Monday, January 12, 2015

Best Practices in Trading: Asking New Questions

I have consistently found that success comes from asking better and deeper questions.  It's the out of the box questions that can lead to fresh insights and answers.  I recently came across a 2014 review of performance from a trader that outlined all the mistakes he had made during the year and the things he wanted to improve.  I asked a question he did not anticipate:  "What did you do well in 2014 and what if you *only* did those things in the new year?"  Asking a fresh question will not always generate new and better answers, but asking the same, stale questions almost certainly will not yield creative insights.

In my current research, I'm looking at overbought and oversold indicators.  How much of an edge do they actually provide?  Do some measures offer significantly greater edges than others?  Over what time frames?  Do overbought and oversold measures offer different levels of edge in different types of markets?  Are there times to utilize these indicators and times we should not be giving them weight?

Notice that such an approach is very different than simply looking at a standard measure such as RSI or Stochastics and pronouncing a given level as overbought or oversold.  It's the tougher, more detailed questions that can yield nuggets of insight.

For example, suppose we track the number of stocks in the SPX index that are making fresh five-day highs vs. five-day lows.  If we go back to 2006 and divide the market into quartiles based upon volatility (VIX), we find that an oversold level in the lowest VIX quartile (1 SD below average) is -62.  An oversold level in the next VIX quartile is -138.  In the third VIX quartile, the same oversold level is -199.  And at the highest VIX quartile, the oversold level is -265.  In other words, what constitutes overbought and oversold is relative to the volatility regime of the market.  Looking at static levels of overbought and oversold across all markets gives us very distorted results.

Does a regime-specific measure of overbought vs oversold breadth offer a greater trading edge than an absolute level?  It's all very testable, but only if we gather the data and ask the question.

Further Reading:  Institutional Participation and Momentum