GDX and IWM Cluster Strengths for June 6, 2011

[TraderD]

The 20 DMF is not computed with the 500 components of S&P 500, but with over 1,000 stocks from all sectors. The 20 DMF signal is not derived proportionally to the signals from all stocks, but is proportional to the signals from all sectors.

Billy,

I wonder why simple averaging of signals from all sectors makes sense for the formation of the 20DMF signal. As sectors' aggregate market caps (and consequently, trading volumes) can be very different, wouldn't simple averaging introduce a skew into the 20DMF signal by over-representing the smaller sectors (and vice versa wrt the large sectors)? I'd be naturally inclined to weigh the sectors' signals in proportion to their market caps, but then, how's that different from just combining 1000 stocks' signals with their respective market caps weights?

Trader D

[Pascal]

Billy,

I wonder why simple averaging of signals from all sectors makes sense for the formation of the 20DMF signal. As sectors' aggregate market caps (and consequently, trading volumes) can be very different, wouldn't simple averaging introduce a skew into the 20DMF signal by over-representing the smaller sectors (and vice versa wrt the large sectors)? I'd be naturally inclined to weigh the sectors' signals in proportion to their market caps, but then, how's that different from just combining 1000 stocks' signals with their respective market caps weights?

Trader D

From back-testing, it worked better that way: same weight for each sector.


Pascal