Indicator hygiene

Sophomore · enrolled learners

This lesson opens with Your own configuration space, counted.

What happened

This lesson's opener is the course autopsy turned on the reader, and it is labeled as such. Bailey, Borwein, Lopez de Prado and Zhu proved that high simulated performance is easily achievable after testing a relatively small number of alternative strategy configurations, that the probability a backtest is overfit rises with the number tried, and that because analysts rarely report how many configurations were tried, a reader cannot evaluate the degree of overfitting in almost any proposal. A standard charting package presents an indicator library of well over a hundred entries, each with adjustable parameters, applicable to any of a dozen timeframes, on any of several price inputs, in any combination. The number of distinct configurations a single user can reach in an afternoon runs to millions, and the tool keeps no record of which were viewed and discarded.

The decision point

The formal result is about backtests and the informal version is what happens every time somebody adds an indicator to a chart, changes a setting, and keeps the version that looks better. The search is identical, the selection criterion is the same, and the only difference is that nothing writes down the attempts, which means the correction cannot be applied even in principle.

Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.

It is free. We do not sell the list and there is nothing to buy at the end of it.