The Perils of Prediction

[SOCRATES]

The whole thing is very peculiar Bramble. You know, for example, when I do not wish to discuss something for whatever reason, I say so and I go on to state the reason. But what has been encountered so far with this is a response so marked as to be alarming, almost as if the individual asked considers it a personal affront to be even asked, it is that dark.

 

[kriesau]

Dead thread - I wonder why !!!!!!!!!!!!!!

 

[DaveJB]

Okay,
here's something to ponder then, on the perils of prediction - chaos applies to non-linear systems... a linear system would be one with a fixed frequency (in market terms I'd suggest the market would only be considered linear if, for example, the same thing repeated every n days or periods on a chart... so I think most people would agree markets are non-linear systems).

An interesting effect of chaos, and one I believe is readily apparent if you are ever driven to view graphs of non-linear system input v output, is that for apparently identical inputs you can get wildly different results... theoretically, if you knew everything about the universe and could completely recreate the moment (so to speak) you can repeat the outcome exactly - but in reality you can't do this. The outputs from non-linear systems (markets) then tend to diverge greatly the further out in time you go.

I think this has got heavy enough already, so here's the point - a non-linear system can have an equation goerning it that is not actually all that complex, but they have things like square and cube functions in there and these functions are what make the outputs differ widely from virtually identical inputs. The possible outputs tend to cluster about specific values, and it's probably easiest and simplest to describe it (roughly) as 'the further on in time you run it, the more of these cluster points tend to appear, and the more random the output appears to become'... the output is not random - the inputs are fixed values (although we may not know them <g>) and the equation is the same every time... but the output chart can look completely random... ring any bells?

This suggests three things to me -
1) Forecasting, especially via PC program, is probably more accurate over the short term. (You might consider 10 bars ahead to be the same for minutes or months, I suspect that's so but the price spread would be larger for the longer term bars).

2) It's entirely possible that you might produce an equation that delivered excellent short term predictions... the trick here is not only to get the equation right, but to figure out how far ahead it started to produce the multiple outputs (price targets at substantially different values)

3) It may be that the human brain, with it's quite amazing pattern deduction capabilities (we can recognise people from seeing small parts of a face, for one example) is an essential part of homing in on the correct equation.

Dave

 

[SOCRATES]

I disagree. In the markets there is perfect order. There has to be otherwise there would be no market at all and prices would go from left to right in a perfect horizontal line on a chart if all information is known and acted upon correctly. Everything would be in a state of perfect equilibrium. The problem is that within this order there is apparent disorder caused by not everything being known and by many participants being unable to act correctly.This creates a conflict which ultimately appears as if all of this is random and chaotic, which it is not.

 

[DaveJB]

Sorry folks,
you are using the colloquial version of chaos, as in 'no order to it, random' which is NOT what I said. I said the market was a non-linear system, which is therefore chaotic (scientific version of the word) one example of such a system is the weather.... if you could know and reproduce the exact initial conditions you could predict with certainty the outcome (ie forecast the weather), but the slightest change to the initial conditions is magnified by those square functions etc to make 'what happens' deviate markedly from 'what was forecast' after quite a short period of time.

That isn't the same as random, unpredictable - what it means is that IF you can derive the equations to describe the market (ie if you can work out a set of rules to follow to produce your price forecast) and you do a good job of it then you will be able to forecast with some certaintly what is going to happen over the near future.... but the further ahead you try to forecast the more the non-linear functions will come to dominate the output and the less and less sure your forecast is going to be, until you don't actually have a clue.

ie you might forecast the price for a specific period on Monday and get it right pretty often, but you'll find it harder to forecast next friday's close, and you'll probably not have a clue what the market will close at in a month from now.
I'd hardly be using TA to try to trade the markets myself if I thought them unpredicatble, now would I?
Dave

 

[SOCRATES]

Markets are the opposite to this, as a consequence of all of it being planned in advance. Equations do not come into it at all. What comes into it is a complete understanding of how everything fits together and the clues given as to the ultimate intent, and sussing this, in advance of the result is the desired state of affairs and not trying to quantify or qualify it just by TA, or fundamentals or other clumsy methods.

 

[peterpr]

Markets are the opposite to this, as a consequence of all of it being planned in advance.

Planned by whom?

Equations do not come into it at all. What comes into it is a complete understanding of how everything fits together and the clues given as to the ultimate intent,

Whose 'ultimate intent'?

... and sussing this, in advance of the result is the desired state of affairs[

Desired by whom?

... and not trying to quantify or qualify it just by TA, or fundamentals or other clumsy methods.

A high proportion of equity/futures buy/sell decisions are made by traders under instructions to aquire/dispose of stock. The decisions reponsible for the largest volume of such aquisitions/disposals are made by professional fund managers who in turn are under pressure from the investments/redemption dynamics of their funds and I can assure you that they commonly make use of what you describe as 'clumsy methods'. This in turn is influenced by a myriad of things, among them the slickness of their principals marketing machine; the funds performace; current public investing psychology; the need to find a home for pension contributions/take pension benefits etc etc - this list is endless although the marketing machine bit has a lot to answer for (I was such a manager through the 1987 crash and therefore can claim to know a little bit about the subject).

If you are implying that there is a group of fixers who decide the movement of any market in advance (short of a sovereign government on the odd occasion or government sanctioned operations to manipulate relatively small. illiquid markets like gold and silver with decreasing and destabilising effect), then frankly you are plain wrong.

Or perhaps I have misunderstood what it is you are trying to tell us.

 

[barjon]

A high proportion of equity/futures buy/sell decisions are made by traders under instructions to aquire/dispose of stock. .
.

Peter,

Doubtless Socrates will respond in his own inimitable way, but haven't you answered your questions with your first sentence?

A trader under instructions to buy or sell a huge block of GSK, say, cannot just stick an "at best" order in for the whole lot. Presumably, they will plan and execute a buying/selling campaign over a lengthy period and the well-trained eye (not mine ) can spot what they are up to.

good trading

jon

 

[peterpr]

Peter,
..... but haven't you answered your questions with your first sentence?

A trader under instructions to buy or sell a huge block of GSK, say, cannot just stick an "at best" order in for the whole lot. Presumably, they will plan and execute a buying/selling campaign over a lengthy period and the well-trained eye (not mine ) can spot what they are up to.
jon

Yep agreed; that's exactly what he will do with fakes and fades and a host of stratagems. But one - or even many simultaneous - such orders doesn't constitute the market. The idea (if that is what is being suggested) that anything other than extremely short term moves can be successfully planned and implemented by some kind of savant inner circle, which can in turn be picked up on by savvy traders, is to my mind absurd. Otherwise it would amount to conspiracy to defraud (which, come to think of it, does bear some resemblance to historical fact actually - tongue-in-cheek smilie need yet again !) But as you say, maybe that is not what is being suggested.

I'd just like to figure out what is, that's all

 

[SOCRATES]

Barjon has adequately answered your question, there is no need for me to add to what he says.

 

[peterpr]

Barjon has adequately answered your question, there is no need for me to add to what he says.

Socs

There were 3 questions; and please accept that I wasn't being facetious.

I have long been a stickler for the precise use of language. Heaven knows it's easy enough to talk at cross purposes without vagueness adding to the problem. So forgive me for suggesting that your contribution to this so far is at best a little vague.

Any reasonably intelligent person could be forgiven for thinking that your meaning is/was as I inferred in my reply to Barjon. If that is not the case then it is also reasonable to ask you to elaborate. So, that's what I am doing.

 

[DaveJB]

I thought the thread title inferred the discussion was related to predicting markets? If the markets are preset Soc, then what will the Dow close at next Tuesday?

In response to an 'its all gone dead' post I then posted something drawn from science on predicting the future course of systems... it struck me that rather than going round in the same circles endlessly some members might like to consider how other people forecast the outputs of complex systems, and some of the limiting factors that affect the reliability of those forecasts.

I don't predict markets, I trade what I've just seen happen and I try to improve that by making my response to what I see on the chart occur as soon as the evidence is visible. The limit of my prediction is that what I see happening is going to continue long enough for me to extract a profit. I am right often enough.

However, I do have some understanding of prediction drawn from other areas outside finance... just as MESA, for one example, results from engineering studies that have been applied to finance, so too might other fields enable the astute to avoid reinventing the wheel.

What I don't understand is how having taken the time to describe the difference between the common use of chaos and the scientific one in the first post, specifically stating that chaos is non-random , the immediate rebuttals all assumed I had described markets as "chaotic therefore random"... queries from anyone who didn't understand that point I'd be quite happy to cover, as they all clearly demonstrate a 100% incorrect understanding of what I posted. I specifically stated the outputs are not random.

One follow up pointed out the real problem in this sort of calculation - unless the inputs are known precisely then the result becomes more and more unpredictable, in fact the longer range the prediction is intended to cover the more the actual result is possibly going to deviate from the forecast.... but you could backtract from the actual result and calculate the actual precise inputs (for all the good that would actually do you). THIS is the point of the chaos as I originally described it.

Dave

 

[peterpr]

What I don't understand is how having taken the time to describe the difference between the common use of chaos and the scientific one in the first post, specifically stating that chaos is non-random , the immediate rebuttals all assumed I had described markets as "chaotic therefore random"... queries from anyone who didn't understand that point I'd be quite happy to cover, as they all clearly demonstrate a 100% incorrect understanding of what I posted. I specifically stated the outputs are not random.
Dave

Dave

I found that re-start post interesting and your explanation of chaos -v- random was clear as chrystal. The only reason I didn't respond was because I found nothing to take issue with.

I too find chaos theory and non-linear systems fascinating. However, unless human behavioural responses (systems) are as scientifically reduceable and predictable as inanimate physical ones, then systems designed to react with and predict them (human responses) will always run up against severe limitations. The obvious outer limit being that if their application results in only buyers (or sellers) then there is no market. Between the extremes, there is probably something to be gleaned from applying them to market analysis.

I try to keep abreast of any 'new' market theory/methods/systems, if only because my trading methodology requires me to decide real-time what/who is driving market action and to second guess how the professionals will react to it. For example, a particular price level becomes crucial/critical/important ONLY if enough traders consider it to be so. It is their (often pre-planned) reaction to a particular level being reached that is the major short-term determant of the ebb and flow of price (IMHO). So how they arrive at their 'important levels' is of great interest to me - even if it is 'Mars is in conjuntion with your rear-end' so-to-speak To that extent I try not to pooh pooh any system; trying to understand the psychology at work is more profitable.

In that regard, I'm sure you will be aware that there is a Bradley sidero-whatsit turn date Tue/Wed next week. Now, I personally find that an itsy bitsy way out but, a lot of people don't ; and a lot of people acting all at once on a shared belief, however irrational, can move markets

So I'd rather be aware of it beforehand - Just another piece of junk in the toolbox.

 

[SOCRATES]

I thought the thread title inferred the discussion was related to predicting markets? If the markets are preset Soc, then what will the Dow close at next Tuesday?

In response to an 'its all gone dead' post I then posted something drawn from science on predicting the future course of systems... it struck me that rather than going round in the same circles endlessly some members might like to consider how other people forecast the outputs of complex systems, and some of the limiting factors that affect the reliability of those forecasts.

I don't predict markets, I trade what I've just seen happen and I try to improve that by making my response to what I see on the chart occur as soon as the evidence is visible. The limit of my prediction is that what I see happening is going to continue long enough for me to extract a profit. I am right often enough.

However, I do have some understanding of prediction drawn from other areas outside finance... just as MESA, for one example, results from engineering studies that have been applied to finance, so too might other fields enable the astute to avoid reinventing the wheel.

What I don't understand is how having taken the time to describe the difference between the common use of chaos and the scientific one in the first post, specifically stating that chaos is non-random , the immediate rebuttals all assumed I had described markets as "chaotic therefore random"... queries from anyone who didn't understand that point I'd be quite happy to cover, as they all clearly demonstrate a 100% incorrect understanding of what I posted. I specifically stated the outputs are not random.

One follow up pointed out the real problem in this sort of calculation - unless the inputs are known precisely then the result becomes more and more unpredictable, in fact the longer range the prediction is intended to cover the more the actual result is possibly going to deviate from the forecast.... but you could backtract from the actual result and calculate the actual precise inputs (for all the good that would actually do you). THIS is the point of the chaos as I originally described it.

Dave

The answer is as follows:~

If the current conditions prevalent in the market do not significantly change then a close higher than Friday's could reasonably be expected.

 

[SOCRATES]

Socs

There were 3 questions; and please accept that I wasn't being facetious.

I have long been a stickler for the precise use of language. Heaven knows it's easy enough to talk at cross purposes without vagueness adding to the problem. So forgive me for suggesting that your contribution to this so far is at best a little vague.

Any reasonably intelligent person could be forgiven for thinking that your meaning is/was as I inferred in my reply to Barjon. If that is not the case then it is also reasonable to ask you to elaborate. So, that's what I am doing.

Answer No 1: The planners.

Answer No 2: The intent is the ultimate intent of the planners, which may have one of three outcomes, up, down or sideways, even if the route chosen to get there seems unnecessarily circutious, or random, or prolonged.

Answer No 3: The consequences of not making the ultimate direction clear, is that nearly everyone is kept guessing, therefore the object of the excercise is to suss the intent, and act in accordance with it, and not against it. The desired state is that of everyone participating in markets whether privy to the intent of the planners or not, to be able to act in accordance with developments.

 

[DaveJB]

Cheers Peterpr,
you've cheered me up no end <g>

I have to respectfully take issue with you Soc on 'up' - I took it that this thread's aim was to discuss the possiblity of forecasting to a somewhat greater extent than 'if nothing changes it'll go up', although that is not to dismiss this as contemptuously as some - a significant number can't pick the market direction currently in view, let alone decide what they expect it to be in the future.

I'm not convinced that markets are a 'special case' Peterpr - yes the wide variety of investors/traders pick a number of price points to react to, and it's arguable that the inclusion of human choice in the process adds a factor of great complexity, but then many of the other systems being predicted have highly complex inputs. The weather forecasters need to know the position, temperature and humidity of every air molecule for example before any hope of an accurate long term prediction could be made. This does not prevent reasonably accurate short term forecasts being achieved. I do think it would be extremely difficult to manage this however, and in trading terms it is more practical to simply observe what people are actually doing, and to then proceed to try to take profit from it.

Dave

 

[DaveJB]

Why is this thread degenerating, as others have, into a personal slanging match?

I disagree with Soc on occasion, I'm not going to persuade him about chaos theory's potential application to market prediction, but so what? There's no requirement to ensure people holding different views end up in agreement - the threads on here are for discussion, swapping ideas, and (I thought) an element of friendly chat between like minded individuals - if somebody disagrees with you, and you can't persuade them to change their opinions fairly quickly, then it's pointless to keep on about it like a dog worrying an old bone.

Dave

 

[barjon]

Well, I see it's week-end knock-a-bout time again . When you've finished you won't object when I delete most of the recent off-topic posts will you (unless you get there first of course )

jon

 

[SOCRATES]

You spoilt all the fun, Barjon. At the same time you prevent the members from being exposed to classic examples of real ignorance, real ingrained, persistent ignorance. In the same way it is valuable for them to be exposed to real knowledge, it is equally important that they be exposed to real ignorance, in order to help them recognise it at a glance, with neither doubt, nor difficulty, nor hesitation.

 

[SOCRATES]

Why is this thread degenerating, as others have, into a personal slanging match?

I disagree with Soc on occasion, I'm not going to persuade him about chaos theory's potential application to market prediction, but so what? There's no requirement to ensure people holding different views end up in agreement - the threads on here are for discussion, swapping ideas, and (I thought) an element of friendly chat between like minded individuals - if somebody disagrees with you, and you can't persuade them to change their opinions fairly quickly, then it's pointless to keep on about it like a dog worrying an old bone.

Dave

That is correct. We may occasionally disagree, but never a cross or rude word between us.

Now this question of chaos theory's potential application to market prediction could lead to an interesting interchange of ideas. Where do you begin with this ? Do you postulate the hypothesis that chaos is an integral element in prediction ? And if so, what is the root core of this hypothesis, if so ? All this provided people behave, otherwise Dave and I will go the PM route with this. Now you have been warned. And then if we are forced to continue our discussions via other routes, then no one benefits.