Profiting from randomness: A money management system?

[beamstas]

Might be a bit off topic, but I am definitely interested in learning how a coin flips outcome is dependent on the outcome of the last coin flip outcome. Every little bit of information I have ever collected tells me that coin flips outcomes are independent of each other.

Its is never dependent on the previous flip, i'd be suprised if he tried to argue that.

I think he is trying to say that the flip of a coin is non random due to thing such as gravity, force being used to flip the coin, the height you toss the coin, the type of coin used etc.

It may be possible to compute the outcome given none of those factors are variables, but since all inputs will be variable for every single toss it is as close to being pure randomness as possible.

 

[motorway]

I'm yet to see proof that entries give you an edge.

With a simple type of trend filter for the market and a simple trailing stop, you'd be suprised at how well random entries on random stocks can work.

I do believe there are people out there who can get an edge with their entry, but the vast majority of people will simply backfit an entry from what they have seen on a chart.

yes I think most back testing for entries is just that


But consider

Given one of these entries
and lets say at an identical time
you get a signal on TLS
and another on XYZ promising small cap ( eg bio tech or mineral explorer )
are the outcomes identical
in terms of expectation

is there more to entries than just HOW ?

and what of TALEB
point of disconfirmation

Can you have a signal to tell YOU NOT WHAT TO GET INTO

Is there an edge to be found there ?
and should you go short there ?
and should your long universe be only those stocks that
are not signalling a NO ENTRY
and yes it is all about Trend Filters
But are there better filters

and are there No-trend filters ?
Would one of these be useful ( again disconfirmation )

Thinking in terms of coin tosses
where the bias on the coin is created by the bets people make on the coin
Is the type of thinking that will take you very far..



motorway

 

[overit]

Just to add my 2cents worth I believe the market is fairly random but has a bias built in. There is an obvious upside bias in favourable market conditions and a downside bias when things are crappy. Ie why does the market go up when the economy is going well.

I did an interesting experiment with my roulette numbers collection. Obviously I would like a larger statistical base than 400 numbers but it is enough to give you the idea. I just wanted to see what a random number graph would look like. I used my hoyle casino pc game as a random number generator.

The 3 line graphs are black vs red, odd vs even and high vs low. I associated value "1" vs "-1" to these. With zero given a "0" value. Ie if it is 50/50 black/red the total sum should be 0. (Correct me if I am wrong.) I honestly thought I would not see a trend develop with 50/50 outcome possible.

Definetly food for thought. Do I see support/ resistance points, double tops... LOL!

 

[motorway]

If it is 50/50 black/red the total sum should be 0 ( But ) + or - 20 ( with 400 and average of mutiple runs)
big HINT how to build a trend filter in that

BUT BE CAREFULL

Epileptic FIT

model that with coin toss too
But the same "Pattern"
while not predicting the next coin toss
will predict the next Fit

Just because a random sequence can produce a pattern
does not mean in another context that the pattern has no meaning too

Also read Mandelbrot on How
stock fluctuations are different to Coin Fluctuations
he covers that difference well

same with earthquakes and coin tosses

and stockmarket and coin tosses
The dynamics of a top of a Bull market are real and exist
there is no top and no dynamics ( regime changes )
on your charts
JUST random walks

motorway

 

[brty]

I think he is trying to say that the flip of a coin is non random due to thing such as gravity, force being used to flip the coin, the height you toss the coin, the type of coin used etc.

You think correctly here..

Not so accurate here though..

It may be possible to compute the outcome given none of those factors are variables, but since all inputs will be variable for every single toss it is as close to being pure randomness as possible.

Why do people equate difficulty to predict with random?? How do you know

since all inputs will be variable

Does "variable" mean random?? What if some input was deliberately changed each toss??

I first woke up to how un-random things are by tossing a coin while trying out a new system for commodity trading. By giving the coin a toss with quite few revolutions, probably about a foot high into the air, I scored 20 heads in 25 tosses. Thinking this ridiculous, I started again, this time with tails up on my finger, I had heads up the first time. The result was about 15 tails in a row before I gave up as I was not getting anything close to "random". Only by deliberately changing the force of flip was I able to get results approaching 'random'.

My argument is about how an input that is controlled can be classified as 'random'.

By controlled, I mean by force applied in the coin tossing case, but everything else that is considered random also has 'force' of some type or description applied. (a computer random number generator is simply a function of the algorithm applied)

I have not seen anything that could be described as 'random' and do not believe in the concept.

brty

 

[beamstas]

Why do people equate difficulty to predict with random?? How do you know Does "variable" mean random?? What if some input was deliberately changed each toss??

I have not seen anything that could be described as 'random' and do not believe in the concept.

This is like you see on news.com.au.
Backfit ideas around why a random event occurs, that would have made the event completely foreseeable if you had the information at the time, but is not useful for predicting a future event.

I think you are confusing calculating and backfitting a random event with actual randomness.

Probability is simply accepting the fact that alternate events could occur, not the actual computation of why they did occur.

 

[brty]

beamstas,

When you understand why the different events occur, you have a better handle on the probabilities.

Could you please give an example of just one "random" event and of course why it is considered "random".

brty

 

[ThingyMajiggy]

Whats the big deal about if its random or not?

Whether you can profit from it is the main thing I would have thought, whether it be random or not?

Charts are just visual representations of the effort and result of the traders/investors involved. Traders aren't random.

 

[MichaelD]

Charts are random. ---Have no reliable reoccurring information for trading

Yes they do.

Trends.

My view is that any given chart looked at in isolation provides nil information as to what will happen next.

i.e. Will tomorrow be up/down/sideways? Not knowable from a chart.
i.e. Has a trend started? Are we mid-trend? Has a trend finished? Not knowable from a chart except in hindsight.

If sufficient people look at the chart and believe it means something, therein lies the genesis of a potential non-random event (= potentially profitable event with appropriate trade management), but these are mere ripples in the sea of randomness.

Trends arise in random series and are exploited by trade/money management. Fortunately the market throws out trends more often than expected as the market does not exhibit a normal distribution of returns. You still exploit them with trade/money management though.

All profitable traders do is ride the "good" outliers whilst cutting the "bad" outliers. All the rest is simply noise.

That's why "everything works, and yet nothing works" in the markets.

 

[motorway]

Will tomorrow be up/down/sideways?

Wrong question to be asking


motorway

 

[brty]

motorway,

Wrong question to be asking

agree 100%

brty

 

[brty]

Michael D,

Trends arise in random series

Could you please give one example of a random series??

the market does not exhibit a normal distribution of returns

Hence not random, by the definition of random.

brty

 

[MichaelD]

"Will tomorrow be up/down/sideways?"

Wrong question to be asking

Totally agree, but it is the question most asked of a chart.

Could you please give one example of a random series??

The sequence of Boys/Girls born in Australia in 2008.

This produces a binomial distribution, not a normal distribution There would be many, many runs of 10 boys or 10 girls in a row in this dataset. i.e. trends within a random series.

Random <> normal distribution of results.
Random = unpredictable outcome.

 

[brty]

Michael D,

Random <> normal distribution of results

This produces a binomial distribution, not a normal distribution

Ok, sounds like I don't have to add anything here.

Random = unpredictable outcome.

A man walks out of the forest for the very first time, with no previous contact with modern man. He sees a strange low flat black rock extending in either direction as far as the eye can see. It has a white line down the middle. He then sees these strange metal boxes with people trapped in them going one way or the other.
Whilst carefully observing he notes what he thinks a pattern with where most come from, but further observation later in the day 'proves' it wrong. The final count being 1000 metal boxes going each way with a random pattern that seemed to 'trend' at one point.

The following morning while observing this strange phenonomen another person walks up to the forest man and mentions "peak hour's a bitch isn't it" .

Not knowing how to calculate all the variables does not make an event "random". The sequence of Boys/Girls born in Australia in 2008 only appears random to you as you do not have all the information necessary to determine the outcome. Certainly many variables to be known to work out such a sequence, but if it was "random" then some of the lengths of pregnancy, with the large sample, would approach "infinity" to be truely "random".

Do you know of any women who have had a successful pregnancy that approached infinite time in length??

brty

 

[Timmy]

Thinking in terms of coin tosses
where the bias on the coin is created by the bets people make on the coin
Is the type of thinking that will take you very far..



motorway

I like this, a lot.

I first woke up to how un-random things are by tossing a coin while trying out a new system for commodity trading. By giving the coin a toss with quite few revolutions, probably about a foot high into the air, I scored 20 heads in 25 tosses. Thinking this ridiculous, I started again, this time with tails up on my finger, I had heads up the first time. The result was about 15 tails in a row before I gave up as I was not getting anything close to "random". Only by deliberately changing the force of flip was I able to get results approaching 'random'.

My argument is about how an input that is controlled can be classified as 'random'.

By controlled, I mean by force applied in the coin tossing case, but everything else that is considered random also has 'force' of some type or description applied. (a computer random number generator is simply a function of the algorithm applied)

I have not seen anything that could be described as 'random' and do not believe in the concept.

brty

Interesting stuff thanks brty. Not many people bother doing an actual test of ideas like this and the results are sometimes surprising. However, would you accept that some things are not knowable/controllable and thus for practical purposes best described as random? I am thinking of an explanation on www.random.org ("The randomness comes from atmospheric noise"). Don't know if I am getting too far off topic here ... or if what I am saying is relevant to trading.

 

[brty]

Timmy,

would you accept that some things are not knowable/controllable and thus for practical purposes best described as random?

Of course there are many things that I don't know or would be extremely difficult for anyone to know, given current technology. Things don't have to be 'controllable' to be non-random. All you need is a past performance and a knowledge that 'something' is having an effect on the outcome. The more knowledge of the 'something' and its behaviour the more predictable the outcome for any individual. However more or less knowledge does not change the nature of the behaviour, just the perception of what is happening from the individual perspective.

For example, with the previous coin tossing of mine, if you had no knowledge of that and met me in a bar and I offered you 2 to 1 odds on the flip of a coin. You get to inspect the coin, to make sure it is not biased, but I get to flip and call. You think it is random, I know it is not. After playing against me, and losing all your money, would you come back again to play tomorrow, or would you have learnt something about "random" coin tosses??


atmospheric noise = chaotic behaviour.

I was wondering when someone would bring that up. What makes something chaotic? To me it equals not having enough information to calculate something, not that the event is 'random'.

brty

 

[Timmy]

Thanks brty

 

[ducati916]

Interesting discussion,

The stockmarket is a stochastic process. That is a sample path leading to an outcome. A random stochastic process does not have an equiprobable outcome.

A sample path in the market would be the movement of US Steel from $225/share to $17.66/share with every [day] price along the way.

Non-equiprobable would refer to say earnings [one example] increasing or decreasing, relative to current price, weighting the probabilities.

Thus, some outcomes will have higher probabilities than others. A stochastic process also refers to the dynamics of events that unfold, or reveal themselves with the passage of time, or the future.

Thus the unbalanced probabilities, are only able to be calculated IF you can predict the future.

If we are agreed that [currently] no-one can predict the future, then, the stockmarket [prices] are random.

jog on
duc

 

[MRC & Co]

I was wondering when someone would bring that up. What makes something chaotic? To me it equals not having enough information to calculate something, not that the event is 'random'.

brty

I would have to side with this guy.

It is staggering how consistently many beat the market, without the use of any kind of traditional money management.

 

[beamstas]

I would have to side with this guy.

It is staggering how consistently many beat the market, without the use of any kind of traditional money management.

If the market is random, not using money management =/= not profiting.

There will be some lucky fools, some unlucky fools, and some people who actually can profit from randomness.

Just to add to the discussion, I'd argue that short term the market is less predictable than long term.

Tomorrow the market could be up or down, it's pretty close to a 50/50 chance.

But in the long term, the chance of the market being up, say over the next 10 years, is most likely higher than a 50/50 chance.

Regards
Brad