Time-Period Basis

An implication surrounding the use of time-series data in which the final statistical conclusion can change based on to the starting or ending dates of the sample data. The concept discourages using a smaller time series, as chance events are more likely to be reflected in the conclusion.

When testing various investment strategies, the time-series basis can be significant. Idiosyncratic events or using a time frame that only captures certain business cycles can make certain strategies appear more profitable than they really are. To help avoid the time-period basis when testing investment ideas, one should use very long periods of sample data.

For example, let's test a momentum strategy in which an investor enters a long or short position in a stock that has a rapid price increase or decrease. The investor identifies a stock to buy for the testing period. If the strategy is tested during a period where the entire market was strengthening, it could make the momentum strategy appear very profitable. However, if the profits were driven only by the conditions of the market, then the strategy would become dangerous in an economic downturn. Extending the testing period to allow full business cycles would reveal the true profitability of the strategy.