Predicting the lottery

Is it possible?

[TheDave]

Well, basically I'd just like to know whether anyone here believes its possible to predict lottery results? The Australian lotto works on drawing 8 numbers from a barrel of 1-45, and the last 2 drawn are suplimentry which makes them worse or better or something. Anyway, with that general idea of lottery in mind, would you believe it would be possible, given past results, to predict the next results? Or would you rather dream them? :)

[mack]

Using past results to predict to predict future results in my opinion is not possible :(

You have the same chance of getting the draw producing the same numbers for 2 weeks in a row as you do for different numbers each week. Totaly random and no calculations could possibly predict it.

Sorry, but we can all dream :)

[JayC]

Absolutely not. Think of it this way: if you flip a coin three times and it comes up "heads" each time, have the odds changed as to what will come up the fourth time?

Past results are meaningless in a drawing done as you describe. The drawing is started with exactly the same set of balls each time, and each has exactly the same odds of being drawn each time.

Unless, of course, the game's been "fixed." :)

[kevinpate]

No offense intended, but hey, if I could predict the lottery, I wouldn't be here at the computer on a Saturday night. I'd be out night skiing on some private slope
8^)

[gopi]

Other than SportsGambling/Horse-Race/stock market in no other game of chance the past result can influence future results :)

I read in an Wired Magazine article some time before that in Hongkong there are groups of hitech gamblers betting on horse races with the help of complex software running in high performance hi-end servers . The interesting thing is they have fair level of success in predicting the race outcome :)

Its true no one can predict random events at any time but across a large number of random events probabilty predictions become true...

As a very simple example we all know the probability of a head or tail in a coin toss is 0.5 but you cannot predict its a head or tail in a single event... but if you flip it continously for maybe 10,000 times there is a greater chance that the head/tail distribution will follow the calculated probability of 0.5

[gopi]

And to add Lotteries and Keno are the worst forms of Gambling in terms of probability :)

[mack]

Horse racing and the lottery games cant really be compared. If one horse apears to be performing better that another, then there must be a reason for this. younger animal, better bread etc. The lottery is just balls and chance (that didnt sound quite right) all the balls are the same so there is nothing to be classed as a variable.

[TheDave]

Thanks for your replies :) Someone told me that they load the balls into the machine in the same way every time, and the varying factor is the time they spin the barrel. To be honest I think I've watched about 3 lottery draws in my life, so I wouldn't know, but do you think that could influence results?

[Duckula]

Sounds like someone needs a little google for combinatorics...

That's a high school course by itself, "choose 8 from 45"; uses a lot that funky n!/k!(n-k)! rule.

As for the machine loaded the same way each time, I wouldn't count on it. Chaos mechanics enter into action each time the balls collide. Even the impact of the steps of somebody walking in the room may affect results.

[dingman]

Now, if they start using a computer with a PRNG to pick the number, then you can *maybe* predict the number. You'd need to know a lot about the machine in question, though - probably more than you could find out early enough to buy a ticket.

[brotherhood of LAN]

i guess part of the question is, is random really random :P

i stay away from the lottery. whats the chances my numbers turn up in another countries lottery? that would not be a fun game to play ;)

the odds in the UK are 13000000 to one, if you buy that amount of tickets on a rollover theres a chance you could pocket a few million (hopefully no one else wins!) ;)

[TheDave]

Well, now that the draw has happened I can tell you, I didnt win. Out of 8 numbers I got 3, which is actually not too bad. What I find interesting is that our work lottery, which we have had for 6 years, has won minor ammounts (4 numbers which I think is the lowest win) maybe 2-3 times, whereas a friend of mine's work lottery wins quite regularly. The best accuracy I have achieved with prediction is a 30% accuracy (2-3 balls a draw). I really need 4-6 balls a draw tho :)

[Fiver]

depends on how poorly the lottery is programmed ;)

(funny story.. stupid casino)
[keno-info.com...]

"if you know both the seed and the number generator algorithm, then you can perfectly predict the coming numbers"

so ya, what dingman said :)

[JonB]

well, if you know all the combinations in the past you can "relly" on probability whcih says that in long term all things come at equal distribution. For example if you throw coin 5000 times then results should be around 50/50 in terms of how many tiems did head or tail fall. Althoug it can happen that around 2000 you can get 20 heads in a row. that is of course no guaranty that those numbers that were not in the past that they will come now. But becasue when you thow coin then chances are agfain 50/50 no mater how many times did you throw before.

The problem is that "chance/luck" doesnt remember how were previous results :)

[Fiver]

ya, it's the classic casino trick. notice how the roulette table has a digital display above it reading off all the the numbers that have been hit recently... as though the roulette wheel has a memory. The guy's running Vegas must really think we're slow (... *cough*)

though there was a story in the news recently about a group switching casino's roulette balls for magnetic ones. now that's taking prediction to another level.

[JonB]

one advice: NEVER play a poker or ANY game that is entirelly digitalised - that means that it is software based.Also games like when you pull the lever and 3 wheels go roudn and then you have to hit 3 of the same symbols (i cant rememebr the name) - because they are NOT fair. i saw once when the guy was reparing one of software poker machines that you can set "fairnes" in %. So totally fair game (like if it would be played LIVE with real cards) is set to 100%. But I saw machines set down to 60%! That means that even if by standard probability or luck you would get let say royal flush - if the game has 60% "fairnes" then computer would CHANGE one of cards so you would NOT get royal fluch. Also with "pull lever" machine they can easily do this so one wheel will "with mechanicall help " go "little further" - causing you not to win.i hope that you undertand what i wantedto say.:)

[bcc1234]

Using past results to predict to predict future results in my opinion is not possible :(

In an ideal system - true. In real life - not quite.

Absolutely not. Think of it this way: if you flip a coin three times and it comes up "heads" each time, have the odds changed as to what will come up the fourth time?

If a ticket to play such lottery cost $1, but the payout is $2 - you could "play" 100M times and check the results. Chances are, in the real life, one side of the coin is heavier and thus, the outcome won't be 50%/50% for heads/tails.
And then, after 100M trials, if you get something like 49.99%/50.01% - that's a solid business and not a lottery any more. Just keep betting on the %50.01 side and keep collecting the difference :)

Read the line above, about the ideal systems :)

No offense intended, but hey, if I could predict the lottery, I wouldn't be here at the computer on a Saturday night.

There are government agencies that constantly check statistics and look for non-linear relationships in the outcome of all major lotteries. As soon and they find anything even remotely resembling some kind of a pattern - the lottery gets audited and the mechanisms get changed.

Its true no one can predict random events at any time but across a large number of random events probabilty predictions become true...

There is no solid mathematical proof that "a random event" actually exists. So far, all that humans could reproduce, was pseudo-random to some degree. The degree (basically, the highest power of a polynomial in a relationship) has always been the computational practical limit.

but if you flip it continously for maybe 10,000 times there is a greater chance that the head/tail distribution will follow the calculated probability of 0.5

My point exactly. It's just a matter of how complex you can go with your calculations. What if you take into account kinetic energies of all particles in the universe while calculating the odds of the coin toss outcome? Again, it's just the level of complexity that we can measure.

Sounds like someone needs a little google for combinatorics...

Again, this is a real world, not a mathematical equation. That's what makes the difference!

[Tapolyai]

I think we can predict the lotto.

(I would prefer to do it say this week with the U.S. Megalotto at $100+ million, would be nice, eh?)

If you study chaos theory, or as previously mentioned randomness, you will find that there is no such thing as random.

The problem is not "can we predict", but "what goes into the prediction".

The wings of a butterfly in Belize would have an impact on the lottery in Australia.

[TheDave]

Some good contradictions to the first posts, and I have some evidence now to back them up :). I ran a test last night, using past results and prediction methods. Picking 8 numbers and comparing them to the actual results, the average accuracy of my prediction method over a set number of draws (50) was 25% (Accuracy being counted in balls per draw, so about 2-3 balls a draw, sometimes none, sometimes 5, rarely 6), while when I had my computer generate random results and compare them to the last 50 results, the average accuracy was about 2% (ie, rarely got 1 ball even).

[weblamer2]

Once, out of boredom, i loaded up a couple months worth of florida winning lotto numbers into an access database.

you can get them here:
[flalottery.com...]

I then wrote a VB front end and tried all kinds of things with the numbers, avaraging, stats, and so on.

Only thing i saw the numbers do is make an almost perfect bell curve.

Then, for fun, i made it creat a VRML page. Each winning combination was a cube, with the first 3 numbers representing the X-Y-X coordnate, and the next 3 numbers the RGB color of the cube.

I came out with a spiffy borg-looking piece of art but, alas, nothign to help me win the lotto.

[JayC]

Some good contradictions to the first posts,

Not really contradictions; they're just not answering the same question -- which was:

would you believe it would be possible, given past results, to predict the next results?

"The next results" seemingly meaning the results of the next lottery drawing.

So, the points made that you could make 100,000 trials and find that what appears to be 50/50 is actually "49.99%/50.01%" don't address the question. That there might be a small margin which you can use over a long period of time to beat some games doesn't mean that you can predict the results of the next lottery drawing, or the next coin toss -- no matter how much you know about past results.

For example if you throw coin 5000 times then results should be around 50/50 in terms of how many tiems did head or tail fall.

Again, true enough. But not really useful in predicting the results of the next coin toss. On the other hand, I'd be willing to bet a cool million that if I toss a coin 100 times, it'll come up "heads" at least 40. :)

[bcc1234]

Not really contradictions; they're just not answering the same question -- which was:
would you believe it would be possible, given past results, to predict the next results?

The question misses a lot fo details.
Just by knowing the previous numbers - no.
Have the machines been changed/upgraded during the previous trials? - This is just one out of infinitely large number of factors that should be taken into account.

For the question: "Is it theoretically possible to predict the future state of a given system and it's history in a practical world?" - The answer would be yes.

[bcc1234]

I then wrote a VB front end and tried all kinds of things with the numbers, avaraging, stats, and so on.

If you like messing with analytical statistics - check out evolutionary algorithms (in particular genetic algorithm theory) in conjunction with artificial neural networks.

You'll love the stuff.

[AgentSmith]

In 6 slot lotto (with numbers between 1 an 50), with 6th number being independent (meaning drawn from a seperate pool)

50*49*48*47*46*50=12,712,560,000

That's with a very basic finite math. Throw in some variables to the equation, the computation could get pretty ugly:)