The t‑test is a decision rule on a mathematical basis that can be used to analyse in more detail a difference between the empirically found mean values of two groups (for example portfolio returns vs. benchmark returns). The standardised sample characteristics are called t‑values, the standardised distributions are the t‑distributions. They do not quite correspond to the standard normal distribution, but are narrower peaked. Using the t‑distribution, an empirical t‑value can be assigned a probability with which exactly this or a larger t‑value occurs under the assumption of the null hypothesis.