Heat Chart for Lotto / PowerBall
- Thread starter bloubul
- Start date
Frank,I wanted to let you know that Your Temp Sorter File has shown great results. As I said, I created a version that Looks at Each Draw Position for Ca. Pick 3. An Added Sheet that looks at a Descending Data Block and Pastes the Results for a Each Draw show some advantages. Turns Out that in 1000 Games the Hotter Numbers hit most (Duh). But one also finds that the Frozen Number is often the 3rd or 4th best hitter. You can see that as a number grows colder (on it's way to frozen) it hits less often. Again (Duh). Each Draw is a snap shot that show Me that 50% of the hits will come from the Hottest 3 and the Frozen 1.
I mentioned in an earlier response a question about when to measure the temperature. I Set up a test that let's me change from what will be to what was. In other words First column is the Skip when the Number last hit instead of the Current Skip. The Results in my last Post are based on "What Was". I just reran the data using "What Is". this shows that Frozen is the 3rd Best hitter in Positions 1 and 2 and the 6th best hitter in position 3. In all cases White hot hits most often, but, "What is" is about 1% less than "What was". I doubt that any number ever falls continuously from white hot to frozen; but, if it did and then was drawn it would log a white hot tag instead of frozen, thus explaining the (I think) useful Increase.
Any Way, Just wanted You to Know how useful and enlightening you Work here is.
Thanks Again for the Cool Tool
Any Way, Just wanted You to Know how useful and enlightening you Work here is.
Thanks Again for the Cool Tool
The Original Setup uses the Current Skip followed by the previous 3 Skips to calculate the current "Temperature" of Each Number. That Gives us "What Is". In other words what the "Temperature" will be if a Digit Hits in the Current Draw. As I have progressed it appears that that is the Way to use this process most effectively.
If the First Column is the last actual skip followed by the previous 3 skips Then you see "What was".
I was suggesting that Comparing the Sores for each option might be useful somehow. You get to see how much and in what direction the Score will move if the Digit is drawn in a Position in the Next Draw.
I am currently focusing on the "What is" or "What will Be" Setup (Frank's Original Setup). I am not sure if I will look further at the "What was" because I have not figured out how to use it effectively yet.
If the First Column is the last actual skip followed by the previous 3 skips Then you see "What was".
I was suggesting that Comparing the Sores for each option might be useful somehow. You get to see how much and in what direction the Score will move if the Digit is drawn in a Position in the Next Draw.
I am currently focusing on the "What is" or "What will Be" Setup (Frank's Original Setup). I am not sure if I will look further at the "What was" because I have not figured out how to use it effectively yet.
Frank,
I've been wanting to reply to your file on mediafire but have had problems logging in, thanks LT for sorting out the problem. I was wondering how you came to your weighting formula. I see from your reply to Jack that you are off to Cape Verde soon, I hope you have a good time, been there 3 times, Sal and Boa Vista, but also visited Sao Vicente & Fogo, always good. I like to dabble with Excel but wish I had your expertise.
Have a great holiday.
Rgds
Cartref
I've been wanting to reply to your file on mediafire but have had problems logging in, thanks LT for sorting out the problem. I was wondering how you came to your weighting formula. I see from your reply to Jack that you are off to Cape Verde soon, I hope you have a good time, been there 3 times, Sal and Boa Vista, but also visited Sao Vicente & Fogo, always good. I like to dabble with Excel but wish I had your expertise.
Have a great holiday.
Rgds
Cartref
Frank
Member
Cartref,
Well I'm back, and have been catching up with stuff over the past few days. You asked how I came about my weighting formula. Well for the recent example, I couldn't remember exactly how I arrived at my original heat scores 22 years ago and could not find the original excel file. So I had to improvise, and came up with that version because I happenned to have some skip values to hand and wanted a quick way to generate scores. I knew that my original method took into account both frequency and recent appearances, but the skip method was much simpler. I did not regard the method of arriving at a score as important as converting the scores to a heat chart, and that was the main purpose of that spreadsheet.
I did stress that everyone will have their own view of how to define hot or cold and could devise their own method of working out a score. The essence of the scoring system is to acknowlege the difference between a ball drawn one draw ago and one drawn say five draws ago. The problem with skip charts is that they don't explicitly define when a ball was drawn, so much as to count how long it's been absent ,and you find yourself dealing with figures in descending order going back in time until they reset at zero in the past. So long overdue numbers have high skip values over recent history and result in (high score = cold balls) and vice versa.
The problem with the skip values is this :-
Suppose the the skip value in the most recent draw was 25. Then 10 draws ago it was 15. In terms of cooling down between 15 and 25 it's percentage cooling based on skips is 10/15 which is a factor of 66%. I don't think that is much over 10 draws, I think it should be a multiple much more than 100%. I'm saying that I think cooling should be exponential not linear.
Thats why weighting was employed for that version of heat calculation, to force cool the ball the further into the past you go without it being drawn. For brevity I did not look back very far, ideally one should look a lot further back than four draws.
Since then I have recovered an old copy of my original work and now remember my method from all those years ago. For the 649 lottery, I looked back 12 draws and instead of using skips, I had a grid of 0's and 1's. 0' meant a ball was not drawn, 1 meant it was drawn. I calculated what I called 'recency rating' from this grid. So if the recent history of a ball was:- ( leftmost is most recent)
0,0,1,0,1,1,0,0,0,1,0,1 my weighting favoured the most recently drawn balls, 12 points for the most recent, 11 points for the previous one, etc down to one point for 12 draws ago. But points only score where there is a 1' in its history. So this would have a recency rating of 0 + 0+10 + 0 + 8 +7 + 0 +0 +0 +3 + 0 + 1. = 29.
You should notice that this method, instead of force cooling overdue balls, boosts the heat of more recently drawn balls instead.
I also calculated another figure based on the balls frequency and the expected frequency for the current draw. This had to be weighted to make it of comparable value with the recency figure ( hence having influence on the final score). For 49 balls my 'occurrence' figure was (actual F X 50) /expected F . Usually a figure in the range 45 to 55 would result for all the balls.
My heat score was the sum of the two components, e.g 29+ 47 = 76. How much importance F had could be varied by changing the multiplier ( currently 50) to say a lower value 20 , making recency of a ball more influential.
You should notice that this way of doing it, high score = hot balls and vice versa, unlike my previous example. This only works with an upside down lookup table as VLOOKUP doesn't like the first column to be sorted descending. There are other display issues too, which can be overcome.
Like I've said all along, there is no right answer to this. It's what you feel is most important:- Frequency ( be that over lotto history or over the past year) or recent appearances, or both - you can massage fiddle factors to reflect your view on this.
bloubul
Member
Well I'm the culprit that started this thread.
Frank now you let loose the proverbial cat among-st the pigeons.
I'm using Excel... WinnaLotto... GH Gold and Plus, and a macro to provide me with "so called" skips. As Excel is a accounting peace of software I take it to be more accurate in value, compared to some software.
This is one of the most awesome spreadsheet sheets I ever came across. Frank you have out done yourself on this. I have all the nrs with in 4 lines, the nrs are a little to much,,, BUUUT it's there all the time.
Frank once again I tip my hat too you.
Now I'm going to stir the "SNAKE BAG", Is "EXCEL" Superior to any Lotto Software or not?????
BlouBul
Frank now you let loose the proverbial cat among-st the pigeons.
I'm using Excel... WinnaLotto... GH Gold and Plus, and a macro to provide me with "so called" skips. As Excel is a accounting peace of software I take it to be more accurate in value, compared to some software.
This is one of the most awesome spreadsheet sheets I ever came across. Frank you have out done yourself on this. I have all the nrs with in 4 lines, the nrs are a little to much,,, BUUUT it's there all the time.
Frank once again I tip my hat too you.
Now I'm going to stir the "SNAKE BAG", Is "EXCEL" Superior to any Lotto Software or not?????
BlouBul
Cartref,
Well I'm back, and have been catching up with stuff over the past few days. You asked how I came about my weighting formula. Well for the recent example, I couldn't remember exactly how I arrived at my original heat scores 22 years ago and could not find the original excel file. So I had to improvise, and came up with that version because I happenned to have some skip values to hand and wanted a quick way to generate scores. I knew that my original method took into account both frequency and recent appearances, but the skip method was much simpler. I did not regard the method of arriving at a score as important as converting the scores to a heat chart, and that was the main purpose of that spreadsheet.
I did stress that everyone will have their own view of how to define hot or cold and could devise their own method of working out a score. The essence of the scoring system is to acknowlege the difference between a ball drawn one draw ago and one drawn say five draws ago. The problem with skip charts is that they don't explicitly define when a ball was drawn, so much as to count how long it's been absent ,and you find yourself dealing with figures in descending order going back in time until they reset at zero in the past. So long overdue numbers have high skip values over recent history and result in (high score = cold balls) and vice versa.
The problem with the skip values is this :-
Suppose the the skip value in the most recent draw was 25. Then 10 draws ago it was 15. In terms of cooling down between 15 and 25 it's percentage cooling based on skips is 10/15 which is a factor of 66%. I don't think that is much over 10 draws, I think it should be a multiple much more than 100%. I'm saying that I think cooling should be exponential not linear.
Thats why weighting was employed for that version of heat calculation, to force cool the ball the further into the past you go without it being drawn. For brevity I did not look back very far, ideally one should look a lot further back than four draws.
Since then I have recovered an old copy of my original work and now remember my method from all those years ago. For the 649 lottery, I looked back 12 draws and instead of using skips, I had a grid of 0's and 1's. 0' meant a ball was not drawn, 1 meant it was drawn. I calculated what I called 'recency rating' from this grid. So if the recent history of a ball was:- ( leftmost is most recent)
0,0,1,0,1,1,0,0,0,1,0,1 my weighting favoured the most recently drawn balls, 12 points for the most recent, 11 points for the previous one, etc down to one point for 12 draws ago. But points only score where there is a 1' in its history. So this would have a recency rating of 0 + 0+10 + 0 + 8 +7 + 0 +0 +0 +3 + 0 + 1. = 29.
You should notice that this method, instead of force cooling overdue balls, boosts the heat of more recently drawn balls instead.
I also calculated another figure based on the balls frequency and the expected frequency for the current draw. This had to be weighted to make it of comparable value with the recency figure ( hence having influence on the final score). For 49 balls my 'occurrence' figure was (actual F X 50) /expected F . Usually a figure in the range 45 to 55 would result for all the balls.
My heat score was the sum of the two components, e.g 29+ 47 = 76. How much importance F had could be varied by changing the multiplier ( currently 50) to say a lower value 20 , making recency of a ball more influential.
You should notice that this way of doing it, high score = hot balls and vice versa, unlike my previous example. This only works with an upside down lookup table as VLOOKUP doesn't like the first column to be sorted descending. There are other display issues too, which can be overcome.
Like I've said all along, there is no right answer to this. It's what you feel is most important:- Frequency ( be that over lotto history or over the past year) or recent appearances, or both - you can massage fiddle factors to reflect your view on this.
Frank,
Thanks for the in depth explanation which I found most interesting and will spend some time investigating further. The formula you use for the Frequency part :- (actual F X 50) /expected F; am I correct in assuming that the actual frequency is how many times a ball has hit in the last "X" draws, where "X" could be 20, 40 , 50 or 100. Is the expected frequency the same for all balls (49/6)?
Did you find that quite often a number of balls had a recency figure of 0, I mostly play the Thunderball and at present there are 9 balls that haven't hit for over 12 draws and 2 that haven't hit for 12 draws.
As you say though it is a matter of trying various factors.
Cartref
Frank
Member
Hi Cartref,
"
Regarding actual Frequency I used to use lifetime frequency from draw 1 so in your defintition "X" would be the actual number of draws from day 1. When you use this defintion then expected frequency is a variable depending on what draw number you were currently on. For UK lotto 649, if I was counting all 7 drawn balls in my analysis then expected frequency was calculated from EF= (X times 7)/49 where X is the currrent draw number. So by draw 1000 it would have been 7000/49 = 143 to the nearest whole number. If you limit your lookback period for counting frequency to a fixed number of draws, say X=104 then you are correct, and this would mean that the expected frequency is now a constant which would be 15 for all 7 balls counted in the 649 lotto.
I did find that the recency rating was sometimes zero and this boils down to how far you decide to look back in your '0' and '1' hit and miss grid. In fact I've just created my version to look back 16 draws for the UK 659 due to the reduced probability of a hit when using 59 balls. I found that 19% of 'attained skips' were 13 or more, but 11% of 'attained skips' are 17 or more, so 11% of my results will have a recency rating of 0. If I didnt have an 'occurence' element to my final figures I would need to look back more than 16 draws (probably 22) in order to get a wider spread of recency values to work with.
I dont have any useable data for the thunderball which would indictate how far to look back in a hit/miss grid to avoid too many zeroes, but i would suggest at 5 balls drawn from 39 the expected cycle time of 7.8 per ball is more than the UK 749 was at 7 per ball and you should therefore be looking back further than 12 draws too.
Good luck!
Hi Frank,
I agree with your views so I am going to go back 20 draws, currently there are no balls with a skip greater than 19 but there are times when there are just a few greater than 20.
I'm also going to look into your suggestion of expected frequency as well as fixed expected frequency, I'll only use the 5 balls as the Thunderball is chosen from a separate pool of 14.
I'm currently building a spreadsheet to calculate the current heat of each ball and the heat of each ball when it was drawn, this should hopefully give me a better insight when I decide which temperature to use. I'm currently using just 4 - Hot, Warm, Tepid & Cold.
I'll get back to you in a couple of days with any findings
Cartref
I agree with your views so I am going to go back 20 draws, currently there are no balls with a skip greater than 19 but there are times when there are just a few greater than 20.
I'm also going to look into your suggestion of expected frequency as well as fixed expected frequency, I'll only use the 5 balls as the Thunderball is chosen from a separate pool of 14.
I'm currently building a spreadsheet to calculate the current heat of each ball and the heat of each ball when it was drawn, this should hopefully give me a better insight when I decide which temperature to use. I'm currently using just 4 - Hot, Warm, Tepid & Cold.
I'll get back to you in a couple of days with any findings
Cartref