[dirtyaura]

A tangential question: It seems that a lot of smart people believe in technical analysis, but to me it sounds like telling the future from tea leaves. Does it really work or are successes just part of the standard randomness of stock investing?

[chollida1]

This is a good question, and I don't think you'll get a definitive answer.

If by technical analysis do you mean watch for patterns to appear, then yes it works. There are patterns all over the place.

Renaissance Technologies is famous for its pattern matching AI. They hired 2 key employees out of IBM many years ago that layed the base for their technology and the rest has been money making history. See: http://www.businessinsider.com/bob-mercer-peter-brown-2010-3 and http://en.wikipedia.org/wiki/Renaissance_Technologies

The world of trading has been a cat and mouse game of pattern matching for awhile. One of the earliest attempt at hiding large orders was an algorithm called POV( Percent of Volume). It's an order that would slice up a big order into smaller chunks and sell it throughout the day. The first variations would just sell every 10 minutes. Its easy to see how someone could find this pattern (Hmm, it seems like 1,000 MSFT are being sold at market every 10 minutes by Goldman Sachs) and exploit it which lead to more intelligent order spreading, and the cycle continues.

However, if by technical analysis you mean looking for patterns like "head and shoulders" (http://www.investopedia.com/terms/h/head-shoulders.asp) then it might be true only in that if so many people/computers believe in it that it becomes a self fulling prophecy. For example, if every one believes that when a stock crosses above its 20 and 50 day moving average then its going to fall, then it will fall just because everyone will start selling because they believe it will fall, which causes it to fall which reinforces everyone's belief that the pattern works and you have a positive feed back cycle.

Does that mean technical analysis works? IMHO this type of investing doesn't work, but who am I to say...

[TDL]

I've come to the conclusion that those who successfully employ technical analysis (i.e. chart reading, etc.) are practicing a form of risk management (knowingly or unknowingly), but do so consistently. In short, they are quickly out when they are wrong (taking small losses) and allow their 'winners to run'.

[dirtyaura]

Yes, I meant the latter version, i.e. manual detection of patterns from the price charts. Given how optimized modern trading is, I find it hard to believe that a crude method like this would work.

I totally believe that with algorithm trading you can detect patterns, but I would assume that the sweet spot - like you described - would be detecting and exploiting patterns of individual actors (or a cluster of actors) instead of the patterns of the chaotic mass.

[joshu]

Pov would be an execution strategy. And exploiting it seems almost by definition not technical analysis.

The rest seems accurate.

[wcgortel]

It really depends how you use it. "Technical analysis" refers to techniques that are as disparate as "cooking". There is evidence of some technical stuff being quite valid--the relevance of momentum to returns, for example--but there are also plenty of people who misuse it.

My own view is that it's a window into the pricing process and adds depth to the practice of reading the market that many analysts go through. Ignoring technical analysis when looking through charts is kind of like watching TV in standard definition instead of HD. Even though it's kind of still the same thing, I enjoy it.

[JTon]

The most compelling argument I've heard is that technical analysis is a self-fulfilling prophecy. I.e. If a large enough segment of players believe the accuracy of a certain indicator, and they all take positions reflecting this belief, and the sum of actions can move the market. Personally I've grown fond of behavioral analysis (I suppose that's a subsection of technical analysis), in the form of the candle stick technique. I think it works more often than not

[jonknee]

If you're doing technical analysis yourself it's a joke and indeed reading tea leaves. However, technical analysis is essentially what high-frequency trading automates and has made many people billions of dollars. You have no chance competing though, by the time you have identified your Fibonacci retracement or whatever other snake oil you're looking for, there have already been millions of actions taken by hundreds of HFT firms that have likely already eaten your lunch before you knew you were hungry.

[encoderer]

I believe there's efficacy to it as one input; it's not an exact science.

Technical analysis is just acknowledging that while price moves are random (and if you plot daily price moves, they represent a bell curve with fat ends), there are still patterns that reappear. And if you have tight stops in place to manage risk, you can (try to) take advantage when a pattern emerges.