Germanic Statistics

The thing I love about German is that it's atomic (and, as I am slowly realising, Chinese too!) lots of words are built from smaller words.

For example.

Glove is 'Handschuhe' - literally sticking two words together, i.e. 'hand shoe'.

Schadenfreude - 'pity joy'.

Clumsy is tollpatschig - 'terrifically splatty'.

Schifffarht - 'journey on a ship'.

English does it now and again too - but Germans do it all the time - building new words from smaller building blocks.

Many of my old students used to try to do the same in English - to which I gave them a stern look - no such creativity is allowed in English.

Unfortunately, English isn't as neatly atomic as German.

English is such a mishmash of old French and German it doesn't lend itself to Lego-like-linguistics.

Finance is at its best when you can break it down into simple atomic blocks and build up portfolios from those blocks.

Statistics is no different.

Let's take a time series,

The venerable S&P 500.

This is how most if not everybody displays a time series of returns.

If you were naive, you would assume each daily return as a somewhat atomic piece of data.

In a way yes, but you could never compare a cumulative return from the '60s to the present day, there's magnitudes of difference.

In order to make cumulative returns comparable we have to demean the time series.

All of a sudden the little dinks we see in the first graph are magnified greatly.

We can easily map the recent data to recent history.

1980's recession - check.

Dotcom crash - check.

2008 - check.

And of course, now it's easier to see what happened in the distant unfamiliar past also.

German and Chinese is far more clean cut than English with some really nice benefits - e.g. easier to read ancient texts - and the same is true after de-trending your time series.