by lottoboy » Sun Jun 01, 2008 10:24 pm
[quote=Bobijohn]
Hi Folks.
Perhaps a little clarification may be helpful.
lottoboy. Sorry, I don’t have a link to Excel spreadsheets. You will need to export the WNH differences summary table and import that into Excel. Then use the standard built-in functions to compute the standard deviation for each of the 11 columns. Elsewhere in this forum there are directions on how to perform the export/import procedure. Once in Excel you can “Google†how to use the standard built-in mathematical and statistical functions to get the standard deviation and many other useful statistics.
PadawanLotto. The standard deviations given are approximate for each column. In other words, the standard deviation for column -0 may be 15.8 and that for column -7 may be 17.5. Their use would be to take ± these values around whatever predicted sum has been chosen. This is just an alternative to using ± 20, as has been mentioned often in the forum. It just provides a statistical basis for the choice – no more, no less. I agree with you that to try and set the WNH difference between 15 and 18 would be useless. That is not what I intended.
Just a comment. As for accurately predicting the sum for each column – well, we would all like to know how to do that. However, I will add that if you just took the sum from the last draw and applied it to the next draw in conjunction with the standard deviations derived for each column of your draw as described above you will have all 11 columns in tolerance about 1.5% of the time (67.5%^11=1.5%) taken over >1000 draws. However, if you use the median values as the predicted sums (disregarding the cautionary comment in the help files) for the next draw and apply the standard deviation derived in Excel this percentage increases to about 2.0%. Anyway, on this basis, for a lottery that plays twice a week you could expect to have the winning ticket once per year assuming you did this consistently. The problem is that for a large lottery this still gives too many tickets for most of us to play. So now we are back to additional filtering and their related risks of losing the winning ticket. This is comment is given just as an example to demonstrate the use of the mean and standard deviation to set a predicted sum and tolerance level respectively. Most of my experimental analysis is performed on a pick 4 lottery just for speed and simplicity. The same principles would apply to other lotteries with slightly different results.
That is about all I have on this topic.
Good Luck Guy’s.
Bobijohn.
[/quote]
Hi, Stan:
I think that Bobijohn's strategy about the mean and standard deviation to set a predicted sum and tolerance level would work for all lotto. And it may be a good supportive tool for the main filter of ± 20/9-11 in EL.
Could you want to check its accuracy further? If the answer is positive, could you add it to our EL as a new Statistics or Filter tool? Thanks.
Best,
lb