About the same amount behind the CG as you'd put the main gear in front for a conventional gear.
Tim's response above is correct. And I am not trying to diminish his answere in any way. However, I offer the following:
That approximate 15
o a forward of the aircraft CG (taking into consideration the true CG including its vertical position for a conventional (tail dragger) LG has been time proven for our models ever since before Bill Netzeband told us about it almost 50 years ago (April, 1957). This also applies to full scale aviation. This rule-of-thumb angle applies for models flying from a paved surface. From grass fields, the gear can be mounted further forward.
As Tim mentioned, for a trike gear, that 15
o angle aft of the aircraft CG also works. However, a bit of research, going back to Bill Netzeband's columns (and others), that rearward angle can be as small as 10
o (Netzeband again in March, 1971). I have seen material written that for a stunt ship with a trike gear, the rearward position can be located just close enough to the CG such that the model will not set on its tail when at rest. This will allow some really nice smooth non-bounce landings. One thing to
NOT do with a trike gear on a stunt ship is for it to give any positive angle of attack when setting/rolling on the surface.
Now, I am going to beat a dead horse. In spite of the insistence of some probably well intentioned but less experienced and technically challenged yet vocal contributors to these forums, the landing gear position for a stunt ship has nothing to do with its relationship to any portion of the wing leading edge. The critical parameter to properly locate the landing gear is its position relative to CG of the model, regardless of the wing planform.
Keith