How we measure performance

Here is the spreadsheet used to calculate performance claims made herein. We are receptive to reasoned arguments about performance measurement, so we invite you to download, examine, and make suggestions about the numbers.  To date, we are not aware of any challenges or corrections to the methodology used or results reported here.

Spinoff & Reorg Profiles suggests a number of ideas each month, but hangs its track record on a single recommendation, typically presented as a one-page case on Page 7 of each issue. It is usually, but not always, a spinoff or parent thereof.

When measuring performance, we want to know, first, how these recommendations compare to spinoffs in general, and, second, how they compare to the S&P.

Our goal is complicated by the fact that new spinoffs are issued at a highly variable rate — there might be 8 spinoffs in one quarter, and zero in another. As a result, if we attempted to calculate a shadow portfolio of spinoffs through time, we would need to make arbitrary decisions about porfolio weighting, which would introduce bias.

As a result, we believe the most accurate, representative measurement is simply to calculate the S&P-relative return of each spinoff over a fixed holding period, and then take the geometric mean of those returns. This generates a single number that can be compared directly, apples-to-apples, with our own stock-picking performance, regardless of market conditions.

A disadvantage of this method is that it cannot be used to construct a pretty, up-and-to-the-right performance graph. But we believe that what this method lacks in sales appeal, it makes up for in accuracy.

We use the geometric mean because it is a more demanding, and more accurate, measure of portfolio performance than the arithmetic average:  for any given portfolio, the geometric mean will be less than or equal to the arithmetic mean, and this becomes more true the greater the volatility.

Our first few picks in 2004 happened to have very high returns — so we threw them out for purposes of this calculation.  If they are included, our CAGR is somewhat higher.