Several challenges undermine our ability to predict the course of evolution: the likelihood of evolutionary novelties, the impact of the adaptive landscape, and the influence of developmental biases. NOBLE (NOvelties, Biases, and Landscapes in Evolution) will make use of cutting-edge machine learning, fossil times series, and biomechanical modelling to generate a synthesis of the predictability of phenotypic evolution. First, NOBLE will use ancestral state reconstruction to determine if origination of and loss of novelties are predictable over macroevolutionary timescales (>113 million years). Trait origination may be predictable from ancestral morphology, while degree of trait specialization may predict loss. Next,NOBLE will investigate whether convergent evolution is a product of adaptation (as is so often assumed) or chance (rendering it less predictable). This phase will quantify the magnitude and frequency of morphological convergence between functionally and structurally homologous traits in different lineages. Then, observed convergence will be compared to converge achieved in simulations of Brownian motion (and biased random walks over biomechanically-derived adaptive landscapes. Finally, NOBLE will determine the impact of integration between traits on 1) the strength of within-trait developmental biases and 2) morphological diversity. This is possible because NOBLE’s model organism, cheilostome bryozoans, have a rich fossil record and a colonial growth form – allowing within-genotype, within-population phenotypic variation to be quantified. The influence of trait integration will be examined at five levels of comparison (from within a single species to across multiple clades). Not only will NOBLE begin untangling the Gordian Knot of evolutionary predictability, it will provide an emerging researcher with an advanced statistical skillset (incorporating phylogenetic and paleontological analyses).