Before reading below, keep in mind that simulations rarely have an accurate CG location, unless you maticulously put in the weight distribution of all your expoxy, machine screws, shear pins, etc. Most people don't do that.
What they give you is a pretty good starting CP. Assemble your rocket in ready-to-fly configuration, just as if you were on your way to the pad. Measure your actually CG by balancing the rocket on your hand or a sawhorse, etc. Use your calculated CP and measure CG to check the actual stability. If you still come up with less than one caliber, read below.
To further elaborate on Frank's post, the CG moves forward as the propellant burns away. So if you start with .85 calibers, you need to give yourself a long enough time on the rod/rail, so that by the time the rocket leaves the rod/rail, the CG has moved sufficiently forward to achieve a one cal. or better stability.
Less than one cal. means that gravity might not be able to overcome wind rotation, which is why the rocket will start to "sky write" (spin in the air). Greater than one cal. leads to weathercocking, which is unavoidable in many cases anyway, but should be minimized.
Supersonic rockets experience a sudden forward leap of the CP at Mach transition, and thus need to start with two or more calibers.
As stated above, short and stubby rockets are more stable by design, and thus don't need a full caliber of stability. And long, skinny rockets (think "Estes Mean Machine," "Fliskits Richter Recker") are inherently overstable and risk bending in the air.
For a DX3, add nose weight to get to one cal.
