what does a prediction formula look like?Prev Topic Next Topic

New Topic New Poll

Developer

Somerset
United Kingdom
Member #9,710
December 17, 2004
189 Posts
Offline

Feb 8, 2006, 7:35 pm

Here is a simple correlation formuler to aid in your quest for a prediction formular
Look Correlation up on Google for more information
So lets say we have the balls 2,3 and 4

Ball 2 = 3 (Appeared 3 times) : B2=2 : B2T=3
Ball 3 = 3 (Appeared 3 times) : B3=3 : B3T=3
Ball 4 = 6 (Appeared 6 times) : B4=4 : B4T=6

mb = 12 / 3 = 4 ::::: mb = (All balls totals sum 3,3,6) / (number of balls, 3) = 4
mt = 9 / 3 = 3 ::::: mt = (All ball numbers sum 2,3,4) / (number of balls, 3) = 3

Sb = the standard deviation of mb = 1.73 (I may post how if i get the time)
Tb = the standard deviation of mb = 1.00 (I may post how if i get the time)

x = (3-4=-1) + (2-3=-1)= 1
y = (3-4=-1) + (3-3=0)= 0
z = (6-4=+2) + (4-3=+1)= 2
w= x+y+z = 3

or you could write it as

x = (B2T-mb=-1) + (B2-mt=-1)= 1
y = (B3T-mb=-1) + (B3-mt=0)= 0
z = (B4T-mb=+2) + (B4-mt=+1)= 2
w= x+y+z = 3

v = (w / (Sb*Tb)) = 3

u = v / Sb = 1.73

v / v-u = 0.87

0.87 is a correlation value used to see relationships between the data sets

a value of -1 or +1 is perfect you should now predict the lottery
a value of 0 is bad and you will not win
The closer you are to +1 or -1 the better

A correlation value of more than +1 or -1 means you did somthing wrong as this can not happen.

Note in this case the correlation value is a mesure of the strength of relationship between the numbers count and the numbers in the data set

Being a programmer the above should be stright forward putting into a application and then maybe plotting the correlation (linear relationship) on a scatter graph of the data set

I hope this helps

It has been said every software has a bug in it, it has also been said that every program can be reduced by one line of code.

There for any program can be reduced to 1 line that dont work.
hypersoniq

Pennsylvania
United States
Member #1,340
April 6, 2003
2,585 Posts
Online

Feb 9, 2006, 5:49 am

Developer, thanks for the info!
is SB and TB a difference like STDEV and STDEVP in excel (sample vs. population?)

Have an EXCELlent day!
truecritic

Michigan
United States
Member #22,394
September 24, 2005
1,583 Posts
Offline

Feb 9, 2006, 6:38 am

Running a curve fitting program over a random scatter-graph has never produced usable data for me. Doesn't stop me from trying. But it has never worked.
Perhaps a Non-linear curve fit?
Perhaps a Polynomial equation?
time*treat

United States
Member #13,130
March 30, 2005
2,171 Posts
Offline

Feb 9, 2006, 8:39 am

If I only have to put down a few bucks per try, I don't mind if the "curve" does not pass through EVERY "point". A few curve-point intercepts here and there will add up to REAL money.

In neo-conned Amerika, bank robs you.
Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.
LottoFan

Arizona
United States
Member #4,595
May 6, 2004
57 Posts
Offline

Feb 11, 2006, 9:02 am

Hi Guys. Long time, no post.
I hate to be the guy from "left field", but I'll share some of my thoughts.
I only play jackpot type games. When presented with two quickpicks: 1-2-3-4-5-6 and 51-52-53-54-55-42, many people "see" 12 different numbers, and track them accordingly. I see only two numbers. That is because I treat each combination as a seperate (unique) entity or number.
Conversely, in Powerball, for example, I see approximately 146 million seperate numbers, and most other people see 55 seperate numbers. For me, in order for bias to occur (in relation to the above quickpicks), either 1-2-3-4-5-6 or 51-52-53-54-55-42 must repeat one or more times. So 2 PB drawings per week = 104 drawings per year = 1040 drawings per decade = 10,400 drawings per century. For me then, I have only "seen" a little over 10,000 "numbers" in 100 years.
Where can I find "bias"? I look at the lowest number of a PB pick. For my two examples, that would be "1" and "51". The lowest number is my "key"number. If I examine all of the "combinations" beginning with "1", I see that there are several hundred thousand. If I examine all of the combinations beginning with "51", I see there are only 42. There is a static, predictable, relationship between all "first" numbers in a combination. It is somewhat proportional, as well as declining (say, from around 400,000 to just 42).
Well, I'll stop here for now. As you can see, my first level of bias is located within the first number (lowest) in a picked combination. Just wanted to let everyone know that someone is working to solve the problem from another angle!
time*treat

United States
Member #13,130
March 30, 2005
2,171 Posts
Offline

Feb 11, 2006, 9:54 am

Heh, left field is getting a little crowded.
It looks like you're using something like the "delta" system, where you take a key number and then the others are based on differences between them and the next number. A whole method can be based on average distances between one number and the next.
I haven't graduated to the 5+1 games yet, but I do assign each possible combination a unique number, in the lotto 5. I guess many of us are converging on the same systems. Then, just like the movie, the Matrix will change.

In neo-conned Amerika, bank robs you.
Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.
RJOh

100

mid-Ohio
United States
Member #9
March 24, 2001
20,272 Posts
Offline

Feb 13, 2006, 5:34 pm

I'm looking for a formula that would have covered the winning combinations 10% of the time in a group as small as one hundred thousand possible combinations o so. I don't think anyone can come up with a prediction formula but it may be possible to uncover a pattern or condition that exist 10% of the time that would include only a small potion of the possible combinations. By playing combinations in those parameters, one would increase the chance of winning when that condition occurs.

* you don't need to buy every combination, just the winning ones *
time*treat

United States
Member #13,130
March 30, 2005
2,171 Posts
Offline

Feb 13, 2006, 5:51 pm

RJOh,
That's the exact next step. Have you considered skip tables? Clusters? (I mentioned that on an older thread). I'm thinking of doing a write-up on it one day

I haven't tried my system-in-progress on a 5/39 yet, but it showed promise on the old IL 5/30 & IN's 5/36.

In neo-conned Amerika, bank robs you.
Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.
RJOh

100

mid-Ohio
United States
Member #9
March 24, 2001
20,272 Posts
Offline

Feb 14, 2006, 10:29 am

I tried a lot of things but the problem is each idea have to be programmed and parameters have to be entered and tested one at a time. I need to program ideas that are self testing, that is I enter the range of the parameters and allow the program to run and at the end display it best results.

* you don't need to buy every combination, just the winning ones *
LottoFan

Arizona
United States
Member #4,595
May 6, 2004
57 Posts
Offline

Feb 14, 2006, 12:47 pm

I'll be playing both QPs and my own picks this Wed. For my own picks, I'll be invoking the "Valentine Deviation" - 02-14-06!
LottoFan

Arizona
United States
Member #4,595
May 6, 2004
57 Posts
Offline

Feb 14, 2006, 12:52 pm

Quote: Originally posted by time*treat on February 11, 2006
Heh, left field is getting a little crowded.
It looks like you're using something like the "delta" system, where you take a key number and then the others are based on differences between them and the next number. A whole method can be based on average distances between one number and the next.
I haven't graduated to the 5+1 games yet, but I do assign each possible combination a unique number, in the lotto 5. I guess many of us are converging on the same systems. Then, just like the movie, the Matrix will change.

You are right, time*treat. My second step is exactly that- determine the significant difference between my picked number(s) and the winning numbers. It's just that back-testing is such a pain-in the-a$$.
time*treat

United States
Member #13,130
March 30, 2005
2,171 Posts
Offline

Feb 14, 2006, 8:31 pm

Quote: Originally posted by RJOh on February 14, 2006
I tried a lot of things but the problem is each idea have to be programmed and parameters have to be entered and tested one at a time. I need to program ideas that are self testing, that is I enter the range of the parameters and allow the program to run and at the end display it best results.
I've never used GWBasic, but just tackled this problem (slightly different parameters on the same data) a few weeks ago, in VBA for eXceL. It was tedious, but I wound up not adding as much hard code as you might expect. I think I will use up a bit more ram when it runs, though. I just worked out the proof-of-concepts, so far. VBA doesn't require line numbers, so editing is much easier than in some of the older langs. I don't know what GWB will let you do, so I don't know if it would be useful/possible for you.

In neo-conned Amerika, bank robs you.
Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.
RJOh

100

mid-Ohio
United States
Member #9
March 24, 2001
20,272 Posts
Offline

Feb 15, 2006, 9:03 am

I have Quick and Visual Basic, neither require line numbers. I even use the QB compiler to compile my GWB programs, I just got stuck in time and too old and tired to update. When I was younger it was easier to switch from TI Basic to Commodore Basic to Tandy Basic to GWBasic, even tried some Pascal. I might be forced to update if someday Windows doesn't run DOS programs.

* you don't need to buy every combination, just the winning ones *
time*treat

United States
Member #13,130
March 30, 2005
2,171 Posts
Offline

Feb 15, 2006, 12:00 pm

RJOh,
You probably will be able to use this with what you already have, then.
I will do a separate thread on it later in the week. That way, hyper can have his thread back. I think I've posted on it more that he has.

In neo-conned Amerika, bank robs you.
Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.
RJOh

100

mid-Ohio
United States
Member #9
March 24, 2001
20,272 Posts
Offline

Feb 16, 2006, 6:21 pm

time*treat,
If you can answer hypersonig's questions, what is a prediction algorithm, what does it look like and what are the steps needed to look at a list of numbers to extrapolate the numbers coming up in the next drawing, why would he mind you doing so in this thread? So far after 59 replies no one has come close, so he will probably be interested in your answers also. Good luck to you.

* you don't need to buy every combination, just the winning ones *

New Topic New Poll

Subscribe to this topic

what does a prediction formula look like?Prev TopicNext Topic

what does a prediction formula look like?Prev Topic Next Topic