still trouble understanding the GAT table

[Sooz]

I'm having trouble understanding the GAT table at the results panel.
My Example: I'm using 6 balls plus one bonus ball. I chose to examine a table which gave 7 numbers requested out of 49.
For total real draws to be tested I had chosen 100.

I manually counted the hits in those 100 draws to be as follows:
28 draws had no hits,
54 draws had one hit,
12 draws had 2 hits,
6 draws had 3 hits.
No draw, of the 100 tested draws, had more than 3 hits. As you can see the total (28+54+12+6) = 100 draws.

However, the hits produced given by GAT were close but not the same:
0 hits - 30%
1 hit - 70%
2 hits - 28%
3 hits - 5%
4 hits - 1%
5 hits - 1%
6 hits - 1%
7 hits - 1%

Why don't these numbers match my count? For example shouldn't 0 hits be 28% ? I promise I counted correctly.
And why does 4 hits, 5 hits, 6 hits, and 7 hits each show 1%. My (obviously faulty) understanding would have them show 0%.
Thanks. One of these days I'll be able to help answer questions instead of always asking them.
.....Sooz

[lottoarchitect]

Hi Sooz, how did you count manually those tested draws since you can't have the prediction of that GAT table for each tested draw which is different anyway? I'm not sure what you have done here but honestly you can't count manually the hits because you are not able to see all the predictions for all the tested draws used. This is only possible if you run version 2.0 or 2.1 and you have written down every prediction progressively displayed of that GAT for every tested draw. In GAT 2.2 we display only the prediction for the upcoming draw which shouldn't be compared to any of the tested draws (if this is what you have done).

[Sooz]

Hi LottoArchitect,
I'm not sure understand your question, but anyway here's what I did - then you can tell me where my reasoning becomes faulty.
I'm using GAT 2.2. After running the GAT engine overnight, I paused, and clicked not to continue. Thus the program arrived at the Results Panel. I chose one of the numbered tables and clicked on it. My question refers to the tab labelled "GAT table". This as you know, also has 2 tabs - "predicted numbers" and "hits produced". I'm looking at "7 numbers requested out of 49". The "hits produced" tab has the following columns: 0 hits, 1 hits, 2 hits, ...7 hits. Underneath is some percentages. Under 0 hits it says "30.00%". Under 1 hits it says "70.00%", etc. Are those percentages not showing the number of hits in the past 100 draws?

I have an excel file which has all the past winning draws. Each number has its own cell. One draw per horizontal line.
I also have a VBA macro with a command button. I did a lot of googling to find it and will paste it to the bottom of this email. I never learned programming so not sure why this program works but YOU will immediately understand it.

Then I typed the 7 numbers supplied by the GAT table, into 7 empty cells on the same excel sheet.

Then I clicked on the VBA command button. The button asks me the range and I select all those 7 cells. Then the VBA button asks me where to look, and I select the entire range of winning numbers for the past 100 draws. (It's a block of cells, 7 cells wide by 100 cells deep. Then the VBA macro highlights in yellow whichever of those 700 cells which had any number equal to any of those 7 numbers.
Then I went down line by line, and manually counted. Remember I have one draw per line. For example if a line had one cell highlighted in yellow, then that was a one hit. If a line had 2 cells highlighted in yellow, then it was a 2-hit, etc. Thus I can see how many times those 7 numbers came up in the last 100 draws. Is this not what the "hits produced" tab tells us??

Here is the VBA thing:

Private Sub CommandButton1_Click()

Dim Rng1 As Range
Dim Rng2 As Range
Dim bMatch As Boolean
Dim origRng As Range
Dim compRng As Range

On Error Resume Next
Set origRng = Application.InputBox("Choose the range to look for", "Range 1", Type:=8)
If origRng Is Nothing Then Exit Sub
Set compRng = Application.InputBox("Choose where to look", "Range 2", Type:=8)

For Each Rng1 In origRng
bMatch = False
For Each Rng2 In compRng
If Rng1 = Rng2 Then
bMatch = True

Rng2.Interior.ColorIndex = 6
'1 black, 2 white, 3 red, 4 green, 5 blue, 6 yellow, 7 pink, 8 turquoise, 9 brown
End If
Next Rng2
Next Rng1

End Sub

[lottoarchitect]

Well Sooz, the problem is that you compared different things really. GAT produces a different prediction for each tested draw.
Tested draw 1 -> Prediction P#1
Tested draw 2 -> Prediction P#2
...
Tested draw 99 -> Prediction P#99
Tested draw 100 -> Prediction P#100
Upcoming draw -> Prediction P#What you see as predicted numbers

P#1 is different to P#2 which is different to P#3 etc which is different to P#100 which is different to the prediction displayed in GAT predicted numbers. So what you did is not correct. Since you don't have access to P#1, P#2 etc up to P#100 you can't make that comparison externally by yourself. Those P# predictions are connected by the underlying signatures detected for that GAT table. Hope you understand what's going on.

cheers
lottoarchitect

[Sooz]

Sort of understand. My brain is tired but it's starting to sink in.
I was counting the hits in the actual past draws (ie the total 100 real draws to be tested), as though the P#100 displayed prediction was not just for draw #101 but was for all the previous 100 draws! I think this means there are actually 99 invisible predictions for each table! What an awesome program!
Thank you again.
.....Sooz

[lottoarchitect]

Actually when testing against 100 draws, there are 101 total predictions for the hits display plus some hidden used for the internal calibration of each GAT table, the first 100 are those compared to the tested draws and the one shown (101) is for the next draw to come (which doesn't contribute to the hits since we don't know that draw yet). Since for each performed prediction the same logic is used, what you have observed in the graphs shown "owes" to continue for the draws to come as well. And of course, if the engine is predicting e.g. draw #1500, any GAT computed to predict that tested draw uses only information of draws up to #1499 as influenced by all previous GATs via the signatures detected, which means GATs do not know at any stage which is the draw #1500 (or the hits produced for that draw by a precedent GAT) till all the GATs produce their prediction for that draw #1500. Then each GAT updates its hit ratio for that draw. This is what gives confidence of what to expect from the engine because it does not use information that shouldn't be used anyway to make a prediction - GAT is a real predictor.

cheers
lottoarchitect