assessment 10).
Test a method on subsequent draws
We select the method by total hits in the earlier period. Later draws are excluded from selection. All tied leaders are shown. The training window and number of selected numbers stay fixed; selections are recalculated before each draw using earlier data only.
Selection period ends: 12.11.2026 22:00 · Later test starts: 13.11.2026 14:00
Method selected earlier
Mean hits before testing
Mean hits during testing
Hits / test draws
Most frequent
5.400
5.200
52 / 10
Changing settings after viewing the test creates another historical comparison, not independent confirmation. This result does not establish an advantage in future draws. This test covers main numbers and does not report payouts.
What matched the latest assessed draw?
17.11.2026 22:00 · #—
1
2
5
8
13
14
18
22
30
35
36
38
39
49
51
60
66
73
78
80
Calculations used earlier draws only: #17128 — #17152, 05.11.2026 14:00 — 17.11.2026 14:00.
Rule
Earlier selection; ✓ marks overlap
Hits
Most frequent
1
✓
3
7
13
✓
18
✓
19
22
✓
26
31
33
37
38
✓
39
✓
42
44
55
57
62
72
77
6 / 20
Least frequent
5
✓
8
✓
11
12
16
17
20
24
28
40
41
50
52
58
60
✓
61
70
71
73
✓
78
✓
5 / 20
Longest gaps
4
5
✓
6
11
16
17
20
21
27
34
44
47
48
49
✓
56
60
✓
61
62
76
78
✓
4 / 20
Repeats
3
8
✓
12
15
18
✓
19
22
✓
28
31
32
33
43
58
59
65
68
71
72
77
80
✓
4 / 20
Partners
1
✓
3
15
18
✓
19
22
✓
26
31
32
33
37
44
57
59
65
68
72
74
77
80
✓
4 / 20
Binary rhythm (2 bits)
1
✓
3
7
19
22
✓
23
26
33
37
39
✓
42
53
56
62
66
✓
68
69
72
77
80
✓
5 / 20Earlier pattern examples: 1–12; selections with no example: 0.
Binary rhythm (3 bits)
1
✓
3
7
14
✓
19
26
30
✓
35
✓
37
38
✓
39
✓
44
53
57
66
✓
69
72
77
79
80
✓
8 / 20Earlier pattern examples: 1–7; selections with no example: 0.
Binary rhythm (4 bits)
1
✓
2
✓
5
✓
7
13
✓
14
✓
18
✓
21
30
✓
37
38
✓
42
49
✓
53
57
69
75
77
79
80
✓
10 / 20Earlier pattern examples: 1–9; selections with no example: 0.
Binary rhythm (5 bits)
1
✓
2
✓
3
5
✓
7
8
✓
10
14
✓
16
17
19
21
30
✓
37
42
49
✓
53
66
✓
77
79
8 / 20Earlier pattern examples: 0–6; selections with no example: 6.
More overlaps in one draw describe that draw. They do not establish a dependable method.
Which rules matched more often in this history?
Earlier steps select the method. Their leaders are marked in the table; the ranking uses only the later assessment steps. The default split is 60% / 40%; you can set the number of final test draws. All rules use the same draws and selection size.
Random-model expectation: 20 × 20 / 80 = 5.000 hits per draw; 10 assessment draws → 50.00 total expected hits.
Rank / rule
Exploration mean
Assessment mean
Difference from model
Assessment total
Whole-period mean
1. Binary rhythm (4 bits)
4.467
6.200
1.200
62
5.160
1. Binary rhythm (5 bits)
4.267
6.200
1.200
62
5.040
3. Most frequentExploration leader (ties possible)
5.400
5.200
0.200
52
5.320
3. Repeats
4.533
5.200
0.200
52
4.800
5. Binary rhythm (2 bits)
4.733
5.100
0.100
51
4.880
5. Binary rhythm (3 bits)
4.333
5.100
0.100
51
4.640
5. Least frequent
4.800
5.100
0.100
51
4.920
8. Partners
4.667
4.900
-0.100
49
4.760
9. Longest gaps
5.000
4.500
-0.500
45
4.800
A positive difference is an observation, not statistical proof of an advantage. Comparing nine rules and changing settings after seeing their results increases the risk of finding an accidental leader. Both periods remain retrospective; they are not a preregistered validation.
Fewer than 30 assessment draws: the ranking can change sharply after a single draw.
Accumulated overlap in the later assessment period — Most frequent
Solid line: actual accumulated hits. Dashed line: uniform-model expectation for the same number of draws and selections. The line is not a confidence interval.
X: assessed draws; Y: accumulated hits.
Inspect the latest 20 chronological steps
Target draw
Earlier training IDs
Earlier selection
Hits
17.11.2026 22:00 · #—
#17128 — #17152
1
✓
3
7
13
✓
18
✓
19
22
✓
26
31
33
37
38
✓
39
✓
42
44
55
57
62
72
77
6 / 20
17.11.2026 14:00 · #17152
#17127 — #17151
1
3
✓
7
13
18
✓
19
✓
22
✓
26
31
✓
33
✓
37
38
39
42
44
47
53
72
✓
74
79
7 / 20
16.11.2026 22:00 · #17151
#17126 — #17150
1
✓
3
7
13
18
19
✓
22
✓
26
31
33
✓
37
39
✓
42
44
47
53
62
66
74
79
5 / 20
16.11.2026 14:00 · #17150
#17125 — #17149
1
3
13
18
19
✓
22
✓
26
✓
30
31
33
✓
37
39
44
47
53
56
62
66
✓
74
79
5 / 20
15.11.2026 22:00 · #17149
#17124 — #17148
1
✓
3
✓
10
13
✓
18
21
22
26
✓
30
31
33
37
✓
38
39
✓
44
47
49
53
56
62
6 / 20
15.11.2026 14:00 · #17148
#17123 — #17147
1
3
✓
10
18
21
26
✓
31
✓
37
38
39
✓
44
47
49
53
✓
56
62
65
66
74
77
5 / 20
14.11.2026 22:00 · #17147
#17122 — #17146
1
3
7
10
18
✓
26
31
33
37
38
39
44
47
53
56
62
65
66
69
79
1 / 20
14.11.2026 14:00 · #17146
#17121 — #17145
1
3
7
10
✓
18
✓
26
31
✓
37
✓
38
39
✓
44
✓
47
51
53
56
62
65
66
77
79
6 / 20
13.11.2026 22:00 · #17145
#17120 — #17144
1
3
7
10
18
✓
26
✓
31
32
37
38
39
44
✓
47
49
51
56
61
62
✓
65
✓
79
5 / 20
13.11.2026 14:00 · #17144
#17119 — #17143
1
3
✓
7
26
31
32
33
34
37
✓
38
39
44
✓
47
✓
49
51
61
62
✓
65
✓
74
79
6 / 20
12.11.2026 22:00 · #17143
#17118 — #17142
1
3
✓
6
7
10
13
26
✓
31
32
37
38
39
47
49
✓
51
61
62
65
74
79
3 / 20
12.11.2026 14:00 · #17142
#17117 — #17141
1
3
6
7
✓
19
26
31
✓
32
✓
34
37
38
✓
39
✓
47
49
51
✓
61
62
65
74
79
6 / 20
11.11.2026 22:00 · #17141
#17116 — #17140
1
✓
3
✓
6
7
26
✓
31
✓
32
34
✓
37
38
✓
39
47
49
✓
51
53
61
✓
65
✓
69
72
74
9 / 20
11.11.2026 14:00 · #17140
#17115 — #17139
1
3
✓
6
7
✓
19
22
26
✓
31
32
✓
34
37
✓
38
39
47
49
51
53
✓
65
69
74
6 / 20
10.11.2026 22:00 · #17139
#17114 — #17138
1
✓
3
6
7
✓
22
26
31
32
34
37
38
✓
39
47
49
51
✓
53
62
69
74
79
4 / 20
10.11.2026 14:00 · #17138
#17113 — #17137
1
3
7
12
19
26
✓
31
✓
32
37
38
39
47
49
51
53
62
✓
65
69
74
79
3 / 20
09.11.2026 22:00 · #17137
#17112 — #17136
1
✓
3
7
✓
12
13
22
26
31
32
✓
37
✓
38
39
✓
44
47
49
51
53
69
✓
74
79
✓
7 / 20
09.11.2026 14:00 · #17136
#17111 — #17135
1
3
7
12
13
26
31
32
37
✓
39
✓
47
✓
49
51
53
62
✓
65
68
69
74
✓
79
5 / 20
08.11.2026 22:00 · #17135
#17110 — #17134
1
3
7
12
13
26
✓
31
✓
32
37
38
39
47
✓
49
51
53
✓
62
65
68
69
79
✓
5 / 20
08.11.2026 14:00 · #17134
#17109 — #17133
1
✓
3
12
13
22
26
31
32
38
39
47
✓
49
51
53
✓
55
62
65
68
69
✓
79
4 / 20
Which draw properties matched the earlier history?
For each step, we build an empirical range from the earlier window only: sorted positions floor(0.1 × (N−1)) and ceil(0.9 × (N−1)), inclusive. We check whether the later actual property lies inside it. Discrete values and ties can make coverage exceed 80%. These are descriptive ranges, not confidence intervals or number selections.
Property
Earlier median, latest step
Earlier range
Actual latest value
Inside, later assessment
Sum
814
738–897 (N=25)
718 · outside
6/10 · 60.0%
Range
75
70–78 (N=25)
79 · outside
9/10 · 90.0%
Even numbers
10
7–13 (N=25)
12 · inside
8/10 · 80.0%
High numbers
10
8–11 (N=25)
7 · outside
8/10 · 80.0%
Repeats from the preceding draw
5
3–7 (N=24)
4 · inside
9/10 · 90.0%
Band 1–10
2
1–4 (N=25)
4 · inside
10/10 · 100.0%
Band 11–20
2
1–4 (N=25)
3 · inside
10/10 · 100.0%
Band 21–30
2
1–4 (N=25)
2 · inside
10/10 · 100.0%
Band 31–40
3
1–5 (N=25)
4 · inside
8/10 · 80.0%
Band 41–50
2
1–4 (N=25)
1 · inside
10/10 · 100.0%
Band 51–60
2
1–4 (N=25)
2 · inside
10/10 · 100.0%
Band 61–70
3
1–5 (N=25)
1 · inside
9/10 · 90.0%
Band 71–80
3
1–4 (N=25)
3 · inside
10/10 · 100.0%
Properties have different units and are not ranked together by number hits. Bands are 1–10, 11–20, etc.; the final band ends at the pool limit. Repeat examples require a predecessor within the earlier window, so their sample is one draw shorter.
Does the historical leader remain the same?
The later assessment is divided into consecutive, non-overlapping periods. All tied leaders are shown; the final incomplete period remains visible. Periods below 15 draws are marked as small.
Changes in the set of leaders: 0/0 period boundaries.
Period
N
Leaders, including ties
Leader total
13.11.2026 14:00 — 17.11.2026 22:00
10 · small sample
Binary rhythm (4 bits), Binary rhythm (5 bits)
62
Changing a period size after seeing the outcome is retrospective fitting. This view does not establish a dependable rule.
Your selection in the same history
Enter your main numbers in the form above or open this view from My number report.
Multi Multi Plus: separate assessment
A separate pool, with its own equal selection size and model. Its hits are not added to the main-number ranking. 1/80. Excluded older draws: 0.
N=50 · Later assessment: 10 · Model mean: 0.013.
Rule
Assessment mean
Latest earlier selection
Latest hits
Most frequent
0.000
54
0/1
Partners
0.000
1
0/1
Repeats
0.000
65
0/1
Longest gaps
0.000
1
0/1
Binary rhythm (2 bits)
0.000
54
0/1
Binary rhythm (3 bits)
0.000
54
0/1
Binary rhythm (4 bits)
0.000
54
0/1
Binary rhythm (5 bits)
0.000
54
0/1
Least frequent
0.000
1
0/1
Save this historical assessment
JSON stores settings and calculated results; CSV contains the ranking. The HTML snapshot can be opened without internet and contains the visible assessment, not a new offline calculation. Exports include your entered numbers: share them only deliberately.
Sum of co-occurrence shares with numbers from the last earlier draw. Each partner count is divided by the anchor number’s frequency; self-pairs are excluded.
For each number, take its final two earlier presence bits, find that pattern in the earlier window, and measure how often a 1 followed it. Add two model-weighted pseudo-observations. A pattern without previous examples receives the model baseline.
Uses the final 3 earlier bits and known historical successors, with the same two model-weighted pseudo-observations as the two-bit rule. No earlier examples means the model baseline.
Uses the final 4 earlier bits and known historical successors, with the same two model-weighted pseudo-observations as the two-bit rule. No earlier examples means the model baseline.
Uses the final 5 earlier bits and known historical successors, with the same two model-weighted pseudo-observations as the two-bit rule. No earlier examples means the model baseline.
Sums, parity, ranges and decades describe a whole draw. They cannot fairly be ranked by number overlap without inventing additional selection rules. This page therefore compares only the nine rules defined above. Each independent draw retains the same odds.
Interactive tools
Explore the same data actively. The tools describe history and the random model; they do not predict the next result.
My selection — check its properties against history
This describes the entered selection and its position in the historical sample. It does not increase its chance.
Choose a year, use the slider or start automatic playback. This is a presentation of recorded results, not a forecast.
Open the module to load the available years.
Random-model simulator — see ordinary random variation
Generate independent draws with the rules of this game and compare observed counts with the common expected value. Every run is different.
How to read statistical measures?
Random model
The random model assumes every number has the same chance of being drawn each time. We compare results against this model to see whether the history departs from it — not to predict the next draw.
Expected value
The expected value is the average result we'd see if the numbers were fully random, repeated a great many times. Real history almost always differs from it a little — that alone is nothing unusual.
Deviation
Deviation shows how far a value is from what the random model expects, measured in units of typical spread. A value close to zero matches the model; the further from zero, the bigger the difference — but with many numbers compared at once, a few larger deviations are ordinary spread, not a signal.
Percentile
The percentile shows what share of results (from the model or from history) fall below the value being checked. Percentile 80 means about 80% of results fall below it and 20% above — it describes a place in the distribution, not a forecast.
Typicality
Typicality describes how close a value sits to the centre of the model's or history's distribution — typical values happen often, extreme ones rarely. It's a descriptive comparison and says nothing about what the next draw will bring.