Has anyone ever tried using the 'law of averages' to turn a profit?

a.k.a martingale betting

Hello there CrowsEye.

To actually answer your question, i vaguely recall reading on the forum that the Magician did something along these lines.

It was an EPL season and there had been an unusually low number of draws to that point in the season and he decided to back the draw on games for the remainder of the season. I think he did it for fun rather than a serious strategy but he made a profit.

That's about as much as i know, maybe he or somebody who knows a bit more could clarify or confirm if I have got this all completely wrong.

Still martingale though - but, I suppose, getting 5/2-ish on the draw (as oppose to the evens shot on roulette) makes it more attractive. There was a thread (but can't remember the poster's name!) about following/backing the big four in the Premier League (couple of years back - Man U, Arsenal,L'pool, Chelsea) to draw. Did well as I remember. Again a bit martingale-ish but the odds for the draw involving those four (especially at home) were high. And surprising the amount og times they drew.

Level headed gambling doesn't make for a fun thread, especially in General Betting!

IVE USED THE LAW OF AVERAGES IN SYSTEMS AND THE LAW OF AVERAGES BREAKS THE LAWS WHEN I USE IT AND I LOOSE

The basic trouble with using the law of averages is that things don't revert to the mean in a straight line, nor necessarily in a short period of time.
Resultantly you get killed by the buy/sell spread and commission.
Also what historical set of results do you base your future betting on ?
You can always find a set of lead up data to support variances either side of the norm.
Which do you chose and why ?

Good point as usual Trevh.
Btw I didn't respond to you on another thread on the climate change thing, as I know better than to go there with you.

Bit of a bummer Trevh having that sort of disease on here methinks.

Magician, if i recall correctly, backed to level stakes on the draws.

Casinos may as well close their doors now.

The op's opening question relates to what statisticians call "the gambler's fallacy". This is the idea that after a sequence of 10 heads on a coin toss a tails result becomes more likely because overall we expect to see an equal number of heads and tails. They have good reason to call it the Gambler's fallacy; gamblers love it and it's false. It's false because each coin toss is independent, previous tosses have no effect on future ones, so the length of a sequence has no impact on the probability of an individual result. So far so obvious.

Further down the thread someone else alludes to the idea of "mean reversion" although I think what he's really talking about is "regression to the mean," which is slightly different. Regression to the mean is the idea that if a variable has an extreme value then the next value in the sequence is likely to be less extreme. For example if a footballer scores 10 goals in one game, he's likely to score less than 10 in the next. Again, pretty obvious.

Unfortunately however confusion does occur between these two ideas because at first glance the phenomena of regression to the mean might be seen to confirm the Gambler's fallacy (which would make it the Gambler's truth). Take the coin toss example: A let's say that you toss a coin 100 times and come up with 95 heads, 5 tails, that's a fairly extreme result, so some would say that regression to the mean suggests that a less extreme result is likely to follow, and since overall we expect to average 50/50 heads and tails, it's more probable that next result will be tails- right? Wrong. You might be correct in saying that 95/100 heads is unusual, but regression to the mean doesn't suggest that the next individual result will be a tail, it just suggests that the next 100 results are likely to show a spread of heads and tails that's closer to the average (50/50) than the extreme result (95/5), which is much more boring, but true.

Mean reversion is a term used in finance to describe how returns are likely to be stable over a long period even though they fluctuate in the short term. It's a more dodgy and less well formed concept than regression to the mean (aka mean regression) which is a statistics term. The two are often interchanged incorrectly.

I agree. The former is sensible, the latter is not.

Actually roulette is quite a good example because it's simple. It's always worth reducing a problem to it's simplest formation. With something more complex (football for example) there's always the temptation to come up with reasons why what you want to be true must be true, but when we reduce the outcome to one of true chance we can see the mechanism very clearly. In this case it's plain to see that the Gambler's Fallacy applies as readily to Football as it does to roulette.

Football isn't random in the true sense, but because on the whole the prices are a true indicator of the probabilities of the outcomes deviations in actual outcomes relative to expected outcomes are effectively random (in that they cannot be predicted) and a truly random model (like roulette) is a good way of understanding them.

'Mean regression' it is then - cheers Mr Robot

stoopid site - I only clicked the reply button once.

I WAS ALWAYS BELOW AVERAGE AT SCHOOL ,AND NEVER WENT INTO THE ABOVE AVERAGE SECTOR

WHEN I WAS A CHILD IF I STARTED TOSSING COINS UP IN THE AIR ,SOMEONE WOULD KNOCK ME OVER AND PINCH THEM

ALOT OF YOU DONT REALISE SOME CHILDREN ON COUNCIL ESTATES NEVER TOUCH REAL MONEY AT SCHOOL AS THEY ONLY HAVE VOUCHERS FOR FREE BUS FARES AND DINNERS

I WASNT EVEN ALLOWED TO PLAY WITH THE CHILDRENS PLASTIC MONEY OUT OF THEIR TOY POST OFFICE ,AS SHE TOLD THEM TO WATCH ALL THEIR MONEY WHEN I WAS ABOUT