Back to basics

[Sooz]

With GAT's help I had good luck back at the beginning, using the defaults for draws to be tested and for total stats to be used, ie 100 and 20 respectively. Then I tried various approaches all with bad luck.
(Maybe "luck" is a poor choice of words. But "randomness" implies "luck". So maybe it's an appropriate word. Maybe we people ourselves go through cycles of luck.)
So I'm going back to the defaults. But this time I'll use min/max of 12 instead of 6, and will stick with this method for now.

.....Sooz

[tucker]

I am having the same experience here as well. In the beginning I tested GAT with the defaults and no avail. Then I used 20 data and 20 draws. It was if I could do no wrong. For 3 draws in a row the hits were piling up. On the one draw I made a nice hit with 5/6 numbers on the 6/49 and scored a $2600.00 win which was real sweet. Since then GAT has been dead. I run GAT constantly using different configurations and nothing special has shown up yet. I play 3 different lotteries with GAT and have only won on the one.

[lottoarchitect]

Hi tucker, it is the first time you mention such a win! Congrats on that, can you also give some more detail on this? Cost played, parameters used, what did you look for to pick GAT etc? Also, regarding hits over the long run, there will be dead times obviously. It is virtually impossible to expect a win on every run, we still deal with random events to a large extend, we shouldn't forget that! Actually, I like to think it the other way; most users have GAT over the last 4 months or so. Most already got 5 hits and several 4 hits. Can this be compared to any such prior-GAT experience? How often a user had a 5 hit in the past? If GAT manage that in 3-4 months of time, it will do that again within a similar time span in the future, maybe even sooner. Even consecutive 5 hits are possible, the system can produce that. What is needed is a steady approach to the game, not swapping strategies every so often. There isn't any strategy that can deliver on every draw, the point is to stick with one that seems to perform well and use it all the time.

cheers
lottoarchitect

[tucker]

First off, my thoughts towards randomness in regards to the lottery has become a fleeting word in my vocabulary during the past couple of years. My observations tell me that "randomness" appears to have some sort of logic towards its goal irregardless of the dynamics of the game. Predictive processes are most definitely the called for action when it comes to this particular game of chance and thus the reason I was drawn toward GAT.

I have been testing GAT with numerous different perimeters. You name it, I have been testing it. This computer hasnt slept a wink since I purchased GAT...LOL. I wish I could run GAT on more than one system, it would sure speed up the testing and seeking process. Anyways, regarding the sweet 5/6 win, I used 20 data draws and 20 testing draws and called for 12 numbers (I have noticed that quite often GAT will pick out nice number sets early in the process and then dump them for other number sets that just do not hit. My best GATS have come from running the program for no more than a couple of hours). From what I understand, the longer you run the program, the better the gats...unfortunately, that is not my experience. When reviewing the GATS, I looked for an upward trend in the graphs (blue line) from the beginning (sharp upward climb if possible). The best GATS always seemed to appear in the 5 number category and what made the GAT an easy choice was the fact that there were usually no more than 2 or 3 to choose from. If there were more GATS in a particular category (5 or more), their predictive ablility was degraded. Once I had some GATS chosen , I reviewed them and chose the ones that had the best hits. If there were still more decisions to be made, I would narrow it down by looking at the hits delayed graph. The one that had the most numerous delays, with 4 or 5 called for numbers, that would be the one I chose. As I said earlier, I have have been looking at numerous configurations, but this is the particular one that gave me the sweet run of wins along with the 5/6 win. Oh yes, and the lottery data was updated with every draw.

After chosing a GAT, I would have 12 numbers. I would then wheel them with WG and make 10 tickets. Tickets for the 6/49 here are 2 dollars, so that is 20 bucks. Nothing special was used for filters other than sums set at 70-250 and history compare. Everything else was set to default settings.

So ya, after that kind of success (which I know was not a fluke) and nothing has happened in a month and a half, you have to wonder...gee, did I break it or what???

[Sooz]

Tucker that's great - and I totally understand the frustration of having a reasonably nice win, then nothing for a couple of months!

I'm now continuing to give GAT 100 total real draws to be tested. One table just came up which has four 6-hits when 12 numbers are selected. What's interesting is that all 4 of those 6-hits are evenly distributed in the 2nd 50 draws. If I had used only 50 total real draws to be tested, that would have been 8%. But it's only 4% of 100 draws. So do I assume this GAT has become hot for 6-hits?

Also, again, this particular GAT with those 4 6-hits, has zero 5-hits in the past 100 draws. (I still think it's weird but Anastasios in another post says it's part of the awesome unpredictability of randomness - I'm paraphrasing him but I believe that's what he meant, right, LottoArchitect?)

Although it's tempting, I won't play this GAT but will monitor it for awhile. I'll continue to play a gat with lots of 4-hits - maybe - LOL.

.....Sooz

[lottoarchitect]

We should be realistic on our expectations. True, GAT can beat random events by far, however it can't beat inherent randomness to the point having one good hit on every draw. Matching 5 correct when picking 12 numbers in a 6/49 game has 1/477 odds, which means, we should play 477 draws, to have once a 5 hit when picking 12! In normal lottery game times, this can be once every 4.5 years (drawn twice a week), or once every 9 years if drawn once per week. The fact that GAT has already gave a few 5 hits to various users within those few months being available to public, plus some nice consecutive 4 wins too in some situations, tells a lot about its ability to actually predict numbers. With normal utilization, GAT can give a few 5 hits out of 12 picked within a year. This is already outstanding performance. The fact that it may didn't deliver over one month of usage a similar hit, or even a lower one doesn't mean something really, even a 4 hit outof 12 picked has about 2.5% chance, or should occur once every 6-12 months. The time you trap a good GAT, you can have a sequence of good hits, then till the next time a good GAT is trapped (which will deliver a similar sequence), this time seems dead. It isn't dead really, there is a GAT somewhere which hits, the problem is to find this one GAT. More tools will come in GAT towards this GAT seeking anyway. Still GAT seems to be very powerful even with its current implementation, given the wins already mentioned by some users over this very short period of time GAT is available to public.

cheers
lottoarchitect

[lottoarchitect]

Adding some more info to the discussion above.

Also, again, this particular GAT with those 4 6-hits, has zero 5-hits in the past 100 draws. (I still think it's weird but Anastasios in another post says it's part of the awesome unpredictability of randomness - I'm paraphrasing him but I believe that's what he meant, right, LottoArchitect?)

When we pick 12 numbers in a 6/49 game, we have the following odds:
1 correct: 1 in 2.67 or ~37 1-hits in 100 draws.
2 correct: 1 in 3.21 or ~31 2-hits in 100 draws.
3 correct: 1 in 8.18 or ~12 3-hits in 100 draws.
4 correct: 1 in 42.42 or ~2.35 4-hits in 100 draws.
5 correct: 1 in 477.2 or ~0.21 5-hits in 100 draws.
6 correct: 1 in 15134 or ~0.006 6-hits in 100 draws.

The above is the normal odds of random play and what we can expect in terms of hits. So, we should normally expect around 37 1-hits, 31 2-hits, 12 3-hits, 2 4-hits and none of 5 or 6 hits. GAT beating odds would just provide a few 5-hits, lets say with 10 times better odds performance, it would deliver 0.21 * 10 = 2.1 5-hits over those 100 tested draws. With a similar odds boost, we would just have 0.06 6-hits (which means 0 6-hits), so even with a 10 times boost we wouldn't expect a 6-hit at all. So, a normal odds boost would just give a couple 5-hitters only. The magic here is that GAT works at least one order of magnitude better and instead of delivering those 5-hits, it produces instead a couple of 6-hitters, where we would normally expect only 5-hitters. So, GAT indeed defy normal distribution of random events odds and produces way better hits performance than naturally expected even with a 10 times boost of odds. So, your example really demonstrates that GAT can actually trap even more numbers than we would naturally expect with a given boost ratio already achievable by the engine. The 5-hits we would expect by that boost, are actually 6-hitters. The fact that this hits performance defy normal chances and it does not produce the normally expected odds distribution is evidence of inherent and focused prediction performance derived by the engine's algorithm; this is the magic part.

tucker, you say that you find good GATs at the early stages of the computations. Have you tested those later GATs in a long run (with the run factor) or just testing the performance of hitting at the very next draw? GAT should be utilized with the X/100 rule in mind. There can be good early GATs hitting but overall the later GATs are more steady in the wins production.

cheers
lottoarchitect

[tucker]

Yes, I have kept track of all of the GATS I have chosen. Only twice have any of those GATs hit a draw or 2 down the road. All of the others have been non productive which includes the x/100 gats I have logged just to keep track and compare.

And yes, while experimenting with the GAT engine, I have on many occassions found gats that produce early in the computation process...but you need to be quick to catch them. And when I mean produce, they may not be long time hitters but will produce on the draw you are computing for. Quite often it will pick out 3-4 numbers and picking 4 numbers from 12 is pretty dam good I might add. In case anyone is wondering why i call for 12 numbers, it is because it seems to give me the best bang for my buck if I wheel them into 10 combinations.

Nonetheless, I am still experimenting with this new tool in my lottery arsenal and I am open to suggestions as I am always reading the forums to see what others are doing.

[lottoarchitect]

It turns out, given the comments made by the users a sweet setting is to look for around 12 numbers and wheel them. So far, most wins announced use this approach and are still within budget I believe. Fewer numbers will not produce often a good amount of hits and more numbers get out of budget control.
About early GATs performance, I have to say that all GATs have the ability to give good hits in the long run, at least those that show at the panorama even if they are replaced later by better GATs (hit-wise). However these initial GATs have a lower total hits count, therefore skipped to favor later GATs that produce more hits overall. True, they can hit at the very next draw but I can't believe they perform better in the long run than later GATs. Anyway, I understand the frustration caused by that and what I plan for the next version is to keep track of all GATs in descending order hit-wise. That means, at the top there will be e.g. GATs that produce the most hits at a given category, lets say the engine found 3 such GATs, then at the remaining cells in that column will be shown GATs that deliver one less hit, if cells still remain empty to show the next lower producing GATs etc till all the cells are used up. The idea for this approach is that, although the top performing GATs return the most hits overall, they may indeed are in a cold cycle however there will be GATs with one less hit count that are in a hot cycle. So, there will be shown the best GATs overall to decide and I am sure you'll find there equally and even better performing GATs to those initial ones. The principle is simple here: they hit more than the initial GATs so they owe to deliver more hits anyway and among them, there will be some hitting at the very next draw. I believe most of the time, the user will decide on a GAT shown at the 10 first cells.

cheers
lottoarchitect