|
|
|
|
| Now...you have the edge. | |||
| Home | My Trading Journal | Archive of Research Notes | Play HyperDonut! |
|
Being a geek, it's only natural that I'm primarily a technical trader. However, I've never been satisfied with the current state of Technical Analysis (TA) because of the lack of a widespread, rigorous scientific approach to the subject. Here are four main problems with technical analysis that I see: 1. Classic chart-reading is too subjective:
"It's a rising wedge of course!"
Yeah, right. The problem here is that human beings tend to look for patterns, so the more patterns that chartists "define", the more likely it is that one or more of them will show up in a chart. There's no confirmed uptrend on the chart? What about a reverse head & shoulders? No? OK, well how about a rising wedge or a falling flag, or maybe a cup with handle, rounded bottom, or double top? Get the idea? If you can't find a pattern, just keep inventing new patterns until one shows up on your charts. This isn't helpful, and the problem is exacerbated by the fact that most of the "patterns" in classic chart reading are not rigorously defined. By this I mean that it would be difficult to write a computer algorithm that describes how to identify every possible instance of the pattern. One of the major research projects that I've worked on over the past couple of years involves a method of classifying and identifying chart patterns in a rigorous way. This method, called "Bar Pattern Analysis" (BPA), consists of coding certain characteristics of price behavior so that any action on a chart has a unique "string" of code that describes it. This way, there is never any doubt about what pattern we are looking at, keeping everyone on the same sheet of music. I'm still working through some conceptual issues with this approach, and will write more about it in future articles. 2. Few TA indicator "signals" are supported by research: Technical Analysis does not just consist of classic chart reading however. Many technicians understood the drawbacks of classic chart patterns that I've discussed, and took a great step forward with the construction of indicators. Technical indicators such as moving averages, MACD and RSI provide clear unambiguous signals such as moving average crossovers, overbought/oversold levels and so forth and helped to eliminate much of the "subjectivity problem" with TA. However, not everyone agrees whether a given signal is bullish or bearish. Even when there is agreement, it is usually based only on common beliefs about market behavior or on anecdotal evidence ("See? The Williams %R flashed an oversold signal on IBM here, and the price went up soon after!"). This is not to say that no actual scientific research has been done on TA indicators, but it's far from a common practice. Much of my own work involves the use of statistical analysis to determine whether the behavior of a given technical signal has any significant correlation to future price movements. 3. Technical Analysis deals mainly with direction: There are certain chart patterns like triangles and flags which are supposed to tell you not only the direction of a future price move, but also how far the price will move. However, most of TA just answers the question "up or down?" That's not enough. When I traded currencies in 2004, I was constantly setting my stops too tight, my targets too far away, or vice-versa. This experience made me realize that "up or down" is not the only important information. Traders need answers to questions like "Should I set a tight stop or a wide stop in this situation?" and "Should I set my price target close for a small but likely profit, or far away for a larger but less likely gain?" In my research into FOREX price behavior, I try to design my research efforts to include not only the question of which way, but the question of how far as well. 4. The effectiveness of indicators changes as market behavior changes: This is a conjectural model of how markets work that I'm studying currently. By "conjectural model" I simply mean that this theory makes logical sense, but I don't have a shred of evidence to support it yet. The general idea is that the makeup of the "crowd" that is currently involved in actively trading a financial instrument changes over time as some people cease trading it and others begin trading it. Each individual trader has their own style, and as the makeup of the "crowd" changes gradually over time, the behavior of the price will change its characteristics as well. I suspect that this is an evolutionary process rather than a revolutionary one. For any given period of time though, the price behavior will have a certain "personality", a phenomenon that I've seen mentioned by many traders. Consequently, I believe that there are actually no "good" or "bad" indicators. There are just indicators which are currently working well for a given market and others which are not. One of the constant challenges for traders is to determine which is which, and to update their trading strategies accordingly. Some of my current research centers around creating tools which allow traders to do this. Instead of a trader saying, "Well I use a combination of moving averages and RSI" and so on, my vision is that traders will be able to say, "I use whatever indicators happen to be working best right now in whichever market I'm trading."
Scott Percival
|
|