possible way to choose which gat from buildup

G.A.T. Engine general discussion
Sooz
Proficient
Proficient
Posts: 246
Joined: Wed Feb 09, 2011 12:24 am
Location: Canada

possible way to choose which gat from buildup

Post by Sooz » Fri Dec 27, 2013 3:35 am

I'm now asking GAT to find 6-hitters with 11 numbers selected in buildup mode. Then this occurred to me: I browsed the 6 category repeatedly, each time lowering the number of numbers requested. There are only 2 gats that still have the 6-hit with only 9 numbers selected. One of these also still has a nice-looking buildup to its 6 hits. Does it make sense to prefer that particular gat, and to only wheel the first 9 numbers? ie can we say that particular gat is smarter or stronger because it was able to build up to a 6-hit with fewer selected?
Sooz

User avatar
lottoarchitect
Site Admin
Posts: 1457
Joined: Tue Jan 15, 2008 5:03 pm
Location: Greece
Contact:

Re: possible way to choose which gat from buildup

Post by lottoarchitect » Fri Dec 27, 2013 10:34 am

Hi Sooz, I don't think I can answer this question. Buildup gathers all the GATs that exhibit a steady/increasing hits production over the last part of the tested draws. That means, underneath each of these GATs has formulated dynamics from the analyzed data that managed this steady/increasing hits production. In the current GAT buildup implementation, a GAT that appears in the buildup panorama is considered "better" if it has a longer such steady/increasing hits production (and appears higher in the column) because this translates as it does maintain even a partial hits performance (which improves over time) and this can be assumed as this GAT did find even better dynamics so to maintain that hits performance over the longer run. The above means, if the lottery don't break the current "dynamics momentum" to the next draw, the longest GAT sequence has a much stronger ability to provide what that GAT illustrates.
However, the current data available cannot give an edge to favor one against another GAT, thus each GAT in a given column really has the potential to be the one to deliver the good hit. The only hard evidence I can answer for buildup is, even if you pick randomly any GAT from the buildup panorama, you have overall quite better chances to pick a number set that delivers above odds[#1]. Another observation for buildup is, if the next draw is "way off focus" from the current overall dynamics found by the engine, then all GATs will produce lower hits. If the lottery draw continues the current dynamics, a lot of buildup GATs will provide higher hits for that draw. Thus we have really cycles even at the buildup, observed in GATs at the same time. This is what we call "momentum" of the draw[#3]. When we are at a momentum, the computed dynamics give very good hits. All this might sound confusing, it is actually.
The bottom line is, I can't offer advice on which GAT is better among those you propose; the information required to answer that is probably beyond what can be gathered by the available data. I would really concentrate on GATs in the middle columns which, in my tests, these have shown a much more steady partial hits production performance of -+1. I have yet to find a GAT that made it to the highest column (means it did produce the winning number set at the last tested draw) and did something noticeable at the following draw; probably we have to scan even billions of GATs to appreciate GATs in the highest column - don't take for granted what I say here for the highest column however. It is just so far my tests have shown better overall good hits produced by buildup in the middle columns[#2].

[#1] - I'll explain that a bit more. Le'ts say you pick randomly 10 sets of 5 numbers and you pick randomly for play one of them. Your chance to have a given hit produced is x%. Now you run buildup and the buildup proposes 10 sets of 5 numbers. You pick randomly one of them for play. Your chance to have a given hit produced is better than x% because buildup GATs deliver above odds, which means among the proposed GATs there will be those that deliver more hits than naturally expected. So, at 10 randomly generated sets let's say we have 5 returned 0 hits, 3 returned 1 hit and 2 returned 2 hits, buildup GATs would be e.g. 3 returned 0 hits, 3 returned 1 hit and 2 returned 2 hits and 2 returned 3+ hits (yes these can be a 4 or 5 hit too).

[#2] - I made an extensive run a while back to gather statistics on where the best buildup GATs fall in the panorama. I actually play with buildup over well more than 20 draws and I pick 10 numbers every time for a 5/45 field. What I did is I pick a GAT that ends at the same position in a given column/row of the buildup for every draw after scanning eg 100000 GATs. Over the last 7 draws I have 6-2hits and 1 3-hit and in 3 cases all the rest were one-offs (it really pains to see that happening so often, to get soooo close!). In two cases (different draws) over those 20 draws, the GATs above and below that column/row position I pick returned a 4-hit. Unlucky so far but even the current hits production I get is well above odds but I am unlucky to the higher wins till now. Buildup excels so far at the 1/20 field I test.

[#3] - Example of momentum is what I get as results for the 1/20 field I play. I simply run buildup and gather the common numbers proposed by all GATs and sort these in order. I then simply pick the numbers I need from that ordered sequence for play. If the draw in question maintains the dynamics momentum, most GATs will propose the winning number at the same time and this number will be among the 2-3 most occurring, even it can be the top number. Thus we have a clear winner here.

Sooz
Proficient
Proficient
Posts: 246
Joined: Wed Feb 09, 2011 12:24 am
Location: Canada

Re: possible way to choose which gat from buildup

Post by Sooz » Fri Dec 27, 2013 4:41 pm

Ohhhh - how interesting. Thank you yet again for your comprehensive reply. Reminds me of years ago when I went to Las Vegas, there were "hot tables" where almost everyone was winning, and "cold tables" where the energy was depressed and almost everyone was losing. Let me try to put what you said into oversimplified layman's terms so I can think my way through it.

There will be times when the lottery is hot - or the GAT program finds it hot - at those times, more gat tables will clue into that hotness and select winning numbers. In that case, it would be good idea to play the commonly selected numbers amongst the different tables. (You might say the draw has high momentum at these times.)

There will be times when the lottery is cold - at those times, most gats will select wrong numbers anyway, so there's not much we can do about it and it doesn't matter how we try to choose numbers to play. (LOL musical analogies come to mind - offbeat, discordant, etc - but I'll just restrain myself here.)

Buildup mode is one way to jump in while it's still hot.

And of course, overall, GAT clues into the hot times significantly better than probability.
Take care,
Sooz

User avatar
lottoarchitect
Site Admin
Posts: 1457
Joined: Tue Jan 15, 2008 5:03 pm
Location: Greece
Contact:

Re: possible way to choose which gat from buildup

Post by lottoarchitect » Fri Dec 27, 2013 10:06 pm

Hi Sooz, you have the magic way to simplify what I say and describe it very nicely! Yes, I can say this is what I mean. Just a note on the above, we talk about buildup GATs, the momentum discussion is only for this category due to the way we gather these GATs. Using GATs from the augmentative/absolute has a different principle and utilization.

kweg
Regular
Regular
Posts: 34
Joined: Tue Apr 10, 2012 3:15 am

Re: possible way to choose which gat from buildup

Post by kweg » Sat Dec 28, 2013 2:32 am

Hi Anastasios

I also have been experimenting with the "buildup mode" but not to the extent mentioned above.
I mainly play Saturday night lotto here in Australia which is 6 out of 45 numbers. As my computer is not the fastest, I have been running the GAT engine to pick 3 numbers and in this way over 2 days I have managed to run 75000000 GATS. I pick my GAT and then use the "+" button to increase the number of numbers to 15. I note your comment above that " I pick 10 numbers every time for a 5/45 field". My question is this - is selecting 3 numbers initially and then expanding this to 15 later not as good an approach as initially selecting 15 numbers? Would not the final numbers in each case be the same as both are from the same GAT?
I tend to pick the GAT by choosing one and exporting it to Excel - I then look at it's hit performance over the last 10 draws and if it looks ok I use it. ( In many instances, over the past 10 draws the GAT I am looking at has multiple 3 and 4 hits and usually at least 1 five hit) Using this approach I manage to pick 3 or 4 numbers , and recently 5 numbers, but when I wheel the numbers I only occasionally get 3 or more in the one game.

User avatar
lottoarchitect
Site Admin
Posts: 1457
Joined: Tue Jan 15, 2008 5:03 pm
Location: Greece
Contact:

Re: possible way to choose which gat from buildup

Post by lottoarchitect » Sat Dec 28, 2013 10:56 am

Hi kweg, interesting approach here. If you keep getting good results with this method of yours, let us know.
My question is this - is selecting 3 numbers initially and then expanding this to 15 later not as good an approach as initially selecting 15 numbers? Would not the final numbers in each case be the same as both are from the same GAT?
The difference is, when you ask from the engine to display the best 3-numbers GATs, you'll get exactly that. This GAT has been evaluated with "best 3-numbers" in mind, therefore, when you expand its numbers selection to make this a 15-numbers GAT, there can be (and will be) other GATs that perform better if directly asked as ""best 15-numbers". If the initial 3-numbers GAT is also very good as a 15-numbers GAT, you'll see that at the panorama for 15-numbers GATs. However, rarely a GAT is good at various req.number categories, so if it is the best for 3-numbers, likely it will not be among the best for 15-numbers.
If however your approach gives good results, no reason to do something else, just let us know if its performance.

Sooz
Proficient
Proficient
Posts: 246
Joined: Wed Feb 09, 2011 12:24 am
Location: Canada

Re: possible way to choose which gat from buildup

Post by Sooz » Sat Dec 28, 2013 3:31 pm

Anastasios that's a wonderful compliment coming from you!
Kweg I had to enlarge my screen to make sure I was counting your zeros correctly! 75 Million! Do you find that the higher id's are indeed significantly better hitters? Maybe I lose patience too quickly. :-)
But now it occurs to me that, for buildup mode, it makes more sense to keep GAT going longer, if looking for numbers which are more commonly selected by different gats. For tonight's draw, the best buildup gat in my run was indeed the highest id at about 16 million.
Sooz

kweg
Regular
Regular
Posts: 34
Joined: Tue Apr 10, 2012 3:15 am

Re: possible way to choose which gat from buildup

Post by kweg » Sat Dec 28, 2013 11:53 pm

Thank you Anastasios for your prompt and very informative reply.
Following on from what you have said and in response to Sooze, yes, I did get up to 75,000,000 GATs but I was only going for 3 numbers. However, I have found using this approach that the top GATS in each column are much earlier than 75M - one was around 2M and the next one was 20M. No GATS in the 60M to 75M range were in the columns ( 20 GATs per column).
What Anastasios said makes sense so I am now running the GAT engine again, this time for 20 numbers and 4 to 6 hits ( This is for Powerball here in Australia which has 6 from 40 and a powerball from 20 numbers).
I have been running this for a day now and at present the program is up to 37,724,161 GATS. The top GATs in the 6 number column are 4,899,951 ;23,050,491 ; and 25,193,560 - the rest are spread from 5M to 37M.
I exported the GAT 4899951 to Excel and it looks very promising - the number of hits are 6,6,6,5,4,2,5,3,4 and 2. I should mention that for GAT analysis I am using Draws =10 and Data =10 - I have to do that as they changed the Powerball draw here recently and there are not a lot of past draws.
I will run the GAT engine for another day and see what I get. I only play 2 lottos here in Australia - Powerball , drawn on a Thursday and Sat Lotto drawn on a Saturday. This means I can let the GAT engine run for a couple of days in relation to both games.
I will post some more comments in a day or so.

kweg
Regular
Regular
Posts: 34
Joined: Tue Apr 10, 2012 3:15 am

Re: possible way to choose which gat from buildup

Post by kweg » Mon Dec 30, 2013 12:17 am

An update on my last post.
I have run the GAT to 75,000,000 and the top 4 GATs in the 6 number column were 71,969,049; 4,899,951; 23,050,494 and 25193660.The only difference between these 4 GATS, apart from numbers selected, was the first one had no occurrances where 0 numbers were selected in the past 10 draws whereas the other 3 each had 1 instance of a 0 in the last 10 draws. In the 6 number column the remainder of the GATS were in the lower numbers with a couple in the 30Ms and one in the 60Ms.

When I ran the first simulation ( only 3 numbers requested initially and then expanded to 20 ))I picked GAT 70,968,085 which gave these numbers - 3,11,7,35,22,28,15,4,12,6,30,1,18,33,8,38,13,40,32 and 37.

For the last simulation ( with 20 numbers requested ) GAT 71,969,049 gave the following numbers - 11,13,15,14,8,27,5,2,28,32,35,7,19,29,24,17,31,33,37 and 21.

Ten numbers from the first GAT are included in the selections for the second GAT

It will be interesting to see which GAT does better in the next Powerball draw due here on Thursday 2 January. I will post the result.

From the above results, it would seem that the majority of GATs in the Panorama are from earlier than later scans. I wonder if Anastasios can make anythying out of these results?

User avatar
lottoarchitect
Site Admin
Posts: 1457
Joined: Tue Jan 15, 2008 5:03 pm
Location: Greece
Contact:

Re: possible way to choose which gat from buildup

Post by lottoarchitect » Wed Jan 01, 2014 10:57 am

kweg wrote: From the above results, it would seem that the majority of GATs in the Panorama are from earlier than later scans. I wonder if Anastasios can make anythying out of these results?
Well, scanning more GATs doesn't necessarily mean there will always be an even better GAT at a later stage. If however there is one, you have the chance to find it. This is what we mean when we say "the more you let the engine run, more chances to find even better overall GATs". It doesn't mean that a higher ID GAT will be always better than a lower ID GAT or there will be one better at any given higher ID range - you may find one better e.g. at 100M GATs scanned. If it was possible, I'd make the engine scan in parallel low ID GATs and high ID GATs at the same time. We can't do that because each ID depends on computations performed on all its previous IDs; if we want to compute ID 2, we have to compute first ID 1. Waiting for your results and Happy New Year!

Post Reply

Who is online

Users browsing this forum: No registered users and 2 guests