Abstract
We exploit ideas of nonlinear dynamics and statistical physics in a complex nondeterministic dynamical setting using the Ruelle-Takens embedding. We present some new insights on the quality of the prediction in the laminar regime and we exhibit the data collapse of the predicted relative first difference fluctuations to the universal Bramwell–Hodsworth–Pinton distribution. Hence, the nearest neighbour method of prediction acts as a filter that does not eliminate the randomness, but exhibits its universal character.