The returns in the previous quarter (x axis) are plotted against those in the next quarter (y axis) and I fit an ellipse to the data.

Every ellipse has two radii or 'semi-axes' which represent the two principal components of the data.

PCs are 'idealised' axes which better fit or explain the data.

The first and largest PC has a slope of -1 and cuts the ellipse in half along its 'fattest' part.

The second is at a right angle to the first and cuts the ellipse in half again.

I likened PCs as 'ideas' before and it is also apt here.

'Ideas' are drivers of returns, portfolio managers should live and die by the 'ideas' on which their portfolios are built.

In this case the first PC or 'idea' is 'mean reversion', if a return before this period was negative it's likely to be positive next period, and vice versa.

Another interesting point is that the top right quadrant of the picture is almost empty.

However are two very large 100% jumps in the quadrant, whereas the downside is more limited. Clearly the Vix generally has a positive skew.

I wonder does the opposite hold true? Next time I'll plot some equities.

Javascript code is here.