in the second table all names shorter than 5 chars or having numbers in them are excluded to filter out "Chinese" markets.
could you please post dollar amounts for these domains (no short & numerics)
I will if I have time later today, but I don't really think this would lead to a good analysis. I understand your motivation in trying to filter out the Chinese hype, but you have to remember that short domains have always been more valuable (there is and has always been an overall correlation between length and sale price).
Look at the top sales charts for previous years, we have them going back to 2004 on the side and you'll see they have always been totally dominated by short domains:
https://namebio.com/top-100-domain-name-sales-all-time
Take 2014 for example, 71 of the Top 100 sales are five characters or less. Should sales like Power.com, Check.com, Malls.com, Munch.com, etc. be excluded as Chinese hype? Certainly not, especially considering it was before the hype started. So by trying to filter out the insanity, I think what you're actually doing (for the most part) is filtering out quality.
That leaves you with longer domains that are still valuable, like DataCenter.com, BitcoinWallet.com, etc. A decrease in volume for these types of sales doesn't, in my opinion, point to a crash. I think it shows a shift away from longer and/or descriptive types of names, and a flight to short/memorable/brandable words. Or maybe even a scarcity in these types of names going on the market, since the set of "longer" names that can still command high prices is pretty small.
Fewer sales doesn't even always mean less interest (a negative sign), it could mean people are holding out for higher prices or are letting them expire less often (a positive sign). Hard to do this kind of market analysis, especially from 30,000 feet. I think it is better to take certain buckets (short dictionary, acronyms, numeric, Geo, etc.) and compare average sale prices year over year to see how the market is trending.