Lottery Guy

The Lottery Number Frequency Myth

November 4th, 2009   ·   Read 14 Comments...

lottery frequency[Q] Should I play lottery numbers that have a higher frequency of being drawn?

[A] There are lots of interesting things you can do with old lottery results. You can apply all sorts of analysis on past results, and come up with lots of intriguing things.

Even the lottery companies do it themselves. Many official lottery sites have things like lottery frequency checker tools on them, so you can see how often each ball has been drawn.

But the problem with this kind of analysis, is that it’s entirely useless in predicting what’s going to happen next.

Here’s why…

Lottery Randomness Looks Weird

I know this sounds a bit Douglas Adams but the big problem with randomness is that it can often look so very unrandom.

For something to be truly random, all results have to be possible.

So for there to be any significance in some numbers appearing more often than others, there must by definition be something wrong with the draw process. In others words it isn’t random.

Yes, that’s theoretically possible – but bear in mind lottery companies put an awful lot of effort into ensuring their draws are random.

By all means fire up your Excel spreadsheet, plug in all the past results, and make charts of hit frequency. If you find that kind of stuff fun.

(No offence intended – I’m a maths and probability geek!)

But please don’t be fooled into thinking this kind of analysis will increase your chances of winning the lottery.

It won’t do your chances of winning any harm to play this way, it just won’t improve them any either.

For more help with winning and real stuff you can do to improve your odds, get my free lottery tips course.

Categories: Lottery Questions · Lottery Tips · Picking Lottery Numbers

14 Comments so far ↓

  • wilf

    Hi
    Has the same set of six numbers ever been drawn again in the LOTTO draws

  • Lottery-Guy.com

    Hi Wilf.
    Yes it has. See my blog post here about repeat lottery numbers.

  • Nick

    Hey Lottery-Guy,

    Interesting info.

  • Shockadelic

    If lottery draws *appear* to not be random, maybe they aren’t (even if there’s not obvious explanation why).

    If there is a higher frequency or increasing frequency for certain numbers, why wouldn’t playing them be an effective strategy?
    Even if you can’t explain *why* they’re more frequent, you can’t ignore the fact that they are.
    Surely those numbers are a better choice than ones that hardly ever happen (again without knowing why)?

    I think people want to *believe* lotteries are “random”, as it let’s their lazy arse off the hook, so they don’t have to make any effort figuring out how to predict it.

    • Lottery-Guy.com

      Ah, but what frequency of numbers being drawn determines it’s NOT random? 5 times more than the average, 10, 50?

      If I roll a dice and get a 6 twice in a row, you’d probably agree that’s just fine. But what about if I roll it ten times and get a 6 every time? It’s starting to look funny, right? But is that impossible in a random system, or just unlikely?

      So can we determine unlikely versus impossible..?

      Well clearly it would be impossible to roll a 7 :-). But there are lots of highly improbable results that are perfectly possible using a proper random dice roll.

      And the same applies to lottery balls. There will always be some that have appeared more than others overall. It’s possible for ball number 12 to have appeared 50 times more than all of the others. It may appear again next week, or it may not appear for another 6 months. And all of those things are possible in a normal random draw.

      We can use statistical analysis to determine if results are starting to look non-random, with certain ‘levels of confidence’. I’ve done this on a couple of big games over various periods of results including full histories – and found nothing statistically significant.

      It’s very likely that any lottery company does this kind of analysis too as part of their internal audits, so if anything unusual was even hinted at you can be pretty sure they would be the first to know. And even though it’s probably still fine, they’d be off changing balls sets and machines, and running 100′s of test draws to verify.

      So here’s how I look at it. Lotteries are random enough as to be unpredictable simply because that’s how they are designed. So unless someone can come up with an explanation as to why one isn’t, and then go on to prove it, then to me that continues to be the only logical conclusion. As yet nobody has proven otherwise :-)

  • Steve

    Hi I’m from South Africa and believe lotto/powerball can only be won when one is lucky. I have tried all the software and have not win lots of money. One thing I have learned is that lotto/powerball is not random.

  • Allan

    Random or unrandom… they are all correct! There are also many variations involved. So this or these has let me come into a conclusion, the lottery companies are monitoring the winnings and secretly changing their FORMULAs aye!

    • LG

      Conspiracy theories aside, why would they bother? Why would a lottery company risk their guaranteed substantial profits on doing anything other than operating a perfectly fair and random draw? It doesn’t really make any sense.

      • Shockadelic

        Maybe they, like you, *think* they’re operating a “random” system, but are not for spooky/cosmic reasons unknown.

        I have seen so much evidence that shows a bias for or against numbers, but it’s short term.

        You wouldn’t predict the weather from last century’s records, neither is long term lotto stats relevant.
        What matters is what’s happening (or not) lately, whether it makes “rational” sense or not.

        If #29 happens 7 times in the last 10 games (including 4 in a row) and #36 hasn’t happened for 43 games, there’s obviously a bias there, even if you can’t *explain* why.
        Don’t deny it. Play the bias.

        (But remember this isn’t horse racing, #36 *will* inevitably happen again. The longer it’s been the *more* improbable for it not to happen soon)

      • LG

        Bias is of course possible.

        But it’s also something lottery companies are very aware of, and do their best to avoid. They don’t make a public announcement every time they replace a component in the draw machine, or even the whole machine. They don’t tell us when they replace a ball, or whole sets of balls. (Or for that matter – how many ball sets are in use, and how many draw machines..?)

        What they also don’t tell us is the ambient temperature in the room, or if a heavy truck drove past during this weeks draw. Or a multitude of other factors, any of which could easily outweigh any detectable mechanical bias in the system itself.

        So whilst it may not be a 100% perfect random system, it’s more than random enough. And despite the volumes of players analysing and predicting, I’ve yet to see any convincing evidence to the contrary.

        RE: ‘Ball #36′
        Surely if bias is the reason #36 has not appeared then why would it become ‘more improbable’ in future for it not to be drawn..? And if it’s not bias, then in a random system results don’t become more likely the longer they haven’t happened. They remain exactly as likely every time, right?

        It’s easy to be drawn into thinking that just because #36 hasn’t appeared for 43 games that it must be down to bias. But in a truly random system is it impossible for #36 not to have appeared for that long? And is it likely the bias you think you’re seeing has remained unchanged? (What if the lottery company thought the same thing 10 draws ago and changed the ball..?)

        If you enjoy analysing past result and making predictions then that’s just fine with me – it’s bonus entertainment value. But if you don’t enjoy it, I don’t see that you’re losing anything by not doing it.

      • Shockadelic

        Yes, the operators may change elements, which is precisely why you should only use short-term data.

        Of course there are many factors that could change the outcome. You don’t know what they are, all you can do is monitor the results.

        #36 would become “more probable” with a lengthy absence, simply because of mathematical probability.
        If 6 numbers are drawn from 45, each number should appear once in 7.5 games.
        A gap of 43 games is already way over the statistical average.
        #36 *must* happen eventually and for it to continue this gap is simply astronomically improbable (no, not impossible, improbable).

        That’s what prediction is all about, probability.
        And seeing the *actual* results (not theoretical assumptions of randomness) clearly shows some numbers are more or less probable (but none are impossible)

      • LG

        You’re misunderstanding probability.

        A ball does NOT become more likely the longer it hasn’t appeared. That’s a common fallacy.

        Just like when you roll a dice. Every time you roll it each number still has a 1-in-6 chance of appearing. It makes no difference what the previous results were – there are still 6 sides to the dice.

      • Shockadelic

        I think it is you who misunderstands probablility in this instance.

        There is the probability a number will not occur in *a* game. (1: 7.5)
        But there isn’t just *a* game, there is a series of games.

        So there’s also the probability it will not occur in 43 consecutive games. (1:322.5)
        That’s exponentially different.

        For a number to not occur in *a* game, no big deal.
        For it not to occur in 43 consecutive games is ridiculously improbable.
        Therefore, it is also ridiculously improbable that gap will continue for much longer.

      • LG

        This is a common mistake.

        You’re calculating the probability as if these are dependent events. But they’re not. Lottery draws are independent events, just like a dice roll or a coin toss. So each game the probability of any number being drawn is exactly the same as any other (bias aside).

        (To think of it another way. They run test draws between the real draws, for which they don’t release results. If these were dependent events, you would not be able to calculate the probability as you don’t have the results for all the draws you didn’t see…)

Please Leave a Comment Here