Weighting Lotto Numbers

[Icewynd]

Or in my case, maybe it should be "waiting" lotto numbers

We all "handicap" or "weight" the lotto numbers once we make the decision to reduce the field from 49 (or whatever) numbers to something more manageable. This might be informal -- "46 looks good", or might be something much more structured that produces a ranking for each number in the lottery which, hopefully, tells one something about each number's chances of success in the next draw.

I am looking at creating a formal ranking system and it appears that many of you have already examined this challenge.

The base of my system will be a chart for the Canadian 6/49 game that I have been maintaining for several years. This chart plots the number of game skips for each winning number vs. its hits in the last 16 games. For example, this chart tells me that the most successful combination is a number with 2 hits in the last 16 games which last hit one game ago (i.e. skipped one game). Numbers of this description make up 3.8% of the winners over the last 265 games. Conversely, a number with 5 hits in the last 16 games which last hit 8 games ago has never come up in 2 1/2 years, so it likely wouldn't be a good bet in the future.

I am interested in adding variables to make this system more effective. Any suggestions would be appreciated as to what might work and (maybe more importantly), what doesn't! Also, I would be interested in any suggested methods -- for example how to combine the variables to create a valid weight? Do I have to create regression equations or are there simpler methods? I know that there are many software programs that do this, but they are 'black boxes'. I would rather know what variables go into my ranks.

Looking forward to a good discussion.

 

[blitzed]

hiya Icewynd,

big games such as lotto tend to have 2draws per week, amounting to 96 per year...I like to check probabilities there, and work my down down through various intervals...looking for patterns, taking note of numbers I like at each interval.

I think many people focus way too much on cold & hot numbers...better to play the numberfield represented by the bulk of probability in my opinion...instead of stuff on the fringe, which will only shift back towards median probabilities over the long-term.

cya,
blitzed

 

[Icewynd]

Thanks for your input, Blitzed. I think I see what you are saying.

For example, on the Canadian 6/49 game, numbers from the last 5 or 6 games have not hit for the last nine draws and have only produced 8 hits in the last 20 games vs. an expected value of 18. This puts their 16 game moving average at 0.4, i.e. less than half of the median of 0.9. Given the way these things move in cycles, I would expect to see a higher-than-normal proportion of numbers from the last 5 or 6 games over the next 10-15 drawings (we will see starting tonight).

But what would be the appropriate way to weight this? For example, I tried dividing by 0.4 but this more that doubles the weight for these numbers--too much! But adding 0.5 (difference between current value and median) doesn't seem strong enough.

Any suggestions?

 

[blitzed]

hiya Icewynd, you are most welcome.

you may want to checkout my old lotto 6/49 spreadsheet ifya have not already:
http://crazynuts.hollosite.com/lottery_spreadsheets/

649_Lottery.ods

it may require Open Office Calc, if you do not have it:
http://www.openoffice.org/

anyway, the P-probability sheet shows some example probability intervals.

D-distro sheets can be rigged for certain probability intervals, and numbers can be conditionally highlighted accordingly...then it is easy to see say a number playfield representing about 60% to 85% of probability.

for example, here is probability chart, for any number being drawn X times in 60draws...
6of49 60Draws
==============
0x 0.04%
1x 0.33%
2x 1.36%
3x 3.67%
4x 7.30%
5x 11.40%
6x 14.59%
7x 15.70%
8x 14.51%
9x 11.70%
10x 8.33%
11x 5.28%
12x 3.01%
13x 1.55%
14x 0.73%
15x 0.31%
16x 0.12%
17x 0.04%
18x 0.01%

numbers drawn 4 to 10 times, represent 83.53% of the number playfield...therefore focusing upon numbers drawn 3 to 9 times may represent the bulk of moving numbers in the next draw too.

as you can see, numbers drawn X times at the high or low end of the scale only represent a very small portion of probability...however, if you are watching a shorter-term interval such as 28draws...then chasing a cold number looks more worthwhile, since there is a 7.5% probability of movement. so sometimes it is good to pick a cold number you like, and sprinkle it into a few picks at least.

blitzed

 

[Icewynd]

Thanks, Blitzed, this is great!

If it won't open in Excel I can create my own, but this is a good way of looking at the data to see what might be coming up. I'll have to do some more digging on your site -- it seems like you have lots of great stuff that I didn't see the first time I visited.

as you can see, numbers drawn X times at the high or low end of the scale only represent a very small portion of probability...however, if you are watching a shorter-term interval such as 28draws...then chasing a cold number looks more worthwhile, since there is a 7.5% probability of movement. so sometimes it is good to pick a cold number you like, and sprinkle it into a few picks at least.

blitzed

Yes, one of the things that I want to build into my rankings is a 'chaos' or random factor. I have too often had the experience of thinking that a certain number or group of numbers would be drawn and then the lottery changed its pattern! I define things in terms of 16 games, and my data shows that about 30% of draws have all the numbers in the middle range, but one or more 'cold' numbers comes up about half the time.

I owe you one

 

[blitzed]

hiya Icewynd, it might be worth trying OpenOffice Calc, it has some functions which are not/were not in Excel, such as B (Binomial-distribution)...and hey it is free, also great how ODS files are compressed unlike XLS.

my website is a bit chaotic...I wanted to make a free lottery workbench with various ways for people to shake numbers out...there are heaps of various lottery web-apps scattered around...some have undocumented features...take for example the MLQP page:
http://crazynuts.hollosite.com/my_lottery_quick_picks/

you can actually copy say the last 16draws from your spreadsheet, paste them into the webpage and it will run some basic stats. generate some picks, then likewise you can copy & paste them back in, recycling them through MLQP for basic stats on them, as well as making a 2nd generation set.

I intend to eventually make an MLQP v2, which would include more stats, and some filter options...maybe I can figure out the probability bits in PHP too.

cya,
blitzed

 

[Icewynd]

"For all you do, this Bud's for you"!

What you have available is great!!! Let us know about updates.

I'm still digesting what you sent me earlier.

 

[Icewynd]

Going wayyyyyyyyyyyyyy out on a limb

OK, here is version 1.0 of my 'predictions' for tonight's draw...

Number Weight
5 59
20 54
37 51
22 50
38 47
8 45
13 44
18 43
9 43
30 42
12 41
31 39
15 38
14 34
25 33
10 33
44 33
17 32
16 31
23 30
2 29
40 29
46 26
28 25
21 25

History shall judge me...

 

[blitzed]

cool, no problem...many of us throw stuff onto the workbench, tinker until we find something that brings reasonable success towards reaching our goals.

goodluck,
blitzed

 

[Icewynd]

OK, here is version 1.0 of my 'predictions' for tonight's draw...

Number Weight
5 59
20 54
37 51
22 50
38 47
8 45
13 44
18 43
9 43
30 42
12 41
31 39
15 38
14 34
25 33
10 33
44 33
17 32
16 31
23 30
2 29
40 29
46 26
28 25
21 25

History shall judge me...

Pretty crappy -- 2 plus the bonus. Would expect 3.5/7 for 25 numbers. So, its back to the drawing board. The problem seems to be that my weights for current/short-term factors weren't strong enough to override the base weights.

 

[blitzed]

hiya Icewynd, bummer...yeah, that is a problem reyling upon such a small draw sample.

for big lottery games, I think the last 120 draws or so is a reasonable interval to start with...then I check at various lower intervals, at each choosing numbers I like, and eventually making a hybrid set based upon the probability intervals.

catchya later,
blitzed

 

[Icewynd]

Another thing that I am going to have to try is some sort of spatial, clustering indicator. Last draw 1 number repeated from draw before and three numbers were one removed from last draw numbers 16-15, 25-26 and 37-36.

This clustering happens quite often, with one or two numbers frequently found this way. I plot the numbers on a graph so I can see the patterns emerging. One pattern that is almost a sure thing is if you have two numbers with one space between them (eg. 13-15), choose the 'gap' number (14 in this example) for the next game to form a triangle pattern. It comes up frequently.

 

[blitzed]

hiya Icewynd, yeah that is one strategy a lot of Keno players use...they watch a sequential trio.

I've been meaning to add various filter options on some of my lottery web-apps...including one which would look for such missing leader delta-formations. perhaps I will have freetime to work it out.

cya,
blitzed

 

[Icewynd]

If you have the time to look at it, let us know how it works out.

You always have a unique take on things...

 

[Teufellj]

Icewynd...

Hello. Icewynd...
I have a suggestion about weighing certain numbers...
1_ Suppose you were to play only ten cards and of those same bets, you were to double one number (can be any of them...eg. 49.) The 9 doubled is
equal to 18 and so forth. (Using neighboring pairs like I do for tracking the pick 3.) The advent of what's happening is that you are still using whichever digits of the original set, yet you are expanding and reducing by at least three different ways.
The fourth way is by using mirror sets that will also reflect the original ones contained within the bounds of ...1...up to ...49...! Obviously, some digits in the mirror setup will not work because anything like 50 on up can't be used unless neighboring sets are produced and the odd digit is welded to another loose digit thus wrapping into the next line of figures.
Boils down to the fact that you would be using static numbers (original set) that can make no movement externally but does move internally to produce
several probable digits that are all part of one original bet! In short creative expansionism!

Enjoy...

Teufellj...

 

[blitzed]

hiya teufellj, cool...in the past I've done test algorithms which seamlessly wrap through the number playfield to work some mayhem.

you are correct...some imperfect mirrors...hmmm, in the case of a 6/49 game, 7 is the root of 49...a pseudo-mirror may work, by rounding to the nearest multiple of 7 I guess.

I have heaps of old code I was working on ages ago in my test folder...I should dust some off and see what I can come up with.


cheers!
blitzed

 

[Icewynd]

That's what's great about this place --getting new ideas and perspectives on things!

I'll have to sit down with my number set and see how these ideas operate on some past draws.

Thanks, Teffullj