If you already have the hull built, then the fin goes slightly behind the mast step, since very generally the rig centre of effort (CE) requires a lead of around 5% LWL forward of the fin CE.
But if you are free to design your boat, then the fin can in principle go anywhere (we're talking here about a keel which is a fin and bulb), while your ballast bulb's centre of gravity (CG) pretty much lines up below your hull's centre of buoyancy (CB). Here is your "blank piece of paper", hull and bulb in place, ready to receive the fin and rig:
Then, wherever the fin goes, the rig must go with its required lead, so the fin and rig form a "package", and are always moved together fore and aft.
Essentially, there are three fin configurations -- "T", "L", and "reverse L" -- and this is what they look like. Of course, in practice, the "L" and "reverse L" (also known as a prognathous bulb) would not be so extreme...
On a practical note, wherever the fin/rig package goes, the backstay must clear the mainsail sail leech. If the hull has no overhangs, this can restrict how far back you can place the fin. And, on an IOM, the jib counterbalance weight cannot project forward of the bows, so you may be restricted on how far forward you can place the fin.
On another practical note, the fin has some volume, and if placed significantly forward or aft will introduce a trim change which will need to be countered by a small adjustment of the bulb's position to return the bulb CG to below the combined CB of the hull and fin.
The "T" keel gives the least twisting stress to the fin when the boat is heeled. So why would you want anything other than a "T" configuration?
One reason is if you actually want the bulb to cause some fin twist when heeled. A little fin twist is like the wash-out on an aircraft's wing, where it reduces the wingtip vortex and consequent drag, and improves the wing stall characteristics. The "L" configuration provides wash-IN to the fin tip, ie the bulb end. It is the "reverse L" which provides wash-out at the fin tip. On the assumption you don't actually want wash-in, then you'll need a particularly stiff fin if you want to go with the "L" configuration. If you are wanting a little wash-out, are looking therefore for a somewhat prognathous bulb, and think the stiffness of your existing fin will be fine, I would suggest checking that you have enough stiffness to prevent or properly damp bulb oscillation when well heeled in waves. I ran a long-fin Ten Rater for a while, and noticed two things. One was that, out the water, the bulb seemed pretty wobbly on what was an older and less stiff fin. The other was that, while racing, the boat would slow and spurt unpredictably in waves, like the brakes were being jabbed and then released... I changed the fin to something much stiffer after a while, and this curious behaviour went away. I believe the bulb's fundamental period of oscillation was excited from time to time in waves, and that made the boat momentarily slow down.
Another reason is that you may want to try and place the fin where it will provide its most constructive wave interference with the hull at a given design speed, particularly if you have a fin with substantial volume. Problem is, no one really knows where this position is, and you would need many hours on a supercomputer and a series of experiments in a towing tank to get the answer. On the other hand, if you have a little time on your hands and an identical hull, you could do some side by side tow testing from a dinghy or from a footbridge over a stream to see if you can tell any difference. Put your two hulls into a yoke and trail them. Move the pivot point of the yoke attachment until the hulls balance each other in their drag. The hull further from the yoke pivot has the lower drag.
A third reason is that some designers think that the bulb is better off in undisturbed flow, and favour the "reverse L" configuration. On the other hand, a somewhat larger number of other designers think it is the fin which is better off in undisturbed flow, and favour the "L". I've heard it said that Bruce Farr claimed the "L" keel showed lower drag on round-the-world racers. Be careful when looking at keels on full-size racers, though, and mentally check whether they have likely been optimised for upwind racing (more "L"), downwind racing (more "reverse L"), or round the cans (more "T"), before deciding what you think the current trend is.
Finally, whether the centre of effort of the fin is ahead or behind the hull CB leads to differences in ability to hold course, turn quickly, and track in waves and/or following seas. Actually, I should say "whether the centre of effort of the combination of hull, rudder, and fin is ahead or behind the combination of hull and fin CB". The rule of thumb I have heard is that the hull contributes 10% towards the boat's underwater CE, and so it is usually ignored and only the combined rudder and fin CE, ie centre of area, is considered.
Update 2 Feb 2012
Pierre Reynaud tells me he has been looking at a "Mast-Fin-Tandem", a close cousin to my "Fin and Rig Package". He says,
©2011 Lester Gilbert