Welcome Guest
You last visited December 6, 2016, 11:05 am
All times shown are
Eastern Time (GMT-5:00)

# Error Correction

Topic closed. 7 replies. Last post 11 years ago by lchoro.

 Page 1 of 1

United States
Member #5565
July 11, 2004
260 Posts
Offline
 Posted: June 20, 2005, 11:31 am - IP Logged

Has there been an algorithm or software developed that looks back at the
variances between the predictions that it makes for lottery picks and the historical jackpot values, and uses that lookback to adjust it future forecasts. I believe in math this lookback is referred to as an error correction function that may look at the variances and calculate the standard deviation about each guess for the jackpot value.

I'm thinking of a base function or procedure that makes the best guesses for each instance of the game. It uses the history of jackpot values within a window. Pretty standard fare for mathematically-based lottery programs.
The addition of an error correction routine would be used to construct a
wheel of numbers to play each game. If the standard deviation becomes
too large, the wheel becomes too unwieldy, and it means that the base forecasting function has to be discarded or that it would be best to stand on the sidelines until the forecasts improve.

Les

mid-Ohio
United States
Member #9
March 24, 2001
19825 Posts
Offline
 Posted: June 21, 2005, 12:28 am - IP Logged

First one would have to establish the effect that historical jackpot values have on guessing the size of the wheel to use and then adjust the variances in the standard deviation to create an error correction routine accordingly.  This all assumes that the forecasting function started out being correct and got worst over time or there would be no guide line for the error correction routine.

RJOh

* you don't need to buy more tickets, just buy a winning ticket *

mid-Ohio
United States
Member #9
March 24, 2001
19825 Posts
Offline
 Posted: June 21, 2005, 3:21 pm - IP Logged

On second thought one could look at the forecasting function and figure the variances of error from the start and use that in the standard deviations to correct that function accordingly after each drawings thus the forecasting function would become more accurate after each drawing.  Players who track hot and cold numbers are doing something similar now, but don't use the variances of error to more accurately define their parameters.  I hope this helps.  If we all put our thoughts together, we will eventually pick a winner.  Good luck to you.

RJOh

* you don't need to buy more tickets, just buy a winning ticket *

United States
Member #5599
July 13, 2004
1184 Posts
Offline
 Posted: June 21, 2005, 4:54 pm - IP Logged

RJOh,

I totally agree, error correction in your forcasting model is very important. This is something that has hardly been addessed here at the LP. I've been able to put together something at the number level, but not at the combination level.....

As most of you know I use a filter approach. Invariably, 1 or more of the filters is in error to the actual draw. So what I've done is rated the numbers after filtering. By tracking the rating I've found certain values keep coming up over and over....which I play.

You are a slave to the choices you have made.  jk

Even a blind squirrel will occasioanlly find an acorn.

Saint Louis, Missouri
United States
Member #17000
June 9, 2005
46 Posts
Offline
 Posted: June 23, 2005, 10:51 pm - IP Logged

And in playing those corrections, are you winning?

C.K.

If the eyes are the windows to the soul, let me see you without the shades drawn

What's a Hades Bunny?  Well let's just say you don't wanna be jumpin' over MY back fence.  But hey, you ain't afraid a lil white rabbit now are ya?

United States
Member #5599
July 13, 2004
1184 Posts
Offline
 Posted: June 24, 2005, 8:47 am - IP Logged

No, I am not winning more. However, the amount of number I match has gone up dramatically. The other thing holding me back, is that I have put a self inflicted betting cap of two dollars per draw.

You are a slave to the choices you have made.  jk

Even a blind squirrel will occasioanlly find an acorn.

East of Atlanta
United States
Member #6191
August 11, 2004
1389 Posts
Offline
 Posted: June 26, 2005, 10:46 pm - IP Logged

There may be something to that idea. I tinkered with a program for a few hours that did something similiar to that. It measured the difference between the numbers that hit, then provided possible future hits based on the variance that occurred over the past hits. Once I get my head around the idea, I will probably do the same thing in an Excel file and then have Excel generate numbers based on an evaluation of similiar data.

It looked something like the following:

 Original Number Start Skip Skip Skip Skip Skip Width Type 08-09-13-16-33 8 1 4 3 17 25 Pick 5 23-35-37-40-42-44 23 12 2 3 2 2 21 Pick 6

Now, as I understand the idea...by averaging this data over a course of time, one could presumably create a winning set of numbers. Sorry if it doesn't fully answer your question, but to me, the concept seem to apply.

Sir Metro

United States
Member #5565
July 11, 2004
260 Posts
Offline
 Posted: June 26, 2005, 11:53 pm - IP Logged

I haven't had a chance to look at the field for sometime. It's a technique in computer science algorithms called successive, or iterative, refinement which is used to solve a system of linear equations.

You might have a group of equations which are used to forecast a series of lottery balls, 6 in this case. There would be a formula for each ball. The forecast for the first ball may be dependent on the history of balls that occurred before it and what the probabilities of the next ball to follow. It may be based on what ball has the highest probability of occurring, whether an odd or an even ball is likely to follow the previous ball, what the normal delta is from the previous ball, and so on.. The forecast for the second ball would be dependent on the forecast for the first ball so that it would have to be at least a number one larger than ball one and it would observe other properties of relationship (odd/even, delta, etc.) between the two balls. It would also require similar probabilities for the second ball as the first equation that are dependent on the probabilites for ball two.

I was looking at measuring the distance from the forecasts to the results and then using those sets of values to construct a wheel that would hopefully envelope the winning numbers in future games.

 Page 1 of 1