Friday, February 16, 2007

My QA = TA Post Sparked a Debate

I knew this was coming. Posting a link to my previous blog entry in a quants forum sparked a heated debate. See what very intelligent people has to say about the merits of quantitative analysis and technical analysis. Some even pointed out that TA has more realistic models than QA.

Link to the QA vs. TA thread

Tags:

Wednesday, February 07, 2007

Quantitative Analysis = "Highly" Technical Analysis (?)

Branding Quantitative Analysis as "Technical Analysis" will probably bring in some violent reactions from quants. But I just want to point out the similarities that they share. In fact, it can be seen that Quantitative Analysis is a higher form of Technical Analysis.

Technical Analysis is commonly described as Charting. It is the study of charts (graphical representation of past price movements) and finding patterns in them. Investment decisions are then based on these patterns. People say this is superstition as price moves randomly and just forms these patterns by chance. Technical analysis also utilize quantitative techniques via Technical Indicators. Technical Indicators aren't just numbers, they are results of some statistical modelling. Indicators like MACD and Bollinger Bands are actually similar to statistical measures used by quants today (mean and standard deviation respectively). These measures are used for momentum and mean reversion strategies. Technical analysis also looks into other quantifiable variables found in the market like traded volume, open interest, bid ask spreads, etc. Technical analysis gives rise to automatic trading rules which is also done with quantitative analysis.

In the Jan/Feb 2007 Issue of CFA Magazine, there is an article ("Perpetual Motion by Susan Trammell, CFA") about a recent study on trends in quantitative investing. Below are some findings:

Phenomena Being Modeled:

  • Fund Capacity: 20%
  • Impact of Trades: 24%
  • Textual Data: 2%
  • Higher Moments: 2%
  • Regime Shifts: 10%
  • Volatility: 20%
  • Extreme Events: 10%
  • Momentum / Reversal: 31%
  • Trends: 28%
Modeling Methodologies Used:
  • Shrinkage / Averaging: 9%
  • Regime Shifting: 4%
  • Nonlinear: 7%
  • Contegration: 7%
  • Cash Flow: 17%
  • Behavioral: 16%
  • Momentum / Reversal: 28%
  • Regression: 36%
As seen in the survey results, trends, momentum, and reversal models are quite popular in quantitative analysis. These are also the same phenomena being modeled by technical analysis but at a less "scientific" degree.

The relationship of Technical and Quantitative analysis can be likened to the relationship between Astrology and Astronomy. One is seen as superstition while the other as a science. Astrology came about due to the lack of sophisticated tools and theories. The same with Technical Analysis -- people relied on charts because it was easier to analyze than numbers. But in the advent of faster and more powerful computers, large amounts of numbers can be analyzed with ease.

To see the survey results, please refer to www.theintertekgroup.com.

Tags: