Since all of the choices are looking pretty juicy from the new codex, I thought it might be helpful to break up the per/slot value and per/point value of each unit available.
-I've determined value here based solely on the ability to score wounds on T4, and armor penetration on AV12,13, and 14. Survivability, speed, range, etc... are not accounted for.
-Markerlights are treated as a seeker missile in terms of effectiveness, this is not at all accurate but gives you at least some guestimation on their value.
-All tank models used burst cannons, as these are ideal for this sort of simulation.
-Railgun submunition was given 4 hits per template on infantry.
-The test subjects were outfitted as follows:
Railhead: Multitrack, Targetlock, Burst Cannons, Decoy Launcher
Ionhead: Multitrack, Burst Cannons, Decoy Launcher
Skyray: Multitrack, Targetting Array, Burst Cannons, Decoy Launcher
3 x Sniper team
2 XV88 team: Team leader, A.S.S x 2, HW Drone Controller, 2 Shield drones
3 XV88 team: Drone controller x 3, 6 Shield Drones
So, damage caused per force org chart slot taken from best to worst: Best > worse > worse = equal > worst
Wounds on T4: Sniper teams > Railhead > Ionhead > 3 XV88 > Skyray > 2 XV88
Result on AV12: 3 XV88 > Sniper Teams > 2 XV88 > Ionhead = Skyray > Railhead
Result on AV13: 3 XV88 > 2 XV88 > Sniper Teams (seekers) > Railhead > Ionhead > Skyray
Result on AV14: 3 XV88 > 2 XV88 > Railhead > Sniper Teams (seekers) > Skyray
Now for damage caused per cost of the unit (both XV88 teams tied here, interestingly):
Wounds on T4: Ionhead > Railhead (shocker) > Skyray > Sniper Teams > XV88s (smart missiling)
Results on AV12: Sniper Teams > XV88s > Ionhead > Skyray > Railhead
Results on AV13: XV88s > Skyray > Sniper teams (seekers) > Railhead = Ionhead (iiinteresting)
Results on AV14: XV88s > Railhead > Skyray = Sniper teams (seekers)
So there you have it. Sniper teams performs quite well with a supply of seeker missiles, XV88s are the kings of vehicle destruction in all categories, Ionheads are very economical, and overall there's a lot of swapping back and forth which means no single choice is really going to become 100% dominant. The numbers are a big mess of Excel spreadsheet code, so don't ask.