One of the properties of acetone is that the relative density is actually closer to gasoline than ether, sitting at .79 g/cm^3 rather than .789 g/cm^3 for ether, both at 20degC, and as temperature fluctuates closer to 0degC or higher than 20degC, the acetone mixture keeps density closer to the .80 g/cm^3 of gasoline, rather than fluctuating in density quickly, as is the case with ethanol.
Another thing to note is that acetone has a 100% solubility rate in water, gasoline, and ethanol, but ethanol combining with acetone, it doesn't retain it's properties, but rather, it develops a more complex ketone, closer to ethanol. So, essentially, ethanol in your tank would reduce the affect of acetone.
The two particular issues with acetone, where I think most of the difference is, is how it acts almost as a 'coolant'. It evaporates slower than gasoline (Index of .11 rather than .20), and the ignition point is higher, while the flash point is nearly the same (-20degC), so it keeps the same properties, but 'keeps' the gas molecules from one and the other, keeping a slower overall burn, and as the body ketone of acetone is burning, all the gas has previously burned off, most likely, and the ignition toward the trailing end creates a heavier flame front, which burns off the unused gas much more efficiently than just letting the gas ignite on its own, either at the cat or afterword, hence lower emissions, and the burning of the acetone is rather clean, only putting out a slight amount of NOX and/or CO, so the actual exhaust pushed out is much cleaner, rather than being comprised so much of unburned gas.
High solubility in most any compound makes it much easier to combine and burn, and really, in cold weather, I'm guessing it would actually help to a degree, just due to acting like a 'dry gas', of sorts, as it is totally miscible in water, and becomes a burnable compound which pulls the water out of the tank, lending it to a density closer to the rest of the gasoline, rather than settling at the bottom of the tank, so it will be burned off successfully.
Lots of things to consider with this, but so far, the experiments are going well. And I'm actually preforming some at home myself (DON'T DO THIS IF YOU AREN'T USED TO IT!), and am very interested in lots of the findings I'm coming across. I'm going to try and find some way to test the NOX/CO/carbon soot output of the material, in multiple ways, both mixed with gas, non-mixed, vaporized, and so-on.