

The deviation in the model of late is interesting but still within acceptable limits (a couple of sigma).
One thing to note is that these data are only showing a correlation to price changes and inventory, simply validating a relationship between prices and supply and demand. Low inventory and high sales will lead to prices being bid up, high inventory and low sales will lead to prices being bid down. Half-over-half price changes happen to be the most highly correlated to MOI. A combination of several delays will likely lead (actually, must lead, for those who think about it) to an even better model but meh.