Technical Analysis (TA) becomes even more popular during times of market turmoil. I find value in both descriptive and predictive TA but both are of little help in forecasting the next market move. For that you may be better off tossing a coin.
Descriptive Technical Analysis (DTA) tells us what is happening now in the market and maybe in relation to what has already happened. For example, there was a failed W bottom in S&P 500 last year and at this time there is another attempt for a W bottom, as shown in the chart below (updated as of the close of 02/12/2016.)
However, just from the above information we cannot know what will happen next. This is information journalists use for the most part. Some traders also use DTA to inform other traders about things that are happening in the market, some of them often hard to observe directly. This is often useful because it is impossible for a trader to know everything. Thus, there is value in DTA but it is limited in terms of forecasting value and for risk management purposes.
Predictive Technical Analysis (PTA) deals with probability and statistics. DTA cannot be placed within the framework of PTA unless there is a model of some sort. A model helps in removing subjectivity and confirmation bias. A good part of DTA cannot be placed in the context of PTA because of lack of sufficient samples, objective modeling and proper application of heuristics. An example of PTA is the 50-200 moving average crossover strategy for long-only position. The performance of this strategy in SPY from inception to the end of 2015 is shown below:
The Sharpe ratio is 0.60. That in turn equates to a t-statistic of about 2.87, which is a significant result, provided that the strategy was not obtained after multiple comparisons (something that we cannot know by the way.) In that case, it may be a random strategy that survived by luck.
It is also important to understand why PTA is not always very useful as far as forecasting the direction of the next move: Even with win rate of 80%, as in the above strategy, the probability of a loss is 20%, which is high. If we assume that the win rate is stationary, then the above strategy is profitable only in the context of ensemble averages based on sufficient samples.
As you can see, even PTA is problematic when applied to forecast the outcome of special situations, except in cases when the win rate is very high, maybe above 95%. This is one reason I always talk about the importance of high win rate. It is possible of course to be profitable with low win rate but forecasting accuracy for the next outcome is low. In this sense, PTA with low win rate has only longer-term value. The problem is that in the short-term it may lead to ruin.
I explain the above problems with DTA and PTA in my book Fooled By Technical Analysis. The failure, or even refusal, of many to focus on these issues is one of the reasons that technical analysis has contributed to a massive wealth transfer from retail traders to market makers. Using TA, DTA or PTA, without an understanding of probability and statistics, is like attempting to pilot a plane without taking any flying lessons, with the know results.
Charting and backtesting program: Amibroker
Detailed technical and quantitative analysis of Dow-30 stocks and popular ETFs can be found in our Weekly Premium Report.