Why does anyone take technical analysis seriously?

[Sergio77]

TA covers a wide spectrum

1. Chart patterns
2. Indicators
3. Candles and price patterns
4. Backtesting
5. Evidence based analysis
6. Descriptive and predictive statistics

This is the best latest book money can buy on this subject

Many charts from recent hyped up but failed TA patterns
Reasons to avoid backtesting constantly
Data-mining and data snooping bias issues

You will get a good idea with what you are dealing with when it comes to TA. Book filled with honesty and clarity.

 

[Troy Bombardia]

I agree. Technical analysis is important but only to a certain extent. It should be treated as a guide, not a crystal ball into the future.

Realise that TA is just a matter of probability. Eg when indicator X reaches 20, there's a 70% chance of the price going up.

http://investingtrack.com/will-the-sp-500-make-a-new-low-in-the-next-3-weeks/

 

[dartmus]

Yes indeed, imagine looking up and seeing it everywhere. It's truly beautiful! Definitely puts a smile on my face to see it!

 

[dartmus]

As I see it, TA is just a specialized form of "descriptive statistics":
https://en.wikipedia.org/wiki/Descriptive_statistics

There is obviously nothing wrong in making such quantifications, just as it makes perfect sense to compute other descriptive statistics.

The issue arises when you wish to attribute inferential value to such observations, like for instance when you wish to interpret it as a "premonition" of a future occurrence:
https://en.wikipedia.org/wiki/Statistical_inference

It's mainly a conceptual issue. Imho, attributing previsive value is, first of all, a crackpot nonsense conceptually, and, second, not supported by any sound statistical evidence (clearly if we exclude pseudo scientific inventions by charlatans and people with no solid statistical background).

This does not mean anyway that it's "wrong" to use them to "decide" entries or other actions. In any case you entry is governed by price movement (and/or other observations) . So in case we model the price as a random process, the "entry" (say for instance, the pair [price, time] ) too is going to be a random variable, and basing the entry on some TA computation is simply a transformation of a random variable, resulting in another random variable. So there is nothing wrong in that, as essentially everybody is inevitably doing the same.

I would just argue that the terminology may be a bit misleading, especially for beginners. (Just as the term "stop loss" is misleading.) But these terminologies are typical of the field where there are so many entities trying to sell something and often do not mind to bend concepts and reason when it is convenient.

Brilliant answer! Thank you!

 

[Laissez Faire]

You confusing the person with the discipline. as you say the subjects listed follow protocol accepted across the board--- when a person fails at them it is the person who fails.

The point I was trying to make is that these subjects are very hard to master even when the curriculum is clearly defined and these people have all resources at hand to help them. It's simply not a matter of merely discipline, but intelligence and savvy.

With trading, you don't have a clearly defined curriculum and you do not have a mentor to guide you except snake oil sellers and failed traders on ET. Thus, my point is that learning to trade is much, much harder and it shouldn't be a surprise then that most people fail with TA. A high failure rate with TA is simply not a fair argument against it's validity.

Further, I think you're the one being naive if you think that very profitable methodologies and complete systems would be shared in the open to the public.

Most of the backtesting on TA that I've seen in academia seems to be based on merely testing patterns or indicators. But this is clearly lacking context and it's very superficial.

SInce TA itself is flawed, its not the person, its the subject.

TA itself is not flawed, i.e., using past prices to forecast the future. I do however agree that subjective interpretations of charts is a slippery slope and one that will fail most people.

PS-- where are you at school?

Norway.

 

[Mtrader]

With effective money management -- random entries based on your market bias -- its possible to make a killing.

So you know how to make killings? Then I have a few questions for you:

1. how did come that you blew up a fund? you should be making killings!
2. why are you not one of the richest men in the US? you should be making killings!
3. why are you writing and selling subscriptions? you should be making killings!
4. if your statement is true you should be trading trading trading trading trading trading, but you are not?????

You know how to make killings, but you do all kinds of things except these killings.
I say that TA works and I am consequent: I trade!

 

[kut2k2]

Why do some people actually base their investing decisions on technical analysis?

Because when done right, it works.

http://www.elitetrader.com/et/index...fitably-backtested.147576/page-5#post-2218032

 

[Mtrader]

Maybe the following is undeniable proof that markets are not random, and this implies that TA can work:

1.If you throw a dice, the smallest possible value is 1 and the biggest possible value is 6. Each time when you throw, the value will be at least 16.7% of the previous one (from 5 to 6) and maximum 600% (from 1 to 6). So the outcome from the next throw will be between 16.7% and 600% different from the last throw. This is a huge range.

2.If you take all the closing prices from the s&p since 1/1/1950 (16544 days of data), then we see the following:

• The closing price from any day is always between 91% and 111.5% of the previous close. There was only 1 exception on that: 19 october 1987, when the markets crashed. Then it was 79%.
• If prices would be random the value should vary between 0 and infinite. This is clearly not the case. The prices even stay in a very narrow range. Even if we would narrow the random range to 0 till 200%, the real values use only 10% of this range. This creates a very high probalitity, which is never found in real random data. Here TA comes in the game.
• For 65 years the price never dropped more than 10% and never rose more than 11.5% from one closing to the next one. If you throw a dice the only thing that is sure is that the outcome will never be more than 600% whereas in trading the outcome will be between 91% and 111.5% of the last close with a VERY HIGH probability. The range of about 20% is also much smaller than the range for throwing a dice which is almost 600%.

 

[NoVoodooHere]

I'm currently a student in university doing a quantitative degree. What I've learned is that there are plenty of people who fail to understand mathematics, statistics and other science related subjects. And these are people who can study a clearly defined curriculum with all the world's resources at their hand to help them.

Still, many of them fail.

Does this mean that mathematics and statistics doesn't work?

Perhaps a distinction between statistical methods & "technical analysis" is in order... Most "traders" that use "TA" are not sophisticated at all; they do not understand the most basic concepts of data analysis, they sit & stare at lines on charts; chart monkeys. Just as an FYI

 

[d08]

There is still no consensus on ET what qualifies as TA, therefore the debate won't lead anywhere. Some say it's using indicators, others are saying it's about chart patterns and then there are the ones claiming it's using any historical data to determine future prices.

Personally, I find indicators to simply normalize price movements, making it easier for code to process. The idea that there is some "magical indicator" is silly and has been explored to exhaustion.

A few chart patterns that I trust work because they describe market psychology on a chart, again, there is magic involved - it's about one side of the market giving up and the other side dominating, leading the price in one direction.