Is Technical Analysis bunk?

[ASH]

Re: Is Technical Analysis bunk?

Lukester --

I agree with you entirely that being dogmatic about what "cannot work" is a form of blindness. It makes no sense to disregard observations just because they are contrary to how one thinks things work -- theories never take precedence over facts.

On the other hand, a theory that gets some predictions right but not others (without any explanation about why it applies in some cases and not others), isn't really proven by the things it happens to get right. A theory must be both predictive and falsifiable. Pointing to cases in which technical analysis made predictions which were later borne out by events isn't sufficient to prove the value of technical analysis -- one would have to show both that technical analysis made unique predictions that distinguish it from other methods, and that the rate of success was significantly higher. (I'm not making an assertion here that this isn't so -- I imagine that academic studies must have been published on this topic, so hopefully a more knowledgable member of the community will post a response that enlightens us.)

The point of my earlier post is that technical analysis seems abhorrent to a certain crowd because it generally doesn't attempt to identify any cause-and-effect relationships. Folks like myself are tempted to dismiss such systems out of hand, for precisely the reason that I don't spend much time at my engineering job considering that "magic" explains what the electrons are doing inside my diodes. I am immediately suspicious of any "explanation" that involves procedure without cause-and-effect. However, to my credit, I am willing to entertain the notion that technical analysis may work (which could be demonstrated by the aforementioned statistical study) -- and I was earlier trying to imagine why that might be.

Anyway, the whole issue is complicated by the fact that there are many variables related in non-linear ways, so the behavior of the markets should be sort of chaotic. Any approach to market prediction -- including both fundamental and technical analysis -- can only hope to predict the envelope of behavior and not the detailed dynamics.

[bart]

Re: Is Technical Analysis bunk?

...
I imagine that academic studies must have been published on this topic, so hopefully a more knowledgable member of the community will post a response that enlightens us.
...

There are indeed studies out there, but I know of none that have been done without preconceived notions and/or taking into account the full picture and also showing the use of multiple TA tools at the same time. Taleb's work is invalidated in my opinion for those reasons.
And the old saw about anything can be proven with stats must also be taken into account when reviewing the various studies.

TA is far from foolproof all by itself either, even when applied in a professional manner... as we've seen lately by all the blowups.
But do also take into account the huge profits of the various big bank and investment house trading desks over the decades, like in "program trading". Well over half their profits in recent years have come from their trading operations... and those are 100% computer based and most of the "rules" are TA or TA based.

And besides, TA rulez and all other approaches drool... ;)

[DrYB/C]

Re: Is Technical Analysis bunk?

I’m no expert on TA, but here’s how I perceive its merits:

1. TA, and all other trading rules (“diversify;” “invest for the long term;” etceteras), create orderly trading patterns among outsiders.
2. If computerized trading programs are created based on common trading rules and TA, then it become child’s play for insiders to game the markets to their profit-making advantage.

Maybe 1. and 2. make today’s market rally an actual April Fool’s joke; 1. and 2. may also explain the behavior of commodities right before Good Friday.

p.s. Common trading rules are also programs for the retail investor's behavior.

[Spartacus]

Re: Is Technical Analysis bunk?

IMHO It's impossible to quantify a lot of TA for a computer's use. [2]

edit: this came out wrong - what I meant to write is that no 2 traders will agree on the quantification. I hope the example below is clearer on the issue than the above.

And even harder to write a program that will take people or institutions using TA to the cleaners.

Just as one example, take head & shoulders[*] - what set of rules will you use for that, sufficiently rigorous to be programmed?

It may make sense for a human when you show it visually, but try programming it.

Is the head 50% above the shoulders? 80%? If the left shoulder is 90% below the head, is do you abandon that for a Matterhorn pattern, or an upside down cup & handle, or does your chart have leprosy?

[*] please !!!!!!!!!!

[2] the programs you see are people making assumptions, and "hand-waving" or "eyeballing" the charts, and guessing at a good set of parameters.[3] and not undertaking the task with rigor. The next guy will choose a different set of parameters, as I show with the head & shoulders example

[3] you have to guess because there are usually too many parameters (aka "excessive degrees of freedom") for proper curve-fitting. Like one of my professors once said, "give me 4 adjustable parameters and I'll hide a rhinoceros in there". He was talking about modeling turbulent fluid flow but the point's the same.

I’m no expert on TA, but here’s how I perceive its merits:

1. TA, and all other trading rules (“diversify;” “invest for the long term;” etceteras), create orderly trading patterns among outsiders.
2. If computerized trading programs are created based on common trading rules and TA, then it become child’s play for insiders to game the markets to their profit-making advantage.

Maybe 1. and 2. make today’s market rally an actual April Fool’s joke; 1. and 2. may also explain the behavior of commodities right before Good Friday.

p.s. Common trading rules are also programs for the retail investor's behavior.

[*T*]

Re: Is Technical Analysis bunk?

Re. automation, it's quite easy to write a trading robot using moving averages, for example. A lot of `systematic' hedge funds and traders do this, for example.

I am a convert.

I used to believe the efficient markets crap, but markets are not mechanics, they are made up of people. People believe in voodoo, act in crowds, and use leverage. TA is a self-fulfilling prophecy. There doesn't have to be a causal mechanism.

After years of staring at prices I have come to accept that there is at least something in support and resistance, trend lines and moving averages. When target-setting on retracements, even Fibonacci levels are not as ridiculous as I once thought them.

If you really want a mechanism, I will note that the eff. market hypothesis (EMH)/ random walk requires perfect knowledge by all parties. That assumption is only approximated in real life because information is transmitted via price action. So price history does have some information content, if you look for it.

As for the book 'Random walk down Wall St' proving EMH, bullshit.
That's like saying A Non-Random Walk Down Wall Street disproves EMH. EMH is just that, a hypothesis, and one refutable by observed fact.

Alternatively, I might propose that once much credit or margin is in the picture, price moves become self-reinforcing. In this kind of scenario `trends' do exist (on any given time scale). Then the future market price is a function of the past market price, and the market has become, to quote Soros, `reflexive'. This is counter to EMH.

In the equity options markets, volatility has a term structure, which is traded. The implication of this is that the markets themselves price in mean-reversion or trending.

[*T*]

Re: Is Technical Analysis bunk?

Elliot waves are BS, though, I must admit.