Try calculating the amount of scrap you'd need and how high you would need to lift it (then the storage area needed) if you want to store a few megaWatt-hours from midday through to midnight. I think the logistics might lead you to wonder if it might be easier by rail...
If it is gravity based storage, you want as much of the movement as possible to be in the Vertical Plane. As soon as you introduce a gradient you are already less efficient than optimal. Reactionary forces on wheels and bearings will be the main frictional forces to overcome, proportional to both force at the contact point and speed.
With the rail based solution friction between the rail and wheels needs to be as high as possible whilst friction between the axle and bearing wants to be as low as possible, this is already creating a compromise as you cannot have both. The mass and angular speed of rotation being the same, you have a lower reactionary force between wheel and rail on an inclined plan because the force is not perpendicular to the rail whilst because the bearing is cylindrical the reactionary force on the bearing would be the same, just in a different location.
With a winched vertical weight solution, so long as you could fix the cables to the drums instead of running them back to a counter weight (which would defeat the purpose of the system), you have eliminated the equivalent requirement of High friction between rail and wheel. So you are just down to minimizing friction on the pulley bearings. This would then be a much simpler trade off between Mass with the corresponding cost of supporting this at height, vs the rate of change of height (proportional to the angular speed of rotation)
The real question is not the amount of scrap you would need to lift as the more you lift in either case the more storage you get but the cost of structures to support the weight in a vertical solution x the efficiency of the vertical system vs the cost of structures to support the weight in a rail based solution x the efficiency of the rail system. Where you accept that the rail system maybe less efficient. but maybe more cost effective.
What would be a perfect test bed for the theory would be a couple of old mines. You could test a vertical mine shaft v's an inclined drift mine. Assuming all things being equal you would be able to calculate each methods equivalent efficiency and use those against the estimated cost of a scaled up system. It is certainly an interesting engineering question.