You are telling me quant funds prefer a heavy-weight and slow ANN to a fast, specialized model? Maybe in totally non-critical environments, otherwise it's simply too slow, regardless of performance.