以「琥珀」署名写作。第一部专著《股票探幽》围绕同一个对象——股票——提出十个彼此独立的问题，并为每一个问题构造一个可以被公开数据推翻的读数。写法上有一条不太讨喜的纪律：把不利于自己的检验结果原样留在正文里。书中十次检验，一次全过，四次被作者事先登记的阈值或安慰剂判负，两次未执行；十个读数里有四个在现行记录体系中根本估不出来，一个被自己写下的作废条款废掉。
这些问题不从金融学内部找答案。三十门与金融无关的学问被请进来——矿物学、校勘学、树轮气候学、滴定终点与等当点、条件等色、癌症筛查流行病学、金属疲劳、验收抽样、人道主义排雷、法医学的死亡判准——不是拿它们打比方，而是让它们各自的判据互相顶撞，撞出一个金融学自己说不出的落点。
Writing under the name Hupo. The first monograph, Gupiao Tanyou, puts ten independent questions to a single object — the stock — and builds for each a reading that public data could refute. One discipline runs through it: unfavourable results are kept in the text as they stand. Of ten tests, one passed, four were struck down by the author's own preregistered thresholds or placebo cuts, and two were never run; four of the ten readings turn out to be unmeasurable in any existing record system.
The answers are not sought inside finance. Thirty distant fields are brought in — mineralogy, textual criticism, dendroclimatology, titrimetry, metamerism, cancer-screening epidemiology, metal fatigue, acceptance sampling, humanitarian demining, forensic criteria of death — not as metaphors, but so that their own standards of judgement can collide.
十次可以失败的检验 · 三部十章
Ten tests allowed to fail · Three parts, ten chapters
关于股票，我们拥有大量正确而无用的知识。凡是能被替代的，都不是它之所以是它的那一样。
Most of what we know about stocks is correct and useless. Whatever can be replaced is not what makes it what it is.
这一栏留给后续的文章。同一条纪律照旧：每一篇都要给出一个能被公开数据推翻的落点，检验不利就把不利原样留下。
This section is kept for what comes next, under the same rule: every piece must offer something public data could refute, and unfavourable results stay as they are.