"GAT ablation"; reducing the GAT size, reduce the wheel cost

G.A.T. Engine general discussion
Post Reply
Starbreaker
Casual
Casual
Posts: 12
Joined: Wed Nov 19, 2014 3:41 am

"GAT ablation"; reducing the GAT size, reduce the wheel cost

Post by Starbreaker » Thu Feb 05, 2015 7:50 pm

Hi everyone, first time poster (but a long time lurker!) here :)

As we all know, the more numbers we request from a GAT, the higher the hits percentages we get. However, requesting more numbers will also increase the costs of wheeling since more tickets are needed to cover larger amount of predicted numbers. If we request 20 prediction numbers out of 45 in the Greek Tzoker game, the GAT will certainly be more likely to score a 3-hit, 4-hit (or even a 5-hit!) than, say, 10-15 predicted numbers. In fact, in the Tzoker, with 20 requested prediction numbers, you've got 33% chance at a 4-hit and 10% chance at a 5-hit. Still, no matter how many numbers we request, at the end of the day only up to 5 of those numbers will hit. Or up to 6, in the case of pick-6 games like Lottario, NY Lotto or the various popular 6/49 games from around the world. So even if our chosen GAT with 20 prediction numbers scores a 5-hit, that doesn't change the fact that the remainder 15 numbers didn't hit. Now, there's no way for us of knowing ahead of time what those 5 winning numbers will be. If we did, we'd all be richer than Croesus by now. However, there is a way of finding and identifying at least some of those 15 "loser" numbers (that is, the numbers that won't hit during the target drawing date). If one were to find out what those loser numbers are, and eliminate them from our set of predicted numbers, we can then ablate, or reduce, the overall size of the pool of numbers that we'll wheel. All without reducing the hits percentage and the effectiveness of our selected GATs. That means if we can identify and eliminate from the play even just 3 loser numbers from a set of 20 predicted numbers, we can then wheel and play the other 17 predicted numbers. All while maintaining the same 33% for 4-hits and 10% for 5-hits percentages of the 20 predicted numbers. There's been discussions here in the forum about finding winning numbers using the Countif function of the MS Excel. Well, for the past 6-7 months, I've been doing something similar, counting the frequency of numbers. Except in my case, I done it in order to reveal the future loser numbers, rather than future winning numbers. I've been using this loser numbers detection method for the New York's Take-5 (a 5/39 game), where I've regularly achieved 3-wins and even few 4-wins.


For this forum, I will use the Tzoker game's 5/45 main numbers field as a demonstration. The tonight's 2/52015 drawing being targeted.
For the process, I set the GAT Engine to display 100 tables per column, (though you can give or take a few tables, if you wish). As opposed to the default 19 tables per column. Make sure the RF is 0. We can disable the Absolute mode to help speed up the scan a little but, since we won't be needing it for this method. Only Build Up and Augmentative modes are needed. I then let the engine run a scan until it reaches 3-4 Million GATs. Then (while in Augmentative mode) we request the maximum amount of prediction numbers possible. Which in the case of Tzoker would be "44 numbers requested out of 45." Next, we highlight the entire panorama. You heard me right. Not the individual column. But the whole thing, hook, line and sinker. With all the columns under all of the 1 through 5 hits categories.
Like this:

http://postimg.org/image/lpo0f4qd3/

You can then export all of the tables to the Countif. Or, you can use this free Wordcounter function on this English Grammar.Org website, like I do

http://www.englishgrammar.org/wordcounter/

Make sure you click and disable the "exclude common words from word frequency count", before you paste the tables into the box and hit the "count words button"

This is what you will get:

http://postimg.org/image/c8ifh72fh/

As you see, there's two separate columns. One on the left, labeled "primary keywords" and the one on the right labeled "common words." Ignore the left column, as that column only counts the frequency of GAT ID's and not the predicted numbers. The right side column is the one you should pay attention to, since that's the one that actually counts the frequency of predicted numbers themselves. Ignore the most frequently appearing numbers on the top, and scroll down to the very bottom and see which predicted number is the fewest appearing one

http://postimg.org/image/yhp8bfq19/

As we can see here, #38 is the lowest appearing number, and therefore it will not hit tonight.
Once you're done and you've identified your loser number (#38 in this case), write it down. Then go back to the GAT Engine and lower the requested numbers down to 43 out of 45 and repeat the process again.
Here's what I got:

http://postimg.org/image/6myjhqosz/

Make sure the whole panorama remains highlighted. I highlighted which prediction number appears the lowest amount of times (#32), and which one is next-to-lowest (#21). Also notice the difference between the two numbers' frequencies themselves (464X vs 462X). Pay attention to this gap. The bigger the gap between the lowest and next-to-lowest numbers, the better. It further reinforces the lowest numbers as that drawing night's loser number. Which you can then deduct from your group of predicted numbers. However, if there are two or more equally lowest frequency numbers, such as this....

http://postimg.org/image/5rhuthbt1/

...it's best to ignore it, and move on.
From my experience, with the various lottery games I've experimented with, it's best to work with with the last 4 permissible amount of requested numbers. Which for example in the case of New York Lotto would be the requested numbers 35 trough 38 (out of 39 total numbers). And for Tzoker, would be the requested numbers 41 through 44 (out of 45 total numbers). Anything lower than that, and it becomes unstable and even the lowest of low frequency numbers may or may not end up hitting. After repeating the process until I reached 41 requested numbers, I've found that #11 won't hit in tonight's Tzoker game as well

http://postimg.org/image/qd3p3zfdd/

Once we're done with the Augmentative, request 44 numbers once more and switch over to the Build Up mode

http://postimg.org/image/51os6f1ub/

And export all the 200 tables under the 4-hits and 5-hits categories, and run them through Countif or the keyword density website

....which will give us two lowest numbers (45 and 6) at the same time

http://postimg.org/image/8inqhk4g9/

In this case, we can either let the scan run for another million GATs, and see if these two numbers will diverge from each other after more GATs have been run through the engine. Or we can leave it alone, and try again for the next draw, hopefully get a gap between the lowest and next-to-lowest numbers next time. And if we find us a loser number, we write it down and see if it appears amongst the predicted numbers of our chosen GAT table, and which we then subsequently eliminate from that night's play.

There are more techniques that I use, for my local state's Take 5 game. However, since this is the first time I've run a scan for the Joker lottery game, I'm only sticking to these limited methods, until I try more of my other methods for future draws and see if they work as consistently for Tzoker as they do for Take 5. As I said before, by ablating the large GAT tables into smaller ones, I've consistently won 3-wins and even a couple of 4-wins with the New York Take-5 game. Like this 4-win, worth $400, from a few moths back

http://postimg.org/image/6bcp1w40p/

Since Tzoker is also a pick-5 game, I believe that this method can work for it too :D

Starbreaker
Casual
Casual
Posts: 12
Joined: Wed Nov 19, 2014 3:41 am

Re: "GAT ablation"; reducing the GAT size, reduce the wheel

Post by Starbreaker » Wed Feb 11, 2015 8:26 am

Update: The winning numbers for Thursday, 2/05/15 were; 13-30-35-44-45. Just as expected, none of the decisively proven loser numbers (38, 21 and 11) hit that night. Encouraged by this success with Tzoker, I ran a new scan and following the previously described methods, found these following loser numbers that won't hit this coming Thursday's 2/11/15 drawing - 1, 14, 28, 39. These were obtained from Augmentative mode. From the build up mode, I got the number 25. Again, these are the numbers that I'm betting on NOT being drawn tonight, on Thursday. And if I had any of these numbers in my GAT of choice, I'd remove them.

User avatar
lottoarchitect
Site Admin
Posts: 1635
Joined: Tue Jan 15, 2008 5:03 pm
Location: Greece
Contact:

Re: "GAT ablation"; reducing the GAT size, reduce the wheel

Post by lottoarchitect » Wed Feb 11, 2015 11:57 am

Keep us informed of the results of this approach. You did create a deep scan methodology to withdraw numbers and sounds very interesting. Perhaps it would be nice to write an overall performance summary of this method, i.e. how many draws tested and how many times this brought up the desirable outcome. It may be possible to use this method as an internal process to the Lite version of GAT. Keep up the good work! ;)

Starbreaker
Casual
Casual
Posts: 12
Joined: Wed Nov 19, 2014 3:41 am

Re: "GAT ablation"; reducing the GAT size, reduce the wheel

Post by Starbreaker » Thu Feb 12, 2015 4:09 pm

Anastasios,

I can assuredly tell you that this ablation method worked and always works with the local New York state games, Take-5 and NY Lotto, anytime and every time. When I first tried out the method for each of those games, I tested for 20 drawings in a row. Came through every time, as not one of the revealed loser numbers hit during each target drawing date. And thus could have been safely and reliably withdrawn from the selected GATs and wheels, without the worry of accidentally cutting out a winning number(s) along the way. I have stuck with the ablation method ever since whenever I play Take-5, my personal favorite game. In fact, for the Take-5, I run two concurrent scans targeting the same draw date. One with GAT Engine set to display 100 tables per column, as usual, and the other with 200 tables per column. With 200 tables, I actually discover even more loser numbers (at least 3 or 4 of them). I haven't tried that other latter part with Tzoker though. So far, my method has passed 1 drawing successfully, from last Thursday. So far, 1 drawing down and 19 more to go. We'll see how well it does tonight. I'd especially keep an eye on the loser #'s 28 and 25, as they were both obtained via requesting 44 predicted numbers out of 45, in Augmentative and Build-up modes respectively. And if scanning with 100 tables per column proves to work with Tzoker, we can then try 200 tables. Already, by now I can reliably find, identify and remove 10 or more loser numbers with Take-5. Perhaps the same can be achieved with Tzoker :)

Starbreaker
Casual
Casual
Posts: 12
Joined: Wed Nov 19, 2014 3:41 am

Re: "GAT ablation"; reducing the GAT size, reduce the wheel

Post by Starbreaker » Fri Feb 20, 2015 2:41 am

Update: I did another scan for Tzoker's drawing tonight, on 02/19/2015. The ablation method yielded these following loser numbers.



in augmentative mode;
w/44 requested numbers: 43
w/43 requested numbers: 26
w/42 requested numbers: 24, 33, 37 (all three of the loser numbers equally appeared 450 times each).
w/41 requested numbers: 33

Now in build-up mode;
w/44 requested numbers: 32

In addition, there's another method that I've used successfully for New York Take-5, and which I also want to try on Tzoker. For this method, you must request 44 predicted numbers in augmentative mode, and allow the GAT Engine to run until it accumulates at least 20 GATs that have 100% 5-hits guarantee. Then under the 5-hits column, highlight and export all of the 20+ GATs with 100% guaranteed 5-hits. Like this

http://postimg.org/image/7c77o0lzx/

Run those GATs through a Countif function, or the Word Counter website to see which predicted number appears the least. In this case, this is what I got

http://postimg.org/image/4jtt6xe3t/

As we can see, #1 is the least frequently appearing predicted number here. So we can add it to the list of tonight's decisive loser numbers, and see how it works.

Total loser numbers: 1, 43, 26, 33 and 32.

lottoburg
Knowledgeable
Knowledgeable
Posts: 117
Joined: Sun Sep 18, 2016 11:06 pm

Re: "GAT ablation"; reducing the GAT size, reduce the wheel

Post by lottoburg » Mon Nov 28, 2016 2:19 am

Hi, Starbreaker:

I like your smart strategy about GAT Ablation.
However, I have two questions as below:

1) You scan the process from 38-39, 37-39, 36-39 and 35-39 for Augmentative mode but why scan 38-39 only for Build Up mode?

2) As you know it's time-consuming to scan the GAT Engine for getting 3-4 million GATs usually.

In the settings of GAT Ablation, you using by:

a. 100 Gat Ids/each column;
b. Draws=100, Data=20;
c. 35-38 requested numbers out of 39;
d. Run the Engine 6 times at least: Getting prediction = 1 time;
Getting ablation=5 times including 4 Augmentative & 1 Build up.

But NY Take 5 is a daily game. I'm afraid that maybe we don't have enough time to run the process above. Could you let me know how can you handle the problem easily?

Thanks for your nice sharing and sincere help!!!

Best regards,
lottoburg

Starbreaker
Casual
Casual
Posts: 12
Joined: Wed Nov 19, 2014 3:41 am

Re: "GAT ablation"; reducing the GAT size, reduce the wheel

Post by Starbreaker » Sat Dec 03, 2016 8:21 pm

lottoburg wrote:Hi, Starbreaker:
1) You scan the process from 38-39, 37-39, 36-39 and 35-39 for Augmentative mode but why scan 38-39 only for Build Up mode?
Because that's what works best for the Build-up Mode, from my experience. When you request the maximum amount of prediction numbers possible, it also brings out the highest chances for predicted numbers to contain 100% 4-wins and even for 5-wins. From what I've seen, when you're working with Build-up mode to find loser numbers, the GATs you collect have to be a sure thing. Anything less than 100% guarantee is not going to work.

lottoburg wrote:Hi, Starbreaker:


2) As you know it's time-consuming to scan the GAT Engine for getting 3-4 million GATs usually.

In the settings of GAT Ablation, you using by:

a. 100 Gat Ids/each column;
b. Draws=100, Data=20;
c. 35-38 requested numbers out of 39;
d. Run the Engine 6 times at least: Getting prediction = 1 time;
Getting ablation=5 times including 4 Augmentative & 1 Build up.

But NY Take 5 is a daily game. I'm afraid that maybe we don't have enough time to run the process above. Could you let me know how can you handle the problem easily?

Thanks for your nice sharing and sincere help!!!

Best regards,
lottoburg
Actually, I usually run the GAT Engine scan only twice: first time in order to obtain the "loser numbers", and the second time in order to find the best GAT that offers the highest chances at 3-wins and 4-wins, while containing the most amount of "loser numbers" to withdraw from. Also, in my experience, in order to find loser numbers, 1 or 2 million GATs scanned has been sufficient for me. I certainly didn't have to do 4 million.

lottoburg
Knowledgeable
Knowledgeable
Posts: 117
Joined: Sun Sep 18, 2016 11:06 pm

Re: "GAT ablation"; reducing the GAT size, reduce the wheel

Post by lottoburg » Mon Dec 05, 2016 3:54 am

Starbreaker,

Thanks for your sincere reply and smart instructions!

I really appreciate your amazing strategy of "Withdrawing loser numbers ".
It greatly enriches our mind and I believe firmly it works very well for GAT Engin.

I have two problems about your Strategy as below:

1) To speed the process if we can run the 5-5 or 1-1 only instead of 1-5 since their GATs are same things?

2) To get more loser numbers safely and fast if we can try the different settings such as Draws & Data, ... etc?

"With more new methods discovered, then perhaps I can reliably identify up to 30 future "loser numbers." As time goes on, maybe it will be possible to identify up to 70% or 80% of the "loser numbers" from the total lottery numbers from any pick-5 or pick-6 games around the world."

About two years have been passed since you discovered the smart strategy, could you share some new methods with us here?

Thanks for your nice sharing and sincere help again!!!

Best regards,
lottoburg

Post Reply

Who is online

Users browsing this forum: No registered users and 27 guests