Momentum & PCA Analysis

Check out the first and second posts in this series to get up to speed.

This is a picture of AAPL's monthly returns over the last 3 years.

It shows ever so slightly more momentum than mean reversion as the ellipse is pointing up right.

In contrast to the previous equity and Vix pictures the plotted data is pretty centred. I would bet that AAPL has pretty much zero skew over this time period.

Momentum is more fleeting and brittle than mean reversion, harder to find.

No doubt there is a good reason why momentum strategies are more popular than mean reversion ones.

But mean reversion is more like a strong force versus momentum's weak force.

Here's the strongest momentum I have found. Tesla's monthly returns over the last three years.

Momentum is about twice as strong as mean reversion, reflected by the ellipse's upward sloped radius.

(Note the very positive skew also - most of the data is found in the bottom left quadrant)

Of course indicies exhibit momentum also, for example large cap value growth (VUG).

That's at a quarterly sampling over 11 years, but using a monthly sampling on the same data we see this,

Clear mean reversion!

If we drop down to weekly sampling we see both mean reversion and momentum balanced, as we would expect from efficient price movements.

What does a switch from mean reversion to momentum mean?

How about looking at more data?

I don't see such a switch with small cap growth (VBK) over the same period.

Neither with Small and Mid Caps since 2000 (IWO) or over the same period.

I also checked the international growth ETF (EFG) and did not find the same anomaly.

Every growth ETF shows momentum while sampling quarterly but not much mean reversion monthly.

Perhaps the VUG result is just a glitch in the matrix.

I need to pick over it more and get to the bottom of it.

Just a note, equity in general exhibits momentum at the quarterly level, not just growth.