Predicting Lotto Numbers??

This article only deals with one specific method of pretended lotto prognoses. » Click here for all my articles about lotto

Via Bloggerei.de and (update:) what was, at that time, on Ramschmarkt.de, I came across something called “Global Scaling”, a, let’s call it view of the world where everything is said to be connected by a standing wave and that apparently claims that all things big and small can somehow be described with logarithmic scales – or something like that, maybe I didn’t interpret all those pieces of information floating around on the web correctly. The main homepage seems to be globalscaling.de (also abbreviated as GS in the following text), the “Institute for Space Energy Research”.

• When physicists and other scientists research in “free space energy” or things like that, I don’t mind at first, things have to be investigated;
• When a “cosmic background noise” and logarithmic scales are kinda seen as new “silver bullet” for anything (logarithms not being uncommon in math and physics, anyway), some skepticism is necessary, I think;
• When “lotto prognosis” is mentioned as one of the fields of application, the whole thing, in my consideration, crashes down to the lowest end on any respectability scale;
• When they charge money for that, my spontaneous opinion sees that going maybe a little too far towards rip-off or deceit.

Yes, I’ve contemplated whether I should spend much attention to such a dubious “system”, but…

1. …well, let’s have a closer look.

(I apologize for this being quite lengthy, but a certain detailedness is necessary. So maybe get yourself a cup of coffee – or go directly to the conclusion…))

Quote from globalscaling.de, my translation:

“How exact is the Lotto prognosis?

The lottery is about the selection of a random sequence of natural (whole) numbers. Because of that, under best preconditions, a lotto prognosis can never be more accurate than ±1.”

This combination of these two sentences, I think, will make the hair of anyone who has ever seen a formula from probability calculus from nearby stand on end. Only with a little heaps of good will – and without knowing the formulae of “Global Scaling” – one could assume that this sentence’s wording is just accidentally a little off, or shortened, simplified too much…

The sentence (in my translation)

“We point out that we give no guarantee for the precision of our lotto prognosis.”

is necessary, of course – if the prognoses were perfect, they already were multiple lotto millionaires and had no need to make the effort to sell their “lotto prognosis”. Anyway, following on that page are a bunch of big numbers, which, the way I read it, boil down to the fact that the more tips you play that do not repeat, the higher your chances to win. Preferably (for GS) with numbers bought from GS, of course…

2. But how good are their prognoses?

For this purpose, they publish (accessible in the customer area without login) a list of 145 prognoses (consisting of 7 numbers each) and the drawn numbers from German Lotto (“6/49” plus “bonus number”) from 29/4/2006 to 26/9/2007 (when I started writing this article; they skipped 3 and have 1 twice, accidentally), plus as analysis how many drawn numbers differed by 1 or less and by 2 or less from the nearest predicted number. I analyzed that with a little more detail and calculated the mean values of the 145 prognoses (the bold numbers are equivalent to those they mention themselves):

exact hits:	0.924 ■ (of 7)
difference ±1 / sum:	1.986 ■ / 2.910
difference ±2 / sum:	1.752 ■ / 4.662
difference ±3 or more:	2.338 ■

3. Doesn’t sound bad – or does it?

Well… By the way, can anyone tell me the practical use of differences of ±1 or more – other than maybe trying to show how great the prognosis seemingly works? Because:

Would anyone really play 54,264 tips (all possible combinations of 6 out of the 21 numbers that you get when you take the prognosed 7 and each number before and after these)? If you had, you had actually scored “6 correct numbers” once, winning 571,084.50 €, on 14/7/2007 (or with a chance of 1:10 for the correct “super number” the half of 4,283,134.40 €), and four more times “5 + bonus”. Nice? Well, each tip costs 0.75 €, that’s 40,698 € plus handling fees per drawing, or 5,901,210 € + fees for 145 drawings… Investing 347,130 € (average for tips per month, without fees) every month for 17 months at 3% interest p.a., you’d have about 6,034,000 € now.

Had you played 7 tips each (all combinations of 6 numbers out of the prognosed 7) with these 145 prognoses, it had cost you 7 x 0.75 € x 145 = 761.25 € plus fees (72.50 € in Bavaria) – plus 468 € for GS’s numbers. Total: 1301.75 €. You had scored only “3 + bonus” on two (12/7/2006, 25/10/2006) and “3” on three more drawings, which adds up to 6x “3+bonus” + 14x “3” thanks to the “6 out of 7” system, giving you prizes of a meager 295.10 €.

(Without guarantee.)

4. Could the numbers be better?

For the fun of it, I compared each of these 145 prognoses to each of the 145 drawings, a total of 21,025 comparisons, as well as to all 713 drawings since 2/12/2000 (can be downloaded e.g. from Lotto Bayern, by the way), that’s 103,530 comparisons:

each prognosis compared to:	145 drawings	713 drawings
exact hits:	0.973 ■ (of 7)	0.994 ■ (of 7)
difference ±1 / sum:	1.984 ■ / 2.957	1.991 ■ / 2.985
difference ±2 / sum:	1.732 ■ / 4.688	1.720 ■ / 4.705
difference ±3 or more:	2.301 ■	2.295 ■

Whoops! That’s even a little better! globalscaling.de seemed to have had a little bad luck…

By the way, there is apparently no trend that the prognoses were getting better over time (even worse within difference ±2):

Update until 17 Nov 07 see here.

But how good are the numbers really? How do they compare to “bilnd”, i.e. purely random tips “7 out of 49”? Since the 1 and the 49 do not appear in the 145 published prognoses (the goal being best precision within ±1, mind you), the highest first number is 11 (all tip rows mentioned here are looked at in ascending order, by the way) and the lowest last 38 (which is often far more extreme for purely random choices), we will regard that too:

	random 1 to 49	random 2 to 48, borders ≤11, ≥38
exact hits:	1.002 ■ (of 7)	0.963 ■ (of 7)
difference ±1 / sum:	1.638 ■ / 2.640	1.609 ■ / 2.572
difference ±2 / sum:	1.239 ■ / 3.879	1.271 ■ / 3.843
difference ±3 or more:	3.121 ■	3.157 ■

I conducted 1,480,000 “prognosis comparisons” each (fluctuations are still possible, of course – and they are computer-generated pseudo random numbers, after all). (It’s of course excpected to see an average exact hit of 1 of 7 when drawing 7 of 49 numbers.)

We see: The GS numbers are, in average, actually closer to the drawn numbers than pure random numbers – whatever that may be useful for. That’s not really surprising when considering that pure random numbers have often quite large gaps in which many drawn numbers can “get lost”.

5. So what’s so special about the GS numbers?

One might suspect that it’s important that the numbers differ by 3 or more, becuase with numbers following each other directly or with a single number gap, you would give away potential for being as good as possible with differences of max. ±1. So let’s analyze the frequency of distances between the numbers, calculating with difference from -1 for the first number for practical reasons (difference ±1). (Green line: GS numbers; orange/red-brown: random)

Aha! It is concentrated on distances between 3 and 10, with the distance of 1 (i.e. one number directly succeeding the other) appearing only once in all 145 prognoses, just like the highest of 20. With the random numbers (1-49), the highest distance was 42, the highest first number 41, the lowest last 9!

6. So let’s try simulating this frequency.

I have used this array of numbers, out of which the distance for the next number in a tip row is chosen randomly (and tip rows whose highest number was below 38 were discarded):
(3,3,3,3,3,3,
4,4,4,4,4,4,4,4,4,4,
5,5,5,5,5,5,5,5,5,5,5,
6,6,6,6,6,6,6,6,6,6,
7,7,7,7,7,7,7,7,7,
8,8,8,8,8,8,
9,9,9,9,9,
10,10,10,
11)

The frequency after 1,480,000 experiments looks as follows (red line; green line the GS numbers repeated from above) – somewhat similar (close enough, since I don’t intend to completely imitate GS anyway):

Nota bene: I do not claim that globalscaling.de uses a simple function like this! I don’t know the algorithm GS uses. I can only say that the result of this little function of mine is at least as “good” (according to the criteria mentioned or implied by GS) – in my experiments even better:

	GS numbers (from above)	my little function
exact hits:	0.924 ■ (of 7)	1.148 ■ (of 7)
difference ±1 / sum:	1.986 ■ / 2.910	2.264 ■ / 3.411
difference ±2 / sum:	1.752 ■ / 4.662	1.866 ■ / 5.277
difference ±3 or more:	2.338 ■	1.723 ■

7. So on to the practical application!

 

If you want, you may use these generated numbers – free of charge – as basis for your lotto tip. Of course I cannot guarantee that you win, just as, for instance, (unmanipulated) dice can’t. Should you actually win something, it would be nice if you donate half of it to me (but that’s no obligation, it’s all up to you). For one or a few tips, these numbers are neither better nor worse than any other numbers.

These numbers and the underlying program code may not be used commercially, i.e. especially selling is forbidden!

If you want to look at the code: cleverrandom.js (right click, “Save target as…” or similar)

So we see: Rows of numbers that meet the conditions of “prognoses with small differences” (as they are also deducible from the prognoses published by GS) can also be created with little effort.

We might have just used 7 fixed numbers with the same distances, but with only 145 or 713 experiments (drawings), that would have been anything but representative.

8. And what’s the use?

Well, if you really want to play 54,264 tips (“6 out of 21”, see above, you may now have come to realize on a slightly intricate way that you should make sure that you don’t play the same tip row twice, since double tip rows would not improve your chances of winning anything at all.

Whether you spend extra money on that or just take any numbers that differ by at least 3, is up to you – your chances of winning are effectively the same, as can be seen from above analysis as wel as all accepted, verified mathematical methods…

That also applies when playing in a “sensible” manner – a few tips per drawing. You do not need to buy numbers or use a number generation function like mine above.

We had already found that the prizes won had been less than the costs when playing according to the published GS numbers in section 3 above.

9. Conclusion

• If you simply feel like playing many, maybe thousands of tips for the same drawing, just make sure that you don’t play the same tip row twice – would you really need to buy numbers for that purpose??
• I do not know how globalscaling.de creates their prognoses, I can only say as result of this article: The published 145 prognoses from globalscaling.de are not “better” (even with their criteria) than a simple function with random numbers that makes sure there are no too small and no too large differences between the numbers.
  (And as far as I can assess with my knowledge as a “scientifically interested layman”, they will never be good enough to be useful, because I consider the “scientific” basis of that to be wrong. If, however, you happen to have or know about a respectable and verified scientific proof – ideally published in a renowned journal – please let me know! It would be a huge surprise…)
• The expenses for tips with these 145 prognoses would have exceeded the earnings clearly.
• And as a reminder: The lotto balls have no memory, the chances for 6 correct numbers are the same for each drawing, 1 : 13,983,816 per tip (and for the main prize, the “super number”, the last number of the playing form, must match, too). Beyond statistics, a specific lotto number prognosis is impossible according to all respectable, accepted and verified mathematical knowledge, anyway.

Update 19 Oct.: The lotto prognosis was only temporarily not linked to on their start page, and there’s new info on the trustworthiness of the whole “Global Scaling” construct: more….