← All methods

Weighted ensemble

the other methods blended by how they have actually scored — it starts as a flat consensus and only tilts as evidence arrives

How Weighted ensemble works

Special method

Blends the other methods, trusting each one according to how it has actually scored.

How the numbers are chosen
  1. Every method starts with equal weight.
  2. Over the last 120 draws, a method whose hits beat the chance line gains weight; one that falls short loses it.
  3. Each number tonight scores the total weight of the methods that picked it; the top 3 are the picks.
  4. Because it's built from the others, it's never counted in the night's totals or 'consensus'.
Too early to tell. It averages 0.38 box hits per number per draw — pure luck would give about 0.32, and the random picks are getting 0.20.
Box hits per number, running average
This method: 0.38Its chance line: 0.32
Changes to this method›
  • 3 Oct 2026 — measured against the chance line of the numbers it actually picks, instead of one line for four-different-digit numbers. The whole record was rescored.
  • 20 Aug 2026 — hits counted on the six Malaysian-licensed boards only (no Singapore or daily draws).

Picks lock before the draw and never change. In a fair draw every number has the same odds — a method only changes which numbers you pick.

Tonight's picks →
120
Picks locked
40
Nights
45
Box hits
1
Straight hits

If you had backed every pick

RM1 iBox Big on each pick, on each Malaysian board that night, priced with the real dividend table. An iBox bet covers every ordering of the number, so exact and permutation hits both pay.

RM720
Staked
RM636
Returned
88%
Return

For every RM100 bet, random luck would pay back about RM66.7 on these same numbers — this method is doing 1.32× as well as that. (Each method is measured against the numbers it actually picked: a number with a repeated digit has fewer orderings, so it pays back less on average.)

Money is the noisiest way to judge a method. One 1st prize is worth 35 consolations, so a short record mostly reflects who got lucky rather than who picks well. The pure-random control arm is running at 1.63× its own line on the same measure.
Special × 15Consolation × 252nd Prize × 31st Prize × 3

Work out your own bet

The figures above assume RM1 iBox Big. Set the bet you actually play and this re-settles every pick above against the real draws.

Coverage

One line covering every arrangement of the number. Cheapest way in, but prizes are divided by the number of arrangements. Not sold on a number with all four digits the same.

Where you bet
RM
RM
RM720
Total spent
RM636
Total won
−RM84
Loss

That's about RM18 per draw.

That is a 88% return, against the 66.7% a fair draw returns on this bet. Money came back on 29 of 40 nights.

Best night: Wed, 23 Sept 2026, up RM137 after stakes.

This is what already happened, not what will. The same method over the next 20 nights can look nothing like this, and every bet here loses money on average — a fair draw returns well under 100% on all of them.

Every locked pick

Newest first. Each number was written to the database before its draw and never changed.

1 night with no hit (Sun, 15 Nov)−RM18›
Sun, 15 Nov 2026−RM18nothing won · RM18 staked
6087no hit
6347no hit
7280no hit
Sat, 14 Nov 2026−RM12RM6 won · RM18 staked
Wed, 11 Nov 2026−RM15RM3 won · RM18 staked
1480no hit
9831no hit
Sun, 8 Nov 2026−RM15RM3 won · RM18 staked
2971no hit
6180no hit
1 night with no hit (Sat, 7 Nov)−RM18›
Sat, 7 Nov 2026−RM18nothing won · RM18 staked
1480no hit
3971no hit
6871no hit
Wed, 4 Nov 2026−RM15RM3 won · RM18 staked
3487no hit
3671no hit
1 night with no hit (Sat, 31 Oct)−RM18›
Sat, 31 Oct 2026−RM18nothing won · RM18 staked
0671no hit
2087no hit
6431no hit
Wed, 28 Oct 2026−RM10RM8 won · RM18 staked
0487no hit
6547no hit
3 nights in a row with no hit (Sat, 24 Oct – Tue, 27 Oct)−RM54›
Tue, 27 Oct 2026−RM18nothing won · RM18 staked
2731no hit
3047no hit
4087no hit
Sun, 25 Oct 2026−RM18nothing won · RM18 staked
3047no hit
3871no hit
7480no hit
Sat, 24 Oct 2026−RM18nothing won · RM18 staked
4831no hit
8531no hit
8631no hit
Wed, 21 Oct 2026−RM10RM8 won · RM18 staked
6271no hit
9180no hit
Sun, 18 Oct 2026−RM12RM6 won · RM18 staked
Sat, 17 Oct 2026−RM7RM11 won · RM18 staked
Wed, 14 Oct 2026−RM15RM3 won · RM18 staked
2674no hit
5980no hit
Sun, 11 Oct 2026−RM15RM3 won · RM18 staked
1687no hit
9831no hit
Sat, 10 Oct 2026−RM7RM11 won · RM18 staked
Wed, 7 Oct 2026+RM95RM113 won · RM18 staked
Sun, 4 Oct 2026−RM7RM11 won · RM18 staked
Sat, 3 Oct 2026−RM15RM3 won · RM18 staked
0187no hit
0247no hit
Wed, 30 Sept 2026−RM15RM3 won · RM18 staked
0347no hit
0671no hit
1 night with no hit (Sat, 26 Sept)−RM18›
Sat, 26 Sept 2026−RM18nothing won · RM18 staked
0187no hit
0287no hit
0931no hit
1 night with no hit (Sun, 20 Sept)−RM18›
Sun, 20 Sept 2026−RM18nothing won · RM18 staked
0187no hit
0947no hit
1294no hit
Sat, 19 Sept 2026−RM12RM6 won · RM18 staked
Wed, 16 Sept 2026+RM24RM42 won · RM18 staked
0894no hit
1394no hit
1 night with no hit (Sun, 13 Sept)−RM18›
Sun, 13 Sept 2026−RM18nothing won · RM18 staked
0187no hit
0394no hit
1394no hit
Sat, 12 Sept 2026+RM32RM50 won · RM18 staked
6948no hit
655412-wayno hit
Wed, 9 Sept 2026−RM7RM11 won · RM18 staked
Sun, 6 Sept 2026−RM10RM8 won · RM18 staked
6948no hit
7301no hit
Sat, 5 Sept 2026−RM15RM3 won · RM18 staked
6948no hit
7301no hit
Wed, 2 Sept 2026−RM15RM3 won · RM18 staked
7301no hit
0187no hit
2 nights in a row with no hit (Sun, 30 Aug – Tue, 1 Sept)−RM36›
Tue, 1 Sept 2026−RM18nothing won · RM18 staked
7301no hit
3167no hit
6948no hit
Sun, 30 Aug 2026−RM18nothing won · RM18 staked
7301no hit
3167no hit
6948no hit
Sat, 29 Aug 2026−RM12RM6 won · RM18 staked
Wed, 26 Aug 2026−RM15RM3 won · RM18 staked
1274no hit
7301no hit
Sun, 23 Aug 2026−RM15RM3 won · RM18 staked
0187no hit
1674no hit
Sat, 22 Aug 2026−RM7RM11 won · RM18 staked

Picks are locked before the draw (first write wins) and scored at read time against the real boards, so a corrected result re-scores this whole page. Box hits pay on the scoreboard but never move any clock — the number drawn was a different number.