Evidence apparently not 集韻考正. Where is it from?

It seems made from 戴's left-top and left-bottom strokes. Should the last stroke look like 點 rather than 撇?

Agree to change the glyph from 撇 to 點.

I really don't think this is sufficient evidence for encoding. I would like to see evidence of actual use in printed text, demonstrating that there really is a need to encode this character. If IRG considers that local-use vulgar simplifications can be encoded on the basis of unverified second-hand evidence such as this, then there are hundreds and hundreds of Part 2 second stage simplified characters for which there is a stronger case for encoding.

The evidence posted is from
江西省公安厅一处 编纂：《江西方言土语汇集》第五册，1991年3月，P224
⿱十八 is a common character used widely in the Sourth of China.
Another evidence:
江苏省公安厅江苏方言总汇编写委员会：江苏方言总汇 上卷，北京：中国文联出版公司，1998年9月，P1860

《中国语言资源保护工程汉语方言用字规范》 (2017) p. 290:

kMandarin dài

《金華方言詞典》 (曹志耘, 江蘇教育出版社, 1996), p. 117

《中国语言资源保护工程汉语方言用字规范》 is largely derived from the dictionaries in the 《現代漢語方言大詞典》 series, including 《金華方言詞典》 cited in #14949. The glyph form in 《中国语言资源保护工程汉语方言用字规范》 is less reliable since it is digitization of previous work. Thus, I also suggest changing from 撇 to 點 for the last stroke.

⿱十八 is discussed thoroughly in _Long Story of Short Forms_ by Roar Bökset (https://vitterhetsakademien.bokorder.se/en-us/article/4450/long-story-of-short-forms), under the 戴 entry.

First paragraph: