財報季忙到吐血？AI助力大摩分析師（GT）

01-29

按：本文基於GoogleTranslate的機翻結果編譯而成。完整原文貼在譯文後，供參考。

原文鏈接：https://www.bloomberg.com/news/articles/2017-10-11/hate-earnings-season-morgan-stanley-robots-give-analysts-a-hand

對那些把超負荷工作的財報季視為夢魘的賣方分析師而言，機器人「佔領華爾街」未必都是壞事。

摩根士丹利研究部正在嘗試用人工智慧取代人從事部分涉及公司財務分析的繁重環節——包括仔細閱讀各種披露文件、參加長達數小時的公司業績說明會等。這家全球最大的股票交易投資銀行每年發布5萬份報告，其中有些報告只不過是把公司披露的信息放在一起。按大摩全球研究部主管Simon Bound的說法，這些任務更適合讓機器去做。

Bound在接受採訪時說：「我們能用人工智慧大幅提高財報季期間分析師的工作效率」，「這是為解放分析師的生產力，讓他們做更有價值的工作，並提高發現洞見的速度；讓他們有更多時間與客戶直接打交道。」

華爾街擁抱新的認知技術讓有些人擔心未來飯碗不保。但至少短期內，在諸如摩根士丹利研究部這樣的賣方研究機構，大家樂見人工智慧助一臂之力——這也會成為分析師打敗同行的技術優勢。

Bound認為，未來機器學習軟體能迅速掃描文檔並撰寫淺顯的數據點評報告，每次都能節省幾個小時的人工勞動。當然他也強調：分析師將在報告正式發布前審查機器撰寫的內容，並加上簡短的主觀評論。這些軟體還能閱讀公司管理層電話會記錄，為主人（分析師）挖掘字裡行間的隱藏信息和情緒分析。

摩根士丹利也在開發研究門戶網站上的虛擬助理。起初，虛擬助理只能回應簡單的請求，例如告訴客戶特斯拉目標價是多少；或給客戶發送蘋果公司的估值模型等。

Hate Earnings Season? Morgan Stanley Robots Give Analysts a Hand

For those who dread the quarterly deluge of corporate earnings reports, the arrival of robots on Wall Street might not be all bad.

Morgan Stanley』s research department is testing artificial intelligence to take over the grunt work involved in covering earnings -- a brutal ritual that leaves analysts poring over filings and listening to executives talk for hours. The world』s biggest equity-trading investment bank churns out 50,000 reports a year, some of which are just summaries of disclosures. Such tasks are better suited to machines, according to global research head Simon Bound.

「We can use artificial intelligence to make the earnings cycle a lot more efficient for analysts,」 Bound said in an interview. 「The idea is to free up analysts for value-added work and accelerate discovery of insights; they can then spend more time in front of clients.」

Wall Street』s embrace of new cognitive technologies is causing some to worry about the future of their jobs. But at least in the near term, at places like Morgan Stanley』s research department, artificial intelligence will likely be a welcome help -- and a way to get an edge on analysts at rival firms.

In Bound』s vision, machine-learning software will scour filings and write basic reports, saving humans a few hours each time. Analysts will review the bulletins before publication and add their own brief commentary, he said. The programs also will read transcripts of management conference calls to generate insights and sentiment analysis for their human masters.

Morgan Stanley also is developing a virtual assistant for the firm』s research portal. Initially, it will be able to perform simple requests, like telling clients what the price target is for Tesla Inc. or sending over an earnings model for Apple Inc.