How Random is Randomness

[TL Staff 10]

What are your thoughts on this study? Seems like he is overthinking something rather simple, if you ask me!

 

I was just doing a small study on randomness and came across some interesting results and would like to get some reflection from others.

 

Generated 1048576 coin flips ( "ZERO" or "ONE") randomly using ROUND(RAND(),0)

 

I did this 11 different times. In other words, the above formula was dropped in 11 different columns and generated 1048576 coin flips for each column.

 

Now let's assume that each column is an individual stand-alone trading strategy. I wanted to figure out what the worst/best case scenario could be for a strategy that relied on random buys/sells. Worst/best case scenario that I'm defining in this small study is how many times a win/loss occurred Consecutively. In other words, how many times did the coin flip equal heads consecutively.

 

"Zero" (randomly generated by excel) could either be a win or a loss

"One" (randomly generated by excel) could either be a win or a loss

 

The consecutive win/loss for the 11 stand-alone strategies:

 

Strat1 16

Strat2 20

Strat3 17

Strat4 17

Strat5 18

Strat6 17

Strat7 21

Strat8 18

Start9 17

Start10 17

Start11 18

 

Those results do not seem random to me. So can we argue that if I created more strategies and more random coin flips, there could be a strategy that has a million flips that resulted in a loss each time (since for a coin flip at each flip the odds are 50/50).

 

Thoughts?

[Seeker 11]

It's a little unclear what you're asking, but this is basic probability and you can estimate the expected consecutive losses or wins (there is a formula for this).

 

Perhaps you could explain more clearly what you're asking.

[SIUYA 11]

thoughts.....

first - a random number generator in excel is not exactly random.

second - a truely random number as seeker says can be worked using a formula for what you expect

third - if the markets are really random and you accept that then no strategy will beat the markets therefore you should adopt a passive style of asset allocation.

forth - yes - even in a truely random universe there will at some stage be 1 million (1 or 0)( heads or tails) (wins/loss) that would occur - that is the nature of probability - its highly unlikely to occur but it is stil possible.

fifth - just cause i like to be filthy with my fifth idea is that if you find yourself in the universe go and get laid because it be fantastic because it the place that really unbeleiveable things happen.

[BlueHorseshoe 10]

 

there could be a strategy that has a million flips that resulted in a loss each time.

 

 

The answer is 'yes'.

 

And with equal probability there could be a million winning flips in succession.

 

But why does this matter in terms of thinking about trading?

 

I disagree with SIUYA's statement: I think that a market that was perfectly random would be incredibly easy to trade. The probability of long runs of similar behaviour would be incredibly small, and you could bet against the continuation of such events.

 

Unfortunately price clearly isn't completely random, and nor is it completely non-random, hence trading is difficult . . .

 

I am taking the needle off my own stuck record now

 

BlueHorseshoe

[TL Staff 10]

Came across this post while browsing some private discussion threads, and found it to be interesting on two levels.

 

1. The topic itself regarding "how random is randomness"?

 

2. That people are obsessed with statistics. I think some people overcomplicate trading.

[jpennybags 11]

MARKETS ARE NOT RANDOM!!! The market moves up... it moves down... but for the last %$^&* time it's not random. Anyone who can follow a tick chart / volume chart / depth of market / time and sales can see that it's not random... it's buying and selling.... it's not #$$^&&^^&^%&&^%^& random.

[SIUYA 11]

I disagree with SIUYA's statement: I think that a market that was perfectly random would be incredibly easy to trade. The probability of long runs of similar behaviour would be incredibly small, and you could bet against the continuation of such events.

 

 

Good point, and i think this is why the whole debate about randomness is missunderstood.

If everyone thought and knew it was random and the odds easily calculated over the long term.....in which case you are right.....you would only be able to trade against those who are simply gambling as everyone would then have perfect information and know they are gambling, therefore over the long run....blah blah blah.....

 

Basically the pricing mechanism for this should be arbed out if everyone knew and thought it was perfectly random.....so in a perfectly random market would still be incredibly hard to trade as it would simply be luck - that is the whole point of a random market.

Even If the odds cannot be calculated over the long term because the randomness is still unable to be accurately measured, captured and traded (by the very nature of the randonmness being unable to be proedicited)

 

In both cases you cant beat the market - therefore you should be a passive investor and not an active one

 

 

or maybe we could amend the old saying "The Market Can Remain Irrational Longer Than You Can Remain Solvent" by Keynes to --- "the perfectly random will still have some series of long runs more often that you can stay solvent" - which is why even Casinos limit some people and bet sizes.

[BlueHorseshoe 10]

Basically the pricing mechanism for this should be arbed out if everyone knew and thought it was perfectly random.....so in a perfectly random market would still be incredibly hard to trade as it would simply be luck - that is the whole point of a random market.

.

 

That's not something I had considered. In theory, I suppose, arbing out randmoness would lead to . . . non-random behaviour? I should stop visiting threads like this before 9am

 

"the perfectly random will still have some series of long runs more often that you can stay solvent" - which is why even Casinos limit some people and bet sizes

 

For anyone who can be bothered with it, there's a detailed explanation of the maths involved in casino limits and long runs in Ed Thorp's 'Beat the Dealer'.

 

BlueHorseshoe

[SIUYA 11]

For those interested there is some good theory from William Feller - Wikiquote

 

Its all mind fodder and as probability is all about possibilities and probability maybe most debates will never be solved.....

this might be particularly interesting given all the random talk....

 

"It should be noted that this duration is considerably longer than we naively expect. If two players with 500 dollars each toss a coin until one is ruined, the average duration of the game is 250,000 trials. If a gambler has only one dollar and his adversary has 1000, the average duration is 1000 trials.

Chapter XIV, Random Walk And Ruin Problems, p. 349"

 

.......

Blue - yep i often need a coffee before reality kicks in too

Thats what makes a good discussion - often what we think seems reasonable until compared to reality, or even better sometimes we accept reality in one form until we see it in a different way. Maybe some one will one day develop a perfect pricing mechanism.

Its an interesting thing particularly when it comes to gambling, rational thought etc and pricing.

You could guarantee even with a purely random market there would be people who would bet against the odds - and in walks lady luck.