# Predicting lottery numbers [closed]

I am working on a model and formula for predicting lotto numbers. Some success. Looking at recent ozlotto and sat lotto draws (pics). There were a total of 19 numbers drawn (2x sat (6 numbers)) and 1x lotto (7 numbers) a total of 19 numbers ..... of those 19 numbers I predicted 14 out of 19 (70%) with an accuracy of +/- 3...... does anyone know of any mathematical model or formula that can predict lottery numbers

• pics facebook.com/… Dec 18, 2016 at 9:45
• It's interesting how you think this is "successful" when, in fact, you are no "closer" to winning with 22 than with 35 if the correct number is 23. Dec 18, 2016 at 9:57
• the question was "does anyone know of any mathematical model or formula that can predict lottery numbers" Dec 18, 2016 at 10:06
• Even if you are claiming that there is a pattern in the lotto numbers, this doesn't qualify as a math question. Maybe consider a data science forum. Oct 24, 2022 at 14:04

By definition of unbiased independent random numbers, there is no memory or pattern, and thus prediction is not possible.

• Fabrice, I suggest you look at pics on fb page link regarding prediction. Dec 18, 2016 at 9:57

[lottery:] "A strange game. The only winning move is not to play"

The only reasonable mathematical model of lottery outcomes is that "Every combination of numbers has the very same probability to be the winning one, no matter what the previous outcomes".

• like i said 70% numbers predicted with an accuracy of + /- 3 for each number i provided pics of draws 22/10 .....17/11.... 10/12 .... 13/12 i won a few too its why i have not got pics of them..... i am after 100% number prediction with an accuracy of + / - 2 Dec 18, 2016 at 10:25
• @meandi: what does "accuracy of $\pm2$" mean? Dec 18, 2016 at 10:26
• ok the 4th number (lowest to highest) this tues ozlotto should be 25 or 27 or 31 i want to be accurate within 2 of the actual number Dec 18, 2016 at 10:34
• @meandi: but that means nothing! The numbers 25 and 27 are "close" (only assuming the natural topology, though, if you know what that means), but by no means the corresponding balls in the urn are "close" in any meaningful sense. You should think of lottery numbers not as actual numbers, but only as symbols (of no intrinsic numerical meaning) that allow to distinguish a ball from another. What kind of mathematical model would you ever think of if the balls were "numbered" with symbols as $\ast$, $\uparrow$, $\sharp$, $\&$, $\%$, ... instead of numbers? Dec 18, 2016 at 10:46
• but lottery numbers are numbers so you should consider them as numbers and then you can predict them facebook.com/… that draw was a lot better..... Dec 25, 2016 at 14:39

I think it is possible to use past data to build a model that fits that data, even get very good 'prediction' rates within that data (when you hold some back and use the rest to predict the left over). The trouble is you only get a very good model for THAT set of data. I've noticed patterns/distribution/bias that I can model well; I could even write a set of rules for an expert system that fit my data set almost exactly, but it only does that batch of historic data. It does not go forward (at least not yet!). It's a bit like the weather in that there are patches of 'this' (eg rain) and patches of that (eg sun) and within limits I can spot those and because it has behaved in a certain way in the past I make a prediction, of sorts, and check. But the patches of 'this' and 'that' could start to change next week and the rules Ihave no longer apply.