A captured bearing arm has an interesting road to hew. On the one hand, the bearings must be infinitely free of play, on the other, they must be infinitely free of friction. As you can tell, the two are mutually exclusive. The accuracy of captured bearings is measured on the ABEC scale, as to how closely they approach these parameters. Essentially, it is a measure of just how far off they are from optimal. That said, modern captured bearing arms are very good, and have addressed these issues quite well. When properly aligned, they also completely limit any tortional forces; these twisting forces will make the stylus meet the groove at an inperfect angle and will increase wear and may even result in mistracking.
The unipivot pretty much solves the issues of play in the bearing and friction as well. The issue of tortional forces is much harder for the unipivot to deal with as the arm itself is free to move somewhat around its own axis. Various methods have been used to stabilize this twisting, oil damping, magnetic damping, etc. Many theorize that cartridges that put a lot of energy back into the arm, usually low compliance, with higher tracking force requirements, are not good matches for this type of bearing as they tend to exacerbate this undesireable motion. But many other factors come into play and I have not found this to be true in all cases in my personal experience.
In the end, as with tubes and solid state, execution tends to swamp the choice of unipivot or captured bearing designs. There are truly excellent examples of both.