Reasons For purchasing .. Great Expectations

[ophiuchus]

Anastasios ...

Thank-you for such a rapid delivery of the software license. It would be interesting to take a survey to understand how long we have all been lotto-sticians. I have been at it at least 15 years. People naturally think if you can predict lottery numbers that your power must be on demand. I believe it takes at least 10 years before you really understand what you are trying to accomplish. I have certainly bought my share of lottery tools. Nothing crazy but I became more skeptical as time went on when all the same analysis gets you only so close (Games Out/Hot Due/Cluster Hit/Cluster Skip). I have tried my luck using Neural Networks. Holograpic Neural Networks specifically. While I was attempting to use Neural Nets, I discovered something interesting that I have not seen until today which Anastasios is calling The GAT Engine. Long story short. I was creating number strings and wheeling them. However, just for kicks I always went back to see if any wheels that I created in the past where still hitting. Over the course of a summer, I had created 10-15 wheels with a total of 500 combinations. What I discovered was that in a 2 -12 game period there was a 4/5 match in my sets. Alot of the same combs that produced a 4/5 repeated down the line. So in a set of like 500 numbers there was like 20 sets maybe that kept repeating. I think I still have the data and notes. i am going back to 1993 when I was just beginning and had lots of thoughts moving in all directions. I believe GAT is that vision. I look forward to what may be discovered with your Tool ....

[lottoarchitect]

Hi ophiuchus and welcome at the forums,

regarding neural networks I believe they are not really helpful towards lottery prediction. I'll try to explain this as simply as I can.
A neural network has the benefit it needs no defined rules to provide an output. That particular strength it has makes it very suitable to identify "close to" behavior of well defined patterns.
An easy example to understand the above statement: consider the letter A. We can write that down in various ways such a different width of lines, a bit curved lines, perhaps a bit different length of lines etc. The important "keyword" here is the word "a bit". No matter how we actually write down the letter A, it will always be an A as long as it has those 3 lines intersect like that. Those 3 lines is actually the pattern defining the letter A. Neural networks are very good at taking advantage of patterns. That is marvelous and has very advanced utilization in cases such as an OCR program. Training an OCR network to understand the letter A is basically the procedure of feeding the network with various representations of the letter A and "force it" to respond with the letter A as output. This training attempts to recompute the internal weights of the nodes of the network so they altogether can respond correctly and produce as output the letter A if fed with a similar representation of A to those used to train the network.
Enough of the training part, let's check what happens in the lottery case. Can we apply such a training process in lottery? What do we need? Defined patterns obviously to do the training. What else needed? Defined output for the defined input. What is our input? A sequence of history draws. What is the defined output? The draw following that sequence[*][/b]. The case here is:
1) we assume that a given sequence of draws always produces the same next draw.
2) we also expect to encounter a "similar" sequence in the future so to get back a "similar" outcome.

Case (2) will really never happen. What means a "similar" sequence in lotto draws anyway? I can't define this really! I also consider (1) as inapplicable - based on what? Finally what is the waiting time to have a "similar" sequence happening again supposedly to take advantage of? I doubt we'll ever encounter such a case to take advantage of (always assuming the (1) is true). Additionally the assumption (1) implies defined patterns that are obvious and unchanged. Can we say that for lotto draws? I'm not so sure we can certify obvious and steady patterns can exist in there. Based on the requirements of neural networks, I consider them insufficient to do something useful in lotto draws. They are well suited for many other tasks but in my opinion not for lotto draws.

[*][/b] Of course someone can say, my input is not the actual draws. He could choose some property to evaluate i.e. sums, lowest numbers etc. This doesn't change the problem at all. The assumption that a given sequence of data always must produce the same next value is the BIG pitfall of neural networks when applied to lotto draws.

cheers
lottoarchitect

[ophiuchus]

Which of course is why I have not gone the distance on pure Neural Networks Alone. I follow your reasoning arguing against NN as a solution. Lets discuss that some other interval. Your tool is what is important now. And I have been running Tables all day. I will not give an opinion left or right until like August of this Year. Gives a chance to make some serious GAT runs before any serious tweeking. This is the GAT string I am scoping at the moment. First one off the hopper. I like the hit pattern all ready. No planned wager with the string.

1 3 8 9 11 12 13 16 17 19 20 21 28 29 31 32 35 36

Keep in mind. I have other methods for picking numbers. But If I can get savvy at using the tables. You might get a better sense of what to keep and what to throw out ...

[lottoarchitect]

Looking forward to your results ophiuchus. Just make sure you understand how to properly use the run factor for subsequent future runs of a particular GAT table so to get correct predictions every time. This has been asked a lot of times around here so it might be beneficial to read a few other posts on this matter.

[ophiuchus]

Result
03/26/2013 - 4,17,22,25,31:b24

1,3,8,9,11,12,13,16,17,19,20,21,28,29,31,32,35,36 - GAT String (2hits - 17,31)

Combs played in last nights game:
4,14,22,23,25 - 3/5
4,14,23,25,28
4,14,20,23,27

It was to soon for me to alter my string. BUT the GAT hit trend looks good. Not sure how to insert images but I would show you the hit chart. GAT table has been running since yesterday. GAT Table 10,900 had the best graph out of the 5 tables listed for 5 hits. Something happened that I dont understand. I am at GAT 4,085,984. But my other tables are gone from the panorama view. In any event, I am sticking with current string.

Now, I have a new data point. If I take into consideration the Run Factor, That would give me a new string. In theory though, I would not look for a new GAT Table unless I am ready to abandon the old one. I just need to make sure I update Database before I run. just making sure I am on track.

[lottoarchitect]

But my other tables are gone from the panorama view. In any event, I am sticking with current string.

Perhaps you just browse the lowest part of the panorama? The right below the top-left cell of the panorama you should see "1", if not, you view a lower part of the panorama. Otherwise, I cannot understand what you see there. You should always have at least one GAT table shown in every column as soon as one has been registered.

Combs played in last nights game:

Are these with your own picking method?

Not sure how to insert images

Use an image sharing service such as http://imageshack.us/. Its free and better to make an account there to have access to the uploaded images (add/remove them). Then get the "direct link" indicated by the service and use it with the IMG tag in the forum editor. This will show the image right where you place it. Alternatively, get the "indirect link" of the image indicated by the service and use the URL tag in the forum editor.

I hope you don't use a very small tested draws value. You have to use the run factor if you want to continue the same GAT table at the next draw.

[ophiuchus]

A...

Day 2 of the Mission...

Just got the results. And yes I do use a hybrid of lotto techniques to choose my numbers. A few of my own Quasi automated analysis techniques. Meaning, once I prep the data I can run the analysis. Which is why I missed tonight's operation. I keep my G.A.T on the target though. Hopefully, I am using the tool correctly but the G.A.T table has already delivered. Although I have not decided on my next chess move. Here is what happened tonight.

3/27/2013 - 1,17,20,28,31:b-11

My GAT string: 1,3,8,9,11,12,13,16,17,19,20,21,28,29,31,32,35,36 [4 Hit - 1,17,20,28]

Who is ready to get off this train or do we believe in this string. In this instance we are tracking to the Hits Chart of Table 10900.

I see two paths. I can test to see the success of finding a new table in time for the next event. I could test to see how stepping with the result hit. By The way there is a 5 hit waiting to happen. I need to hit one of those to really be in the money. You can make money with 4. You can even make money with 3.

I am for grins going to take the string and run it through a few wheels to see how I could have fared...

[lottoarchitect]

Hi ophiuchus,

I understand that you keep the initial prediction of GAT. Sometimes this can deliver a good hit and it is known as temporal effectiveness but my feeling is it does not deliver as many hits compared to the designed manner (with the run factor). However the statistics and graphs are computed by the means of taking the new prediction of a GAT table (using the run factor) for every new draw. A detailed example on how to continue using a particular GAT table so to continue its illustrated performance:
viewtopic.php?f=23&t=587

Another post on explaining the run factor usage
viewtopic.php?f=23&t=610

Also what is the game played?

[ophiuchus]

A...

Of course, This is my learning curve. I am playing the MD0539(bonus). My next attack is probably Friday. I am trying to catch up on my data collection and analysis. Stay tuned.

[lottoarchitect]

3/27/2013 - 1,17,20,28,31:b-11

My GAT string: 1,3,8,9,11,12,13,16,17,19,20,21,28,29,31,32,35,36 [4 Hit - 1,17,20,28]

I see here all the numbers predicted including the bonus ball, it looks a 5+1 hit to me, not a 4 hit