The background sounds like it is not necessarily going to be an actual physical scene, but an image of "their sport". The idea behind green screen is that the background is a contrast to the tones in the model, making isolating and masking the model easier and automatable (is that a word?). Even if you pull a great key and a perfect mask, you may have to edit the appearance of the model in the shot to match the lighting or color palette of the background plate onto which you are compositing the model to create a convincing comp that does not appear "cheesy."
If you capture all of the images in front of a non-uniform background, like a backdrop with sponsors' logos, creating a mask to eventually pull the subject out of the image and comp them onto another background could be a potential nightmare that will take a lot of effort to make a good mask. If you shoot against a generic green screen, you can comp them into any background, including one with sponsor logos on them.
Another possible, super clever way to attempt this would be to gather all of the background images that are going to be used and set the model up in front of a projector screen, onto which you could project the appropriate background image. This way you are effectively making the composite image in-camera. Depending on how you light your model, you may be able to gel the projector so that the color temperatures of the projected image background and the subject lighting match.
just a thought. I have never tried this, so I have no idea if it is feasible. Seems like it could work though.
kirk