another possible method how to use tables

[Sooz]

Hi,
Maybe someone can comment on the pros and cons of this simpler method.
I was planning to select a table under "20 numbers requested out of 49" and to wheel those 20 numbers with my filters in WG.
But this just occurred to me. At the moment, GAT is showing 22 tables under "6 numbers requested out of 49". (It's still running in the background so there will probably be more than 22 tables by the time it's done.) Why not just use each of those tables on a ticket - ie buy 22 tickets? Or buy even fewer tickets buy selecting only a few tables from those 22 shown?

.....Sooz

[draughtsman]

Hi Sooz,

Those tables 6 from 49 will have a hit rate of perhaps 2 or 3 numbers thus within your 22 you are quite likely to achieve a similar result plus with some luck thrown in you may get several more numbers in any one of those 22 tickets. I have been exploring also with just 5 or 6 numbers and looking for just 3 GATs where there is good chance the next draw will give two numbers only but then combining these numbers in a Matrix combination in WG - where you will achieve 2 numbers from the first GAT, 2 from the second GAT, and 2 from the third GAT. There is the possibility of a few numbers being common across the 18 from the three GATs so the wheel size is not all that large and then with your filters and guarantee settings you can produce a suitable ticket size within your budget.

I like these suggestions coming forward - between us all here I am sure we shall arrive at several worth while methods of employing this amazing GAT tool.

draughtsman

[Sooz]

Hi Draughtsman,

I remember reading about that method somewhere on this blog. My concern is that each table maybe represents a cluster of numbers that should stay together. Maybe that's why Anastasios uses the terms attractors and repellers. So maybe it defeats the purpose to mix tables. But obviosly I'm just speculating.

By the time the GAT engine finished testing about a million GAT's, for "6 numbers requested out of 49" there were 49 tables under the 6 hits column, no longer 22. But they were still at only 1.00%.

So I came up with yet another simple way to choose a table. The most difficult for me was to decide how many numbers to request. Finally I decided on only 15. I chose 15 because, after staring at the charts of the tested draws, the red line, the blue line, it seemed that, by the time I got to 15, there were significantly fewer 0-hits.

Anyhow, there were only 5 tables to choose from in the 6-hit column.
So I examined each of those and chose the ones that had the most 5-hits. This narrowed my choice down to 2 tables. From those 2, I selected the one with the most 4-hits. (That's all there was to it - it now seems like a no-brainer.)
Also if I squint I can see a wavy pattern in the blue line. The blue line for that table isn't quite at the bottom of the hill, but it is at a level from which it has turned up.

Right now this seems like a pretty boring blog entry, but if I win a nice prize, I'll happily add even more details.

.....Sooz

[lottoarchitect]

Generally the proposed numbers have to be treated as a cluster - altogether. Sooz, your thinking on requesting 6 out of 49 is viable but there lurks an underlying issue which comes down to the regularity of hits. Indeed those tables shown there have hit once the winning combination over the tested draws. The problem is we don't really know when they'll produce that hit again. Even if the improvement ratio is e.g. 1000 times better, considering that a 6/49 game has 1 in 14 million, a 1000 times better performance will bring this down to 1 in 14000 draws. Of course at any given point, if a table is to give the winning combination, then most, if not all others, in that category will not give that hit on that tested draw, thus the need to play the proposed numbers of all GATs displayed. Theoretically, assuming a 1 in 14000 chance to produce that hit again (1000x improvement), if we keep playing these predictions of all those GATs played (lets say we play 50 GAT tables shown), we'd hit the winning combination 50 times in the next 14000 draws. Of course the problem here is the average waiting time of 14000 draws - or till the first of those 50 wins come - probably a person's lifetime is not enough for this. This is basically the reason why we can't really use those tables shown at the max hit of requesting 6/49 numbers when only one hit is produced. If the hits are 2 or more during the tested draws, then we can have faith a winning combination hit will occur again over the future draws equal in amount to tested draws. So if for example we test 1000 draws and see 2-3 times the winning combination produced, over the future 1000 draws we can expect this table to provide that hit again at least once. Under that condition, it makes sense to keep playing the numbers proposed by those GAT tables, 1000 future draws might be 5-10 years waiting time however. So it turns out that until we can have a GAT table that produces at least twice the winning combination over e.g. 100-200 tested draws only, keep playing those numbers 6/49 suggested will not yield any benefit under normal circumstances. I try to improve the hit performance of GATs but I can't tell if this will be ever possible even if GATs can generally give the winning combination with the bare minimum of numbers. Combining GATs that provide an exact hit count e.g. 2 hits most of the time and generate a matrix covering targeting the jackpot however is much more applicable and realistic plus the benefit that you'll get also hits of higher divisions during the run till the jackpot comes which is not the case of hitting 1,2 and rarely 3 correct when playing those GATs shown at the max hit 6/49 category. If you ask me, why those GATs there show that behavior of hitting mostly 1-2 correct only most of the time, I really don't have an answer to that especially when compared to other GATs that regurarly give good hits (but not the winning combination).

lottoarchitect

[Sooz]

Hmmm. So many things to consider. I now understand it's not a good idea to choose a table from the 6-category column when the ratio % is low.
I've already bought tickets at once for the next 4 draws so I have 2 weeks to decide how to choose. Tonight had no winners.
The GAT engine is now running again with the latest winning ticket added. It seems to me that, overall, the ratio % numbers are already much better than last time. Could this be because everytime GAT runs, it gets smarter?

.....Sooz

[lottoarchitect]

The problem really with the 6/49 highest category is that till the winning combination hit again, we generally have 1-2 hits produced which do not provide any wins, neither the trends shown give us any clue when to expect that hit - it looks like the good hit comes out of nowhere. The above analysis is a theoretical estimation of what to expect if we decide to use GATs from that category, it may well produce again that hit much sooner. We really need a huge test history to see exactly the performance of winning combination hits but generally this can't be expected in only just 1000 tested draws to hit again (that would be millions times better performance!). All GATs have the potential to produce the winning combination sometime in the future, it just happens they do not produce that during the tested draws so to be shown at that hits category - and of course we really can't tell what sort of GATs will emerge if we ever manage to go way beyond several million GATs analyzed.
All the above assume of course we plan to play with the bare minimum needed numbers only. When we move to 10 numbers or so, things get much easier. For now, still the best approach is to utilize a few more numbers and play them in a wheel to stay within budget.
Finally GAT does not get smarter, what you observe is just the evolution of randomness along the GAT tables - there can be situations that randomness does not evolve as well and the opposite where lotto draws have much reduced randomness (and more hits emerge). On average, whenever you run GAT at any given history, this is the average amount of hits you can expect. The only way to increase hits performance is by letting more GATs analyzed which translates as better "randomness understanding".

cheers
lottoarchitect

[pusha]

@Anastasios,

the same idea came to me too lately. But e.g. 6 of 49 Lotto there are about 13.x million combinations. Theoretically we have to let GAT run 13.x million GATs to achive the "understanding" of that single number group GAT have choosen or am i misleading myself here ?


Cheers,

pusha.

[lottoarchitect]

Well, not really pusha, the case is there is really a maximum limit in hits accuracy that we can achieve. I don't know if the current GAT results is the best that can be achieved but trying to evolve randomness has an inherent issue; we still deal with randomness! So really the more GATs analyzed we just approach that upper limit of randomness understanding. Still remains open the scenario of computing billions of GATs but I don't expect to find GATs that constantly produce many wins at every draw tested even if that range is reached - a hitting and non-hitting cycle has to be there because this is what randomness do (change over time).
The only reason we may get even better hits (besides letting more GATs analyzed) is to have a lottery with much reduced randomness. The opposite is a lottery that is truly random. In that case GAT will barely offer any good improvement over natural chance. So what you see in GAT is sort of "proof" that lotto draws are not truly random but they are still random enough so not to be ever possible to hit constantly the winning combination. Actually GAT is designed around this principle and tries to take advantage of but still there is a limit on what can be extracted by the inherent randomness of lotto draws.
GATs are not related to total combinations available. If that was the case, then dealing with a 1/20 field would need just 20 GATs to find the best possible. As you can see, this is not the case!

cheers
lottoarchitect

[pusha]

@Anastasios,

thanks for the info. To my understanding we try to find a GAT which has good hit cycles, kind of a parameter to me. How about the idea to add some likely parameters which act "intelligent" or more adapting. Some of the following are my own ideas, some i've gathered by other sources:

1. Sum Range of the draws.
2. Last digit properties.
3. Picking interval.
4. Frequency of Numbers/Powerballs.
...

We could approach those like the Hot/Cold Cycle of a Gat where a Hill of Hits starst or move down. In my mind there's a kind of signature that returns or occurrence in time. Maybe those "parameters" might help GAT to get decent numbers.


Cheers,

pusha.

[lottoarchitect]

Hi pusha,

I don't know yet what more can be used as evaluation mechanism. The current logic tries to rate each number via the attractors/repellers so really the engine picks actual numbers only and displays them in the order of preference decided by each GAT. In that sense, anything that has to do with sets of numbers will not work (i.e. sums) or partial information (i.e. last digit) for the actual evaluation.