It is really tough to track. If you follow the timeline of a single distressed home it is easy. First the NOD, then the NOT then the trustee sale. In general the trustee sale happens almost 4 months (a little less actually) from the NOD.
However many people that get the NOD may cure the default only to get another NOD down the road and start the train all over.
My “guess” is that the reporting you see is simply the number of recordings that come out in that time period. So say the sample is for one month. In that month there are x NODs and y NOTs. There is not any correlations if any of the y NOTs were due to any of the x NODs. In fact more then likely the y NOTs were due to NODs in the previous months, mostly about 2-3 months prior.
Wards predictions are pretty accurate and I like his site. I kind of wish he had more details about the data itself and more information about cures, repeat offenders and such. Good site though.