romansh wrote:I am reminded of discussions with my father in law. His system was to watch a roulette wheel for a long time and then bet on the number that has not come up and then bet on that number. Keep telling him telling him … if he is going to bet on that sort of foolishness, he may as well bet against that number. The wheel might not be true.

Warning: anecdote alert

When I was a kid in the 70s, when I'd learned how to calculate mean and SD but knew nothing else about statistics and probability, I won £6.23 on the Littlewoods Football Pools using a system similar to this.

My parents used to play with a small stake (2 lines of 8 from 10, I think, which cost about 50p) every week, and I got interested in the form book. Having no interest in football, I calculated the mean number of times and its standard deviation that each number came up as a score draw, and found that the means were all nearly the same, and the SD was very small.

I reckoned that if all the numbers came up nearly the same number of times, if I just picked numbers that hadn't come up, I'd get more and more likely to win. So I persuaded my dad to let me pick the numbers for a season.

Amazingly, I plotted the graph of my score each week, and it did increase fairly steadily, until we won a small payout, which was only a little bit less than the total outlay that season. Given that my dad had never won anything on the pools, he was entertained enough by my success to let me keep the £6.23.

I never did find out what it was about the way the matches were allocated to numbers each week, but it did seem strangely non-random at the time. Once there was a mix of numbers that had come up 0, 1, or 2 times, I found it harder to pick them, and I got bored and stopped, so never became a big winner.

Sorry, I was briefly possessed by the Nostalgia Demon there. As you were.

My father in law has stats going back for the pools, nags and dogs. OK in he is in nineties now, so he has slowed down now; but he has been the same for the forty years I have been married to his daughter. My father in law never became a big winner either.

romansh wrote:My father in law has stats going back for the pools, nags and dogs.

Of what is that diagnostic, except OCD?

Cito di Pense wrote:

romansh wrote:My father in law has stats going back for the pools, nags and dogs.

Of what is that diagnostic, except OCD?

I avoid that speculation as it might run in the family.

romansh wrote:

Cito di Pense wrote:

romansh wrote:My father in law has stats going back for the pools, nags and dogs.

Of what is that diagnostic, except OCD?

I avoid that speculation as it might run in the family.

I think if you are averse to speculation, then you test negative for the affliction.

The football pools. Saturday night 6'oclock? (this was 1955) a pint of whelks watching the results around a 14" tele in Putney, London.

We never won anything. My grandfather would put a fiver on the coupon. A lot in those days.

In theory it works. Just always bet enough that if you win, you get all of your money back plus a small profit. But that's why casinos have maximum bets. And as soon as you hit the maximum, the strategy no longer works. It's basically like asking your friend to go double or quits until you win. How many times are they going to agree to it before they see what you're doing and tell you to take a hike?

I don't know the maths of it, but I'd imagine it's not a strategy that leads to a huge profit either, so if you had the sort of money that would allow you to do it, there are probably better ways of making it pay out.

I'm With Stupid wrote:In theory it works.

I'm not quite sure what you mean by that. If you place bets at 50/50 odds (which you can't in a casino) then your expected return playing a Martingale over any finite number of bets is 0, your expected return if you have a finite bank roll and play until you win some fixed amount or go bust is 0 and similarly if the house has a limit to bet size your expected return is 0.

If the house has an edge (like all real casino games do) then your expectation is losses.

The important mathematical result relating to Martingales is The Optional Stopping Theorem.