Factor-Modification of Performance Benchmarks

Performance benchmarks in the orthodox sense are not, and can never be a measure of achievement. What they represent is the result of an unselected universe of more or less subjectively assembled investable assets. Whatever universe of investable assets a benchmark may contain, it is not selected for a specific purpose, other than representing that desired universe.

Benchmarks show the performance of all that is/was on offer, not what should have been bought, or held. One can only prepare fresh asparagus soup if asparagus was available for purchase. But an absence of asparagus is no reason to make soup from mouldy carrots if these were in generous supply. What is so obviously common-sense in cooking, should also apply to investment management.

In investment management, the critical professional skill is the ability to selected in favour of potential return, while discriminating, as much as possible, against undue risk. Benchmarks, representing merely an ‘inventory list’ of markets, but not a  ‘quality-choice’, can only serve as  simplified proxy for the conditions present, when a specifically contracted expert (the investment manager) had to make selection, in order to meet -as much as possible- specific needs for specific purposes (clients).

It is indeed logical to use an appropriate benchmark, but only as the raw data, to be modified so that it broadly reflects reasonable expectations regarding the value of professional selection skills (i.e.: modification factors), but also the environment (market conditions). One client may prefer reduced risk over possible returns, another will expect maximum return with a given level of risk. Modification factors should be set to reflect a reasonable range of combinations, whereas any constant also deployed allows more fine-tuning.

The table below a conceivable array of modification factors, and constants, applied to three distinct risk-tier performance benchmark indices. Each of the indices is being modified with all of three factor modification levels. These are modifications applied to monthly rates of change. They may look very small, but they do compound. The decision to apply modifications to monthly rates of change gives greater flexibility than doing the same to quarterly, or annual rates of change would.

This is an example, and not meant to be perfect. But it is also intended to challenge a status-quo, where benchmark returns are offered as a kind of ‘gold standard’ to aspire towards, while in fact, benchmarks in their raw form represent the complete absence of skills.

Anyone who may consider this approach unrealistically demanding is probably just spell-bound by the general glorification of mediocracy (and outright failure), so omnipresent in the financial services industry. One feels reminded of the tale of the Emperor’s new clothes.

For each of the benchmarks used to illustrate the concept, a table shows values for the benchmarks in raw form, and the three levels of factor-modification.

Shown are values at start and end, overall change of value, annualised returns, observed risk, and to illustrate the impact that factor modifications have, a few qualitative metrics.

Each graph plot performance of the raw benchmark and three levels of modification, in yearly increments, for twenty years between 31.12.1999 to 31.12.2019.

Here, the display of median, and upper (median plus 1 STDEV) as well as lower inflection points (median minus 1 STDEV) should be shifted further into positive territory by the impact of investment management skills.