Capital Ideas & Analysis

Using our premise that weekly MACD trends represent bullish and bearish cycles in the stock market, i sought to look at daily MACD trends to see how long cyclical and counter cyclical moves last. This is a study of time and momentum, not  necessarily price.

Though i will eventually go back further in time, for now Im doing this work from 1995-2008.

Definition: When the weekly MACD crosses above 0, that is considered a switch from a bearish market to a bullish market. a cross below means the inverse. When the daily MACD moves in the direction of the weekly MACD, that is defined as a trend move. when the daily MACD moves opposed to the weekly, that is a counter trend move.

In this timeframe, Ive found 44 trend and counter trend swings. In the past 5 years–since 2003–trending moves have averaged 12 weeks. while counter trend moves have averaged 3.8 weeks. From 1995-2003, cyclical moves lasted 10 weeks, and counter cyclical moves averaged 4.3 weeks. The caveat here is–from dec 1994-jan 96 there was a cyclical move higher. this lasted 112 weeks. Since this is so out of character, i did not include it. It is an important note of how the unexpected/impossible can happen though.

• longest trend move ;112 weeks
• shortest trend move; 1 week
• longest counter trend move; 11 weeks
• shortest counter trend move; 3 days
• Average Trend range; 10-12 weeks
• Average Counter Trend range; 3.8-4.3 weeks

This gives us a good template of what to expect for specifically counter trend moves. no counter trend move–in momentum–has lasted 3 months. Thus, until the weekly MACD turns positive, we can expect daily MACD positive crossovers to fail within 3 months, and lasting on average only 1 month.

Going against the Grain;

I also sought to look for MACD divergences, and how they played out. I filtered them by making it a requisite that ROC 39 also confirmed this divergence. What i found was that in those 13 years, there were only 9 divergences that were confirmed by ROC 39. If one were to have sold a confirmed divergence, and waited till daily MACD crossed the 0 line, then had the MA’s cross over as the sell signal, they were 100% profitable (so the next one will be a 20% loss, right?).

• There were two that went with the trend, 7 went against the primary trend.
• average time span for the divergence to take place was 2.77 months. longest was 7 months, shortest was 1 month.

Thus, though there is probably an instance of these divergences not working, they are surely worth taking the trade. There have been a few others of MACD divergence, but no ROC confirm. some were winners, some were losers. One recently happened in fact.

Ill continue to crunch the numbers back in time, but this should give us a great idea of what to expect when this indicator moves.