Technical analysis: why is it used in some markets (commodities) more than others (IR)?

[gc]

I wonder how many of you use technical analysis for trading, and if so what is the value you assign to it as a trade-opportunity indicator.



I am not a trader and I always dismissed it to be little more than a poor brother to astology. But having seen how widespread its use is on our trading floors, I am starting to think that I must be wrong.



The fact is that if I approach three traders and:

- the first one talks to me in terms of relative value analysis (even if that reduces to simple linear regression between instruments that have fundamental reasons to be co-dependent) and points out a trading opportunity,



- the second one talk in terms of arbitrage free pricing and points out that based on some model (even if they are as simple as G2 or  Heston) an instrument can be dynamically replicated in a cheaper way than the market prices, and points out a trading opportunity,



- the thrid one shows me some charts with some fibonacci numbers, combinations of heads and shoulders, all made it more scientific by over-imposing oscillators,



I immediately think that the ideas of the first 2 may not be correct, but at least there are some logic foundations for their reasoning and I'm willing to assign their bets some edge.



If I talk to the third one, I think that - in the best case - he is fooling himself, in the worse he is using charting as a way to build smoke about confused ideas that he cannot elaborate in a quantitative way almost certainly because of a lack in analytical skills and study. I would immediately think he is a charlatan.



Saying that, every time I go to the bathroom I cross the commodities trading floor and the screens of the traders are always filled with all kind of charts, lines, candlesticks, trend-lines, etc. And they seem very keen about discussing about the proximity of the current point to a support line, etc.

What is even worse for me, is that some of them look at charts (assuming that past information is relevant for current data) and then talk of volatility in a Black-Scholes-Merton framework (assuming that past information is irrelevant for current data). So I think that they don't understand the foundations of technical analysis and of APT to see that they are mutually exclusive, and they are very confused.



The fact is that there are many commodity traders and they (not me as an analyst) make the money, so if one is right it must be them!



Should I rethink about technical analysis? If so, in which roles is it useful?



My field is interest rates, and when I talk and discuss with our traders, nobody uses technical analysis anymore and discussions are more mathematical.



This opens another question: are our interest rate traders smarter/more analytical/more thoughful than our commodity traders?



Or are commodity and interest rates markets so fundamentally different that a more thorough mathematical description of commodities at the level it is done in interest rates is impossible?



thanks

[AndyM]

Hey GC, long time no see.



The short answer is that fundamental value is a more elusive concept in commodities. Is the fundamental value of a barrel of oil $20, $130 or $250? Hard to say. There are very few agreed valuation metrics, unlike IR or other products. Even if you build a fundamental model, the demand and supply curves are so inelastic, especially in the short-run, that small changes in supply and demand fundamentals can have enormous price impacts. So better to quietly shelve these models and go with the flow.



Hence the proliferation of technical analysis. It fills the vacuum.

[cygnet]

I'm no expert on technical analysis, but certainly some of it appears to work. For example, support / resistance at round numbers seems to be present in FX markets.

[Johnny]

Technical analysis pre-dates modern statistics by about half a century (moving averages, support/resistance, etc) or several centuries (candlestick charts). A kind interpretation is that technical analysis is an early form of statistical analysis adapted to use on financial markets in which volatilities are not constant and regimes change. Moving average cross-over is a low-pass filter, break out is a test for change of noise distribution, etc etc. It took statistics a long time to catch up, but now I think it's fair to say that modern statistical methods are much more effective than technical analysis, if only for the reason that the barriers to entry are higher.

[Jim]

I've found technical analysis to be of use in at least two situations:  First, when I can use a given indicator to measure the presence of a specific market micro-structure behavior, and secondly, in the FX and IR markets where monetary policy objectives differ from what is actually happening.  In both those cases there are motivations beyond profit maximization, and I think technical analysis can help detect some of those situations.

[FDAXHunter]

Other threads with a similar theme.



[url=/Show%20Post.aspx?PostIDKey=6851]What exactly is TA, and has merit in TA and what does not?[/url]



[url=/Show%20Post.aspx?PostIDKey=106839]Are Technical Trading Strategies Considered "Quantitative"[/url]

[tristanreid]

Technical analysis can be a really useful tool, if you use it correctly, I think over the long term you can correctly predict the direction of the market with about 50% accuracy.



But if you don't use it right, you're wrong half the time.



-t.

[FDAXHunter]

I'll take 50% correctness as long as I have overall positive expectation Wink

[gutenberg]

Andrew Lo (of MIT) suggests it's largely an issue of terminology. HE gives the following example, compare:



The presence of clearly identified support and resistance levels, coupled with a one-third retracement parameter when price lie between them, suggest the presence of strong buying and selling opportunities in the near-term.



with:



The magnitudes and decay pattern of the first twelve autocorrelations and the statistical significance of the Box-Pierce Q-statistic suggest the presence of high-frequency predictable component in stock returns.

[nnja]

I'm no expert in the area, and I generally believe that technical analysis applied to any old time series is nowhere near as explanatory as more modern statistical analyses, but if a large enough number of traders are all, say, looking for the moving average to cross some volume threshhold before putting in sell orders (again, I don't know anything about this), then guess how the price is going to move. As AndyM pointed out, in areas with little "weight" in fundamentals, it's all just shapes and colors and whatever arbitrary rules that the market has self-organized itself into will be, within limits, sustaining.