The relationship I'm suggesting is that most investors will successfully achieve modest returns, but only few investors will achieve outlierly high ones.

What about people who achieve spectacularly low returns, like -99%?

I'm not kidding.

The "expected distribution of returns" is one of the key areas of study in economics.

Obviously you can't lose more than 100%, and your potential gain is unlimited.

I assume most people have heard of the "normal distribution" also known as the bell curve. It predicts things well for linear processes (e.g. sum of several die rolls) but not as well for geometric processes.

One distribution that fits geometric processes well is the Lognormal distribution (the logarithm of the return is normally distributed). Looks something like:

|
|       ...
|     ..   ..
|    .       ..
|   .          ...  
|  .              ...
|  .                 ....
| .
| .
|.
+-----------------------

A very few people lose it all, most people perform in the average range, and a very few people do spectacularly well.