As pointed out by Shiller (“Irrational Exuberance”, 2000), equity risk premia won’t always be positive. However, that is only a straw man; nobody actually claims that.
For different risk premium thresholds (0.00%, 1.00%, 2.00% and 3.00%), the achievement likelihoods are based upon periods between 1953 and 2019, using UK return differences (equities v long conventional gilts) over periods ranging from 1 year to 15 years (charted here). The numbers reflect the 6 different experiences considered.
Across all 6 periods, for 15 years, the risk premium would be negative 22% of the time or below 3% 47% of the time, which seem pretty good odds. The period 1985-2019 is quite different to other periods but is not necessarily going to be where we remain. The likelihoods for periods of 15 years are shown in the following table (probability higher than 66.67% or probability higher than 33.33%) and are interactively charted for all periods.