Speaking from a HEDGE FUND STRATEGY perspective, I have to agree with TraderZone. The base strategies the funds trade are very unlikely to be a technical analysis based model. In alot of cases, the Models are not automated and provides a degree of freedom for the traders to base their trades. The individual traders executing them may use technical analysis to add-value to their performance. Grey... Black... Model is a model.
TA is IMHO best compared to descriptive statistics (as opposed to statistical inference). The problem is when the jugdement-based graph reading is reduced to mechanical rules. This is often done in a way that ignores methodological problems such as biases, data snooping etc. Secondly: Much of TA seems to be centered around trend-following as opposed to the mean reversion strategies that are mentioned here. This was posted in another thread: - ultrashort timeframe (seconds to a minute or two) - mean reverting - (think scalping or marketmaking) - intraday (3 minute bars to 60 minute bars) - trend following (i.e. think ORB) - Days to a weeks or two - Mean Reverting (i.e. - a really short period RSI system) - weeks to months - trend following (i.e.- the Turtle Strategy). It corresponds to what is my experience and explains the connection between strategy and trade frequency / time frame.
I don't really see the difference between TA and statistical analysis, most TA is just a derivative of a moving average. Its not like traders came up with the idea of a moving average, its stats 101 stuff. The problem is that in most cases a moving average is like bringing a butter knife to an automatic weapons duel. Most people don't have the math background and/or probabilistic mindset to go beyond this though. Thats why you don't see Analysis of Financial Time Series by Tsay at Barnes and Noble in the trading book section..no one would buy it.
My response would be that the FAR MAJORITY of TA that is peddled is completely subjective and tends to draw inferences from anecdotal observations. When cornered to look at the universe of alternative outcomes, practitioners tend to become defensive and argue that you just don't have the "gift" of vision. Or when their model fails, they explain it away by saying that it needed to be re calibrated to adjust for XXXX. After diligently studying a few TA books, the neophyte eventually encounters nagging thoughts like, why is it that when I tried the author's explanation on a different chart, the results came out different, they begin to realize something is lacking. This is where the true knowledge begins. Whereas, statistics attempts to look at the entire body of information and attempts to draw conclusions as objectively as possible. As an objective approach it is usually a more sobering view of truth, as it does not have the luxury of cherry picking observations to draw conclusions (although it does have a word for it; termed bias). Statistics is not without it's warts, but it is immensely more useful to describe information in an objective and truthful manner IMO. That's not to say you can not approach conventional TA using statistical tools. Very few have taken the time to do this. Very few books advocate this. In summary, no: TA as it is typically defined in the myriad of books is not equal to statistics.
Though, in some cases it is. A friend of mine loves to play ranging markets, and then play the breakout. He firmly believes that the breakouts are accelerated because of a rapid covering / selling of the positions that have queued up due to the ranging market. I explained to him that when he played the resistance and support of ranging markets, he was really playing the probability that price would not extend further -- based on the recent price distribution established. He also inherently knew that the distribution had a fat-tail -- so if the 'resistance' or 'support' was broken, he had a good chance of excess gains piling in the direction of that tail. He just didn't realize that what he was doing could be broken down into statistics... I think a lot of TA is that way...you're either playing the drift or you're playing the noise. Either way, there is a distribution present...
Things you can research: goto www.deshaw.com see the description of their strategy. research peter muller at Morgan Stanley - the founder of process driven trading. Look into information propagation. Books- How I became a quant- Schacter Calculated Bets - Skiena Tradezones and others are correct- all TA books are garbage- you want to capture price action in whatever valid way - TA is like taking the cross section of a dead animal in a lab- you can see some of the guts, and thats about it.
Ya they've made 20%+ a year with 30 billion but some grunt on ET expects to be spoon-fed their strategies. Renaissance is even quieter- the mice in long island make more noise.
Their most successful strategies are connections and access to information which we can never get; Until after the fact of course.
Why would someone go long GS short LEH or any other spread? When LEH falls hard, GS will fall too and not rise. The trend of the spread will be shorter than going short LEH and skip the GS leg. More profit for less buck and less commissions. A spread always has lesser trend and more chop. So u trade it antitrendwise and wonder when things dont reverse to the mean. I still believe in the saying Spreads = half profits, double risk.