What are training variable models that we can work towards tying to predict? It's not easy to do; yet, its a highly valuable commodity. Don't think so, just go ask about every team in the NFL who has been riddled with injuries during the past few seasons. How did lack of training prediction models affect athletes' injury rates, performance, outcomes, team revenue, and coach's job security?
The easier route is just to take someone's word for it, right? Well, text or personal references aren't meant to be the end of discussion, but merely a point of discussion. What you see is not always what you get in terms of what's going on with an athlete and being able to compete at a high level (to produce). We really have to try and look under the hood, to see what's really going on and what's the cumulative effect of both training AND competing.
To generalize that athletes should train mostly for power and explosiveness is a misunderstanding in my opinion as these attributes are by-products of other quantifications. In this context, how important is underlying strength, endurance, and skill. How about genetic predisposition? Yet, it can be tempting to take training variables out of context and sexy them-up to make them the focus of training. The body is smarter than that in that you can't develop what you don't have.
We should get better at making educated prediction models of the individual needs of the athlete and the team in general. Prediction models that are made on-going, in real-time, and based on what we see and what we don't readily see. Are you doing this? If not, why?