英国《金融时报》报道，交易算法（Trading algorithms）对我们，尤其是企业高管的遣词用字，正起着深远的影响。
这几年来，由机器驱动的美国企业监管文件（公司的季度和年度财务报告）的下载频率激增，预示着对冲基金正在培训即时读取大量监管报告的内容，以做出快速证券交易决定的智能算法。这种自动化决策算法驱动的资讯汲取与分析量，是传统人类基金经理专才所望尘莫及的。
一篇题目为《如何在机器监听的年代说话：人工智能时代的披露》的论文指出，许多公司热衷于展现公司最光辉的一面，以吸引投资者的青睐。（论文链接：https://www.nber.org/papers/w27950）许多企业尽可能依据机器容易读取的资讯格式，来调整个别公司的业绩报表。
Source: https://www.nber.org/digest-202012/corporate-reporting-era-artificial-intelligence
该篇论文的作指出：“越来越多的公司意识到，企业强制性和自愿性披露的公司信息之资讯受众，不再仅是人类分析师和投资者。[...] 大量的股票交易，其实由机器人和算法提出的买卖建议所触发的。而这些机器人和演算法，是使用机器学习工具，以及自然语言处理套件来接收、处理和分析信息。”
曼集团（Man Group）的卢克·埃利斯（Luke Ellis）是其中一位接受了“机器阅读”识读培训的众多首席执行员之一，学习避免使用某些（会让机器算法负面评价其公司表现的）短语和单词。
这正在形塑一种新的业界现象。一些专业服务人员表示，一些公司的投资者关系部门正在应用自然语言处理（NLP）模型，来测试多种版本的公司声明，以查看哪种（经过专业用语调校的）公司声明，会让主流交易演算法赋予最高的评价“分数”。无论如何，来自金融和企业研究平台Sentieo的Nick Mazing认为这是一场绝望的战斗。
大多数复杂的自然语言处理（NLP）系统并不倚赖静态的敏感单词的列表，（而是在建模的过程中）自学能够识别有反映公司问题，和展现公司乐观景象的单词组合。举个例子，如果一个惯常使用“具挑战性”字眼的首席执行突然不再使用这个字眼，这可能带来更多信息含量。反过来说，一名不曾使用“具挑战性”字眼的企业领导人如果突然引用这个字眼，它的语境会传递关键信息。
还有就是，机器还不能读取企业高管的非言语提示，例如在高管在回答提问之前，呈现了抽搐（泄露了他的看法）。当然，曼集团（Man Group）的卢克·埃利斯（Luke Ellis）认为“（算法能够捕抓和读取非语言提示所蕴含的含义），只是时间上问题。”
[Excerpt] In an article titled "Robo-surveillance shifts tone of CEO earnings calls", the Financial Times reports that trading algorithms leave a mark with deeper focus on the spoken word.
There’s been an explosion of high-frequency machine downloads of US regulatory filings in recent years, as quant hedge funds simply train algorithms to instantaneously read and trade thousands of reports - volumes that no human portfolio manager could ever hope to read.
The paper, "How to talk when a Machine is Listening: Disclosure in the Age of AI, points out that companies are keen in showing off their business in the best possible light （https://www.nber.org/papers/w27950）. They have steadily made reports more machine-readable, for example by tweaking the formatting of tables, as a result of this evolving analysis.
“More and more companies realise that the target audience of their mandatory and voluntary disclosures no longer consists of just human analysts and investors,” the authors of the paper Sean Cao, Wei Jiang, Baozhong Yang and Alan Zhang note. “A substantial amount of buying and selling of shares are triggered by recommendations made by robots and algorithms which process information with machine learning tools and natural language processing kits.”
Man Group’s Luke Ellis is one of the CEOs who has as a result of machine reading been coached to avoid certain phrases and words.
This is becoming a phenomenon, with some specialised services saying that some companies’ investor relations departments are running multiple versions of statements through NLP models to see which “scores” best with the algos. But Sentieo’s Mazing thinks it’s a hopeless battle.
Most sophisticated NLP systems do not rely on a static list of sensitive words and are designed to both identify both problematic and promising combinations of words and teach themselves. "For example, one CEO might routinely use the word “challenging” and its absence would be more telling, while one that never uses the word would be sending as powerful a signal by doing so.
Machines are still unable to pick up non-verbal cues, such as a physical twitch ahead of an answer, “but it’s only a matter of time” before they can do this as well, Mr Ellis says." [/Excerpt]
Reference/资料来源
[1] 首要参考资料出处/Primary Reference: https://www.facebook.com/groups/strongartificialintelligence/permalink/3451419928245559
[2] 次要参考资料出处/Secondary Reference:
https://www.ft.com/content/ca086139-8a0f-4d36-a39d-409339227832?fbclid=IwAR25kwIM3Y52040GNuj6H4cgtfrvWBnN2S3SFkuBm0oGeimQBlkel5D588I
Posts
0 Comments