Nice info.
If you can pull any form of word out of the first 2-5 letters then it's keyword based. The whole keyword does not have to be there to be a keyword based name. Don't go by what the BrandBucket newsletter says because their stats are totally wrong. They don't know how to place keyword or partial keyword based sales in their respective categories. Most of the sales that occur at BB are keyword or partial keyword based.
The letter frequency chart is a bit useless since vowels A E I O U will always be the highest amount most of the time.
I would also never include any 4 letter sales because they always skew the numbers and make the stats not useful when they are included. The sale of Dany.com at $65,000 totally messed up all the stats. The average sales price of domains over 6 years old was $6,163.21 which sounds amazing but if you don't include the Dany.com sale it would probably be half that number so the actual stat becomes useless. There were only 5 four letter sales in July which is only 6% of the total volume so not including them makes sense to get a much more usable stat. You should never include any sales that are total outliers. A $50,000 sale thrown in with a bunch of $2,000 sales will render all the stats useless.
You should make the categories completely invented names, partial keyword based names ,and unaltered keyword names.
Another stat that you should include is all sales that are 2 years old or less. Many with BrandBucket and their most avid followers say that aged domains are better sellers. It makes sense until you realize that 40 out of the 81 sales or 50% were handregs that are less than 2 years old. Never underestimate handregs.
You should stick to a one month table to keep it more relevant but it would be nice to see a table for the past months also.
Every bit of brandable info is useful. Thanks for taking the time to do this.
