Edward Lorenz summarized chaotic behavior as: "When the present determines the future, but the approximate present does not approximately determine the future."

So yes, you could predict the locations computationally to an arbitrary point in the future if you knew their starting locations and velocities with perfect precision; but in practice of course you cannot know anything's position with perfect precision, so your simulation would become inaccurate relatively quickly.

edit to add: and I believe that what this paper discusses is not a solution to the above, but rather a way of getting around it by modeling some types of three-body behavior as if it were truly random, rather than chaotically deterministic.